B R E N N A N
CENTER
FOR JUSTICE
voting system failures:
a database solution
Lawrence Norden
Brennan Center for Justice at New York University School of Law
about the brennan center for justice
e Brennan Center for Justice at New York University School of Law is a non-partisan public policy and law
institute that focuses on fundamental issues of democracy and justice. Our work ranges from voting rights to
campaign finance reform, from racial justice in criminal law to presidential power in the fight against terrorism.
A singular institution – part think tank, part public interest law firm, part advocacy group – the Brennan Center
combines scholarship, legislative and legal advocacy, and communication to win meaningful, measurable change
in the public sector.
about the brennan center’s voting rights and elections
project
e Brennan Center promotes policies that protect rights, equal electoral access, and increased political participation
on the national, state and local levels. e Voting Rights and Elections Project works to expend the franchise, to
make it as simple as possible for every eligible American to vote, and to ensure that every vote cast is accurately
recorded and counted. e Centers staff provides top-flight legal and policy assistance on a broad range of election
administration issues, including voter registration systems, voting technology, voter identification, statewide voter
registration list maintenance, and provisional ballots.
e Help America Vote Act in 2002 required states to replace antiquated voting machines with new electronic
voting systems, but jurisdictions had little guidance on how to evaluate new voting technology. e Center convened
four panels of experts, who conducted the first comprehensive analyses of electronic voting systems. e research
proceeded over a period of nearly two years and culminated in two path-breaking reports: e Machinery of Democracy:
Protecting Elections in an Electronic World, which focused on voting system security, and e Machinery of Democracy:
Voting System Security, Accessibility, Usability, and Cost. In the years since the Brennan Center published these two
reports, the Brennan Center has helped election officials and jurisdictions ensure that their electronic voting systems
are as secure and reliable as possible.
about the author
Lawrence Norden is Senior Counsel in the Brennan Centers Democracy Program and director of the Brennan Centers
Voting Technology Project. He has authored several nationally recognized reports and articles related to voting rights,
voting systems and election administration. In April 2009, Mr. Norden completed his duties as Chair of the Ohio
Secretary of States bipartisan Election Summit and Conference, authoring a report that recommended several changes
to Ohios election administration practices and laws; the report was endorsed by most of the States voting rights groups,
as well as the bipartisan Ohio Association of Election Officials. Mr. Norden was the Keynote Speaker at the Sixth
Annual Votobit International Conference on Electronic Voting (Buenos Aires, 2008), and the 2009 Electronic Voting
Technology Workshop/Workshop on Trustworthy Elections (Montreal, 2009). In June 2009, he received the Usability
Professional Associations Usability In Civic Life Award for his pioneering work to improve elections.Mr. Norden is the
lead author of the book e Machinery of Democracy: Protecting Elections in an Electronic World (Academy Chicago Press)
and a contributor to the Encyclopedia of American Civil Liberties (Routledge 2007).
© 2010. is paper is covered by the Creative Commons Attribution-No Derivs-NonCommercial” license
(see http://creativecommons.org). It may be reproduced in its entirety as long as the Brennan Center for Justice
at NYU School of Law is credited, a link to the Center’s web page is provided, and no charge is imposed. e
paper may not be reproduced in part or in altered form, or if a fee is charged, without the Center’s permission.
Please let the Center know if you reprint.
acknowledgements
As always, the Brennan Center and the author are exceptionally grateful to Laura Seago for putting in many long nights
to provide editorial and drafting assistance, as well as ideas that were integral in defining the direction of this report. John
Travis performed exceptionally in steering this report to completion. We also thank Susannah Goodman of Common
Cause, for pushing us to develop and write this report, and putting us in touch with the many regulatory specialists
who reviewed and commented on the ideas in this document. Paul Riley, now at Milbank, Tweed, Hadley & McCloy,
provided invaluable research and drafting assistance. Joe Hall, Aaron Burstein, and David Wagner of the University of
California at Berkeley and ACCURATE, Flavio Komuves, former Deputy Advocate for the New Jersey Department of
the Public Advocate and currently with the ACLU of New Jersey, Justin Levitt, Matt Robinson and Susan Lehman of the
Brennan Center and Sean Flaherty, Pam Smith and Warren Stewart of Verified Voting all generously gave many hours to
review drafts of this report and provide critical feedback, which has been incorporated into thenal document. anks
also to Scott Kareff of Schulte Roth & Zabel LLP for his careful reviews of this report and his thoughtful feedback.
A special debt of gratitude is owed to Scott Nelson of Public Citizen, who provided us with guidance and insight
into how better regulation and oversight could improve voting systems nationwide. We received similarly invaluable
assistance from Joan Claybrook, former President of Public Citizen and head of the National Highway Traffic Safety
Administration, as well as Pamela Gilbert of Cuneo Gilbert and LaDuca, former Executive Director of the Consumer
Product Safety Commission.
Douglas Kellner, co-chair of the New York State Board of Elections, John Gideon, Susan Greenhalgh and Ellen eissen,
among many others, have long advocated for a better clearinghouse for voting system problems, and their perseverance
in pushing this idea was an important inspiration for the report.
is report would not have been possible without the many election officials who agreed to be interviewed, review case
studies and provide feedback regarding the substance and recommendation of this report. Among the county election
officials, we especially thank: Betty McGary, Executive Director, Butler County (Ohio) Board of Elections; Carolyn Crnich,
Humboldt County (California) Clerk; Denise Lamb, Chief Deputy Clerk for Elections, Sante Fe County (New Mexico);
Cherie Poucher, Director of the Wake County (North Carolina) Board of Elections; Joanne Rajoppi, Union County
(New Jersey) Clerk; Rokey Suleman, Executive Director, Washington, D.C. Board of Elections and Ethics; Jane Platten,
Director of the Cuyahoga County (Ohio) Board of Elections; Matt Damschroder, Deputy Director of the Franklin County
(Ohio) Board of Elections, and Gail Siegel, Communications Director for David Orr, Cook County (Illinois) Clerk.e
following state election ocials and offices were also exceptionally helpful: Lesley Mara, Deputy Secretary of the State of
Connecticut; Lowell Finley, Deputy Secretary of State of California and the office of the Ohio Secretary of State.
We are grateful to the Election Assistance Commission, and in particular, Jeannie Layson, Director of Communications
and Congressional Affairs, Matt Masterson, Deputy Director of Testing and Certification Program, and omas Wilkey,
Executive Director, for graciously agreeing to meet with the author, discuss the ideas in this report, and promptly
answer his many questions, as well as the EAC’s current efforts to share with state and local election officials important
information about election administration and voting systems.
Kitty Garber of the Florida Fair Elections Commission, Professor Penny Venetis of the Rutgers School of Law Newark,
David Zvenyach of the District of Columbia City Council, Professor Candice Hoke, Cleveland Marshall College of
Law, and Noel Runyan all provided essential assistance in drafting and reviewing case studies and sidebars integral to this
report.
e author thanks Susan Liss, Jeanine Plant-Chirlin, and Wendy Weiser of the Brennan Center for their guidance
throughout the drafting process. Of course, any errors in the report are the author’s alone.
e Brennan Center is grateful to the Carnegie Corporation of New York, Democracy Alliance Partners, the Ford
Foundation, the Irving Harris Foundation, the Mitchell Kapor Foundation, the Open Society Institute, Quixote
Foundation, the Rockefeller Family Fund, the Tides Foundation, and two donors who wish to remain anonymous for
their generous support of our Voting Rights and Elections Project.
e statements made and the views expressed in this paper are solely the responsibility of the Brennan Center.
table of contents
EXECUTIVE SUMMARY 1
Core Findings 2
Central Recommendation: Creation of a National Database
for Voting System Problems 3
Additional Recommendations 4
I. INTRODUCTION 6
II. THE CURRENT PROCESS FOR PUBLICIZING
AND ADDRESSING VOTING SYSTEM DEFECTS
7
III. FAILURES OF THE CURRENT SYSTEM: CASE STUDIES 10
Butler County, Ohio, March 2008 10
Humboldt County, California, November 2008 12
Orange County, Florida, November 2006 13
Pulaski County, Arkansas, May 2006 14
Florida, November 2006 14
Broward County, Florida, November 2004 15
Florida, June 2004 16
Alameda and San Diego Counties, California, March 2004 16
Bernalillo County, New Mexico, November 2002 17
Wake County, North Carolina, November 2002 18
Fairfax County, Virginia, March 2009 20
District of Columbia, September 2008 21
New Jersey, February 2008 22
Indiana, May 2006 24
IV. A BETTER WAY TO TRACK AND ADDRESS
VOTING SYSTEM PROBLEMS 27
A Publicly Available, Searchable Centralized Database 27
Provision Details 28
Responsible Agency 29
Analogous Regimes 30
Key Benefits 31
Vendor Reporting Requirements 32
Provision Details 32
Responsible Agency 33
Analogous Regimes 33
Key Benefits 34
A Federal Agency with Investigatory Powers 35
Provision Details 35
Responsible Agency 35
Analogous Regimes 36
Key Benefits 36
Enforcement Mechanisms 37
Provision Details 37
Responsible Agency 37
Analogous Regimes 38
Key Benefits 38
V. CONCLUSION 43
APPENDIX A : IMPORTANT DEFINITIONS 44
APPENDIX B: REPORTS OF VOTING SYSTEM ISSUES 46
Available in the online version of the report at www.brennancenter.org.
APPENDIX C: DUPAGE COUNTY ELECTION SUMMARY 47
ENDNOTES 48
Brennan Center for Justice | 1
 
Failed voting machines, frustrated voters and lost votes: these have been a constant in news reports
following every recent major election cycle. at should not be surprising. e voting systems
1
used in
the United States today are complicated machines; each runs on tens of thousands of lines of software
code. As with automobiles and airplanes, automatic garage door openers and lawnmowers, occasional
malfunctions are inevitable – even after rigorous product testing.
When it comes to system failures, however, voting machines are different from automobiles and airplanes,
and other products, in at least one important respect: for the vast majority of voting systems in use
today, (1) manufacturers are not required to report malfunctions to any government agency, and (2)
there is no agency that either investigates such alleged failures or alerts election officials and the general
public to possible problems (let alone requires voting system manufacturers to fix such problems).
As this report demonstrates, the consequence of this lack of oversight is predictable. Voting systems
fail in a particular county in one election, and then again later, under similar circumstances, but in
a different locale. ese repeated failures disenfranchise voters and damage public confidence in the
electoral system.
e Brennan Center reviewed hundreds of reports of problems with voting systems in the last eight
years, and closely studied fourteen of them. Our study shows that election officials and the public
are often completely reliant on the private companies that sell and service this voting equipment and
related service contracts to voluntarily keep them aware of potential problems with those systems.
As one election official we interviewed noted, vendors are in the business of selling machines, and
often dont have an incentive” to inform present and future customers of certain problems with their
systems.
2
e core thesis of this report is simple: we need a new and better regulatory structure to ensure that
voting system defects are caught early, officials in affected jurisdictions are notified immediately, and
action is taken to make certain that they will be corrected for all such systems, wherever they are used
in the United States.
Based on our review of regulatory schemes in other industries, we are convinced that the focal point for
this new regulatory system must be a clearinghouse a national database, accessible by election officials
and others, that identifies voting system malfunctions that are reported by voting system vendors or
election officials. If this database is going to have any real benefit, voting system vendors must be
required to report all known malfunctions and election officials must have full access to the database.
e Election Assistance Commission (EAC), the relatively new federal agency charged with the task of
creating a testing program for new voting system has, within its limited federal mandate, made great
strides in the last two years increasing quality control for some of the country’s newest voting systems.
However, to fully address the problem of underreported and unaddressed voting system problems, the
EAC or other federal agency should be given statutory authority and resources to fully implement the
kind of database recommended in this report. Such a database would make our electoral system stronger.
It would be easier for election officials and others to ensure that their equipment is as user-friendly and
accurate as possible. It would also make voting machine vendors more accountable to public officials
and taxpayers, incentivizing manufacturers to enhance internal controls. Given the billions of dollars
spent by federal and local governments to purchase and maintain new voting equipment over the last
several years, this is no small thing.
2 | Brennan Center for Justice
core findings
ree fundamental findings result from our study of past reported problems, review of current law and
contracts for the use and regulation of voting systems, and interviews with election officials:
1. ere is no central location where most election officials can find comprehensive
information about problems discovered with their systems before each election.
• Stateandlocalelectionocialsweinterviewedtellusthattheymustrelyalmostexclusively
on the voting system vendors for information about malfunctions, defects, vulnerabilities
and other problems that the vendors have discovered, or that have occurred with their voting
systems in other states.
• Achangeinelectionadministratorscansometimesmeanalossofknowledgeaboutallofthe
potential problems with a voting system as well as procedural safeguards necessary to prevent
those problems.
• ereareapproximately4,600separatejurisdictionsacrosstheUnitedStatesthatadminister
elections.
3
2. Vendors are frequently under no legal obligation to notify election officials or the public
about problems with their systems.
• While purchase or service contracts sometimes bind election ocials to inform vendors
of malfunctions, vendors are not always similarly obligated to inform officials of problems
reported to them.
• Votingsystemvendorsareundernolegalobligationtonotifyanyfederalagencyofproblems
they discover with the vast majority of their systems in use in the United States today, despite the
fact that hundreds of millions of federal dollars have been spent to purchase such equipment.
3. e same failures occur with the same machines, in one jurisdiction or another, election
after election.
• Mostoftheelectionocialsweinterviewedinconnectionwithourreviewofreportedproblems
claimed to have had no prior warning of the issues we discuss. By contrast, in most cases, the
vendors were (or should have been) aware of the problems – often because the same problem
had been reported to them earlier by another election official.
• Frequently, these malfunctions – and their consequence, disenfranchisement – could have
been avoided had election officials and/or public advocates known about earlier problems and
had an opportunity to fix them.
Brennan Center for Justice | 3
central recommendation: creation of a national database
for voting system problems
Given the nature and importance of voting systems to our democracy, we need a new regulatory structure
to ensure that voting system defects are caught early, disclosed immediately, and corrected quickly and
comprehensively. Accordingly, this new regulatory system must center around a mandatory national
clearinghouse, administered by a federal agency empowered to investigate violations and enforce the
law.
Based upon our interviews with election officials and regulatory experts, and our review of analogous
regulatory structures in other important industries, we conclude that the clearinghouse must include
four key elements to work effectively:
1. A Publicly Available, Searchable Centralized Database
Election officials, in particular, would benefit from a publicly available, searchable online
database that includes official (i.e., election official-reported or vendor-reported) and unofficial
(i.e., voter-reported) data regarding voting system failures, and vulnerabilities, and other
reported problems and establishes criteria for the databases contents and organization.
2. Vendor Reporting Requirements
Vendors must be required to notify the appropriate government agency of any known and
suspected voting system failures and vulnerabilities, and other reported problems, including
customer (i.e., election official) complaints, warranty claims, legal actions and/or actions taken
by the vendor to satisfy a warranty or investigate a reported problem.
3. A Federal Agency with Investigatory Powers
e best way to ensure that vendors address potential problems in a timely manner is to empower
the appropriate government agency to investigate all voting system failures and vulnerabilities
listed on the database, grant the agency subpoena power to facilitate its investigations, and
require vendors to, among other things, maintain records that may help the agency determine
whether there are indeed voting system failures or vulnerabilities, and whether the vendor has
taken appropriate action to address the failures or vulnerabilities.
4. Enforcement Mechanisms
e appropriate government agency must have the power to levy civil penalties on vendors
who fail to meet the reporting requirement or to remedy failures or vulnerabilities with their
voting systems.
We detail these recommendations more fully on pages 27 - 38 of this report.
4 | Brennan Center for Justice
additional recommendations
While a national clearinghouse along the lines we suggest in this report is ultimately the best way to
ensure that problems with machines are publicized and corrected throughout the country, there are
important interim steps that county and state governments, in particular, can begin taking immediately
to increase the chances that election officials are notified of problems with their voting systems and can
avoid some of the kinds of problems detailed in this report:
1. Negotiate Better Contracts with Vendors
Provisions in many voting machine contracts make it much more difficult for election officials
and the public to get detailed information about system problems reported in other parts of
the country, or to hold vendors responsible for problems when something goes wrong. To
increase voting system reliability and maximize vendor motivation to minimize the risk of
such problems, counties and states should begin demanding certain key contract terms. Pages
39 - 40 of this report discusses these more fully.
is recommendation is particularly relevant to jurisdictions using Premier voting systems.
ES&S recently purchased Premier, and pursuant to the proposed Final Judgment for the
antitrust action brought by the Department of Justice in March 2010, customers using Premier
equipment will have the option of choosing between ES&S and Dominion for future service of
those machines.
4
is will provide them with an opportunity to negotiate new contracts.
2. Implement Stronger State Regulation
e legislature in at least one state, California, has passed legislation requiring vendors selling
systems within its borders to notify the Secretary of State and all local election officials using
its systems of any defect, fault or failure” within 30 days of discovery.
5
As of the writing of
this report, the legislation is currently awaiting a decision by the governor, who had vetoed an
earlier version in 2009. In 2005, North Carolina passed a similar bill into law.
6
e California
model presents the best legislative attempt we have seen, to date, to address the problems we
discuss in this report. We hope more states will adopt this model.
3. Create a Voluntary Database
e appropriate federal agency should create a searchable database to which election officials,
vendors, and voters could voluntarily report problems. Absent action by the federal government,
a non-governmental organization (like the National Association of Secretaries of State) or even
a state government could create such a database.
ere would be no way to force vendors to report to this database, or to provide election officials
with whistleblower protections for making voluntary reports – two important suggestions for
the mandatory clearinghouse detailed in this report – but it could still serve as a useful interim
resource for election officials.
Brennan Center for Justice | 5
4. Pressure Vendors to Voluntarily Post Information on eir Own Sites is Year
One drawback of the three previous recommendations is that they probably cannot be
implemented in time for this fall’s election. In contrast, vendors could create their own databases
relatively quickly, significantly reducing the risk of embarrassing problems. Ideally, vendors
would create a central, easily accessible and searchable site where election officials could review
all previously issued product advisories, software patches and workarounds, election official
complaints, warranty claims, and lawsuits about their systems (together with the result of any
vendor investigation, explanations, and actions taken to address these complaints).
County and state officials can and should demand this voluntary action from vendors now, in
time to make a difference for Novembers election.
6 | Brennan Center for Justice
Since 2002, the Federal and State governments have invested billions of dollars in new voting equipment,
transforming the way our nation conducts elections and tallies votes. is has had many positive effects.
We have replaced many outdated and unreliable systems. Most political scientists agree that the new
equipment has dramatically reduced the kinds of voter errors common in Palm Beach County in
2000,
7
and, advances in technology have made it possible for many disabled voters to vote privately and
independently for the first time in their lives.
But the change has also given an even greater role
in our elections to the private companies that
manufacture voting machines. e new voting
systems run on tens of thousands of lines of
proprietary software code. Voting machine vendors
create these systems, program, and maintain them.
More than ever, election officials and the public must
rely on private companies to ensure that citizens
votes are recorded as they were intended to be cast,
and that they are counted correctly.
is report details the consequences of lack of
regulation and oversight of the voting machine
industry. Voting machine manufacturers unlike
many other kinds of manufacturers selling products
in the United States are not required to report
malfunctions of most of their systems to any government agency. Nor is there a government agency
that either investigates mechanical failures or alerts election officials and the public to possible problems
for most systems (let alone requiring voting system manufacturers to fix such problems).
While there has been an increase in government oversight of voting systems in the very recent past
and in particular for new systems introduced since 2009 we conclude that the current process for
publicizing and addressing voting system defects nationally is inadequate.
e Brennan Center closely studied 14 reports of voting system problems during the last few years. In
most of these cases, the reported problems resulted in the temporary or permanent miscount or loss of
votes. e numbers range from a few dozen to tens of thousands, but in all cases better oversight and
reporting requirements could have prevented the problems from occurring at all.
e report that follows is broken into three main sections: first, we describe the law and regulatory
structure as it currently exists for addressing voting system failures; second, we document the need to
fix this regulatory scheme by providing selected examples of its current failures; and finally, we offer
suggestions for changes to the law and regulatory structure that would redress the systems current flaws,
based largely on models that have proven successful with other commercial products.
i. 
“i adamantly support the
recommendation of the creation
of a national, searchable database
that election officials could use as
reference to voting systems.”
jane platten, director of the cuyahoga
county board of elections, ohio’s
largest election jurisdiction
Brennan Center for Justice | 7
e Help America Vote Act of 2002 (HAVA) resulted in the replacement of voting systems across the
country. It also created new standards for the certification and use of these systems. It established the
EAC as an independent agency of the federal government and charged it with the task of creating a
testing program for the new voting systems and holding hearings and functioning as a clearinghouse
for election administration information, among other things.
8
Section 202 of HAVA states in relevant
part that “[t]he Commission shall serve as a national clearinghouse and resource for the compilation of
information and review of procedures with respect to the administration of Federal elections . . . .
9
Some argue this clearinghouse function should include
reporting on the performance of voting equipment
purchased with funds granted by HAVA.
10
e EAC
has not publicly embraced this interpretation for
systems it has not certified, and there is no question that
its power to oversee voting system manufacturers has
been severely limited by federal statute and resources
provided to it.
In spite of this, as discussed below, the EAC has recently
taken several positive steps to make information about
voting system problems more readily available to election
officials and the general public. While admirable and
important, we believe these steps fall short both in
scope and timeliness of what is necessary to avoid the
kinds of recurring problems detailed in this report. is
belief is in no way meant to disparage recent efforts
made by the EAC to ensure that problems with its
certified systems are tracked and corrected. To the contrary, as discussed more thoroughly in Section IV (A
Better Way to Track and Address Voting System Problems), current federal law does not allow the EAC or any
other federal agency to take many of the steps we recommend to reduce voting system errors. Nor is the EAC
or any other federal agency currently provided with funding necessary to take all of the steps we recommend.
e EAC’s budget in Fiscal Year 2010 was $17,959,000, minus a $3,250,000 pass through to the National
Institute of Standards and Technology for a total of just $14,709,000.
11
e EAC is in the midst of drafting of a new clearinghouse policy,
12
which will be subject to public comment
and approval by the EAC’s commissioners. Jeannie Layson, Director of EAC Communications and
Congressional Affairs, has recommended a pilot program limited in scope and duration to allow the EAC to
determine resources needed to operate the new clearinghouse.
13
e Brennan Center has asked the EAC to
comment on the extent of its powers and obligations under the clearinghouse provisions of HAVA. e EAC
has declined to state whether its new clearinghouse policy will require more reporting on the performance of
voting equipment purchased with HAVA funds pending final adoption of that policy.
14
However, in the past,
the EAC has taken the position that it does not have the authority or resources to track and resolve problems
associated with voting systems it has not certified
15
which, as discussed below, represents nearly all of the
voting systems in use in the United States today.
Separate and apart from its soon-to-be released clearinghouse policy, the EAC has recently adopted a number
of important reporting requirements for both voting system manufacturers and testing labs that participate in
a meaningful and useful clearinghouse
function is particularly appropriate as
a federal responsibility. it is much more
effective for a single federal agency
to have primary responsibility for
identifying voting system problems and
to recommend remedial action.”
douglas kellner, co-chair of the
new york state board of elections
ii. the current process for publicizing and addressing
voting system defects
8 | Brennan Center for Justice
its newly established Voting System Testing and Certification Program.
16
Pursuant to the Quality Monitoring
Program established in the Voting System Testing and Certification Program Manual (the VSTCPM”) the
EAC will post on its website “test reportsfor all systems tested for EAC certification, regardless of whether or
not they are ultimately certified. ese test reports will include a list of discrepanciesidentified during the
testing.
17
It will also post information related to site audits that it conducts on manufacturers who participate
in its program.
18
Under the VSTCPM, vendors must report to the EAC malfunctionsof EAC certified systems. e
VSTCPM defines malfunctionas a failure of a voting system, not caused solely by operator or
administrative error, which causes the system to cease operation during a Federal election or otherwise
results in data loss.
19
e EAC will also post this information on its website. e EAC recently
informed the Brennan Center that it intends to post a map showing all jurisdictions that use EAC
certified systems, with links to all vendor reported anomalies for such systems.
20
Finally, of relevance
to this report, election officials may voluntarily report anomaliesfor such systems if they result “in
some disruption to the election process,provided the election officials provide their name, title, and
jurisdiction, among other information.
21
is new system had two recent important public successes. e first occured on June 25, 2010, when
the EAC put out a Voting System Technical Advisory(VSTA) for the ES&S Unity 3.2.0.0 system,
which has been certified by the EAC. e advisory came two months after Jane Platten, Director of
the Cuyahoga County Board of Elections, notified the EAC that during testing of the machines prior
to a May primary election, approximately 10 percent of the machines started powering down and then
freezing.
22
After extensive consultation with both ES&S and Cuyahoga County, the VSTA was sent
to election officials using the same system, advising them what steps to take in the event this freeze or
power failure occurred during opening or closing of the polls, or during voting.
23
On August 23, 2010, the EAC issued a VSTA for the MicroVote EMS 4.0B, noting that the voting
panel for the systems Direct Recording Electronic device would not operate with certain flash cards.
24
While the recent steps by the EAC are unquestionably valuable, there are a number of factors which limit the
usefulness of this reporting system. ey are discussed in greater detail in Section IV (A Better Way to Track
and Address Voting System Problems) of this report. A summary of some of the most serious limitations of the
current system follows:
• Perhapsmostimportantly,theEAConlycertieditsrstvotingsysteminFebruary2009 meaning
that almost none of the machines currently in use in the United States are covered by VSTCPM
reporting rules, or any federal reporting requirements, for that matter. Of the approximately 4,600
election jurisdictions in the United States, we are aware of only a few dozen
25
that will use EAC certified
equipment in 2010. In other words, approximately 99 percent of U.S. jurisdictions in 2010 will be
using equipment that is not certified by the EAC and therefore not covered by this program.
• AsmostpollingplaceequipmentinuseintheUnitedStateswaspurchasedafter2002,andbecause
many jurisdictions replacing equipment are likely to continue to use non-EAC certified equipment
in the future, we expect it could be decades before even a large majority of jurisdictions in the United
States are using EAC certified systems.
26
In fact, only twelve states require federal certification for
new systems, so absent changes at the state level it is not certain that the EAC’s program
will ever cover most jurisdictions in the United States.
27
Brennan Center for Justice | 9
• MandatoryreportingbyvendorsisrequiredonlyiftheEAC-certiedsystem“malfunctioned”
during a federal election. us, if a vendor becomes aware of a problem that occurred when
there were no federal candidates on the ballot, it is apparently under no obligation to report
the problem to the EAC.
• Reportingunderthissystemislimitedtovendorsandelectionocialsforaveryspecictypeof
problem. For instance, it is not clear that manufacturers would have to report potential flaws they
discover before they result in actual loss of votes on Election Day, or “merely” because they cause
delay and long lines rather than a loss of data.
• Independentinvestigatorsandvoterswithcrediblereports,nomatterhownumerousorserious,
are not entitled to report problems.
• Evenwherecountyelectionocialsvoluntarilyprovideanomalyreports(exposingthemselves
to potentially unhappy vendors, as discussed on pages 25 - 26), the EAC is not required
to provide this information to other users of such systems unless various criteria are met,
including verification from “the relevant State’s chief election official.
28
• SomeelectionocialshavecomplainedthatneithertheEACnorthevendorsarerequiredto
notify election officials immediately upon learning of a malfunction. Douglas A. Kellner, co-chair
of the New York State Board of Elections, in a letter to the EAC praising them for issuing their
first Voting System Technical Advisory last June, noted that it came two months after the EAC
was first notified of the problem and urged the EAC to put in place a system that would allow an
immediate preliminary notice to be distributed to all jurisdictions using the equipment involved
as soon as EAC staff has been able to verify a report.
29
For these and other reasons, most state and local election officials we interviewed tell us that they must still rely
almost exclusively on the voting system vendors for information about malfunctions, defects, vulnerabilities
and other problems that the vendors have discovered, or that have occurred with their voting systems in other
states. Vendors are frequently under no legal obligation to provide such information. While purchase or
service contracts sometimes bind election officials to inform vendors of malfunctions, vendors are not always
similarly obligated to inform officials of problems reported to them.
30
As Jane Platten put it, “One of the
more frustrating aspects of encountering problems [with voting systems], often while preparing and testing
for elections as well as on election day or during tabulation, is that the vendors themselves often know about
the problems and never disclose any details whatsoever prior to the moment of crisis.
31
Of course, vendors do frequently notify election officials of problems when they occur, and often provide
software patches or other procedural safeguards to ensure that such problems do not occur in the future.
Unfortunately, in at least some instances, vendors have appeared slow to acknowledge such problems.
32
More to the point, there is no centralized location where election officials can find information about anomalies,
malfunctions, usability concerns,
33
and other problems discovered with systems they are currently using
before each election. A change in election administrators can sometimes mean a loss of knowledge about all
of the potential problems with a voting system as well as procedural safeguards necessary to prevent those
problems.
34
e result, as this report demonstrates, is that all too frequently the same failures in the same voting
systems occur in one jurisdiction or another, election after election. Often, these malfunctions and
their consequence, disenfranchisement – would have been avoided had election officials and/or public
advocates known about previously encountered problems and had an opportunity to fix them.
10 | Brennan Center for Justice
Press reports from the last several years contain hundreds of reported cases of voting machine
malfunctions. A subset of these cases is summarized in Appendix B of this report (available in the online
version of this report). News items about voting system troubles tend not to include many details; this
makes it hard to identify from these reports the precise cause of a particular malfunction. Whatever the
causes of a particular problem, it is fair to assume that their occurrence in one jurisdiction will often
eventually be repeated in another unless election officials throughout the country are made aware of
both the causes of the problem and how to avoid them.
Of the hundreds of reports of voting system malfunctions and vulnerabilities, we collected and closely
studied fourteen. ey are summarized below. Most of the election officials we interviewed in connection
with these summaries claimed to have had no
prior warning of the problems we discuss. By
contrast, in most cases, the vendors were (or
should have been) aware of the problems often
because the same problem had been reported to
them earlier by another election official.
1. Butler County, Ohio, March 2008
In March 2008, as they reconciled vote totals
from the State primary in their offices Data
Department, Ohio officials noticed that several
votes were dropped from memory cards even
though their final report stated that votes
on these memory cards were counted.
35
A
subsequent investigation by Ohio election
officials determined that at least 1,000 votes
were undercounted in nine of Ohios forty-four
counties using Premier touch screen or optical
scan voting systems.
36
In an editorial several months later, the New York Times noted that Premier
(known as Diebold Election Systems prior to rebranding in 2007) had subsequently notified more than
thirty states using its systems “to be on the lookout for missing votes.
37
Less widely reported was the fact that this same problem was apparently discovered in DuPage County,
Illinois in 2004. In a county election summary (obtained by the Illinois Ballot Integrity Project and the
relevant portions of which are annexed to this report as Appendix C), a technician who serviced the
machines noted what appears to be the very same problem:
GEMS Upload Failure on York 58 is memory card had a failed upload transmission
on election night that was not detected until the next day when reports were on the
precinct, and zero results were found for each race within the precinct. e status of
the memory card upload within the GEMS was successful” but the upload record
showed the ballot count to be zero. It is rather discomforting [sic] that this failed
transmission was not detected on election night.
iii. failures of the current system: case studies
“one of the more frustrating aspects
of encountering [voting machine]
problems . . . is that the vendors
themselves often know about the problems
and never disclose any details whatsoever
prior to the moment of crisis.”
jane platten, cuyahoga county board of
elections
Brennan Center for Justice | 11
e publicity around the problems in Butler County, Ohio in March 2008 may have saved thousands
of votes on Election Day the following November. It is impossible to know how many votes were lost
before the problem was so widely publicized.
Nor was the mere reporting of the problem to the vendor in 2008 enough to guarantee that the 29 other
States using this system that year would have known how to protect themselves from similar problems.
As the rest of this case study shows, it was the extreme vigilance of the Butler County Board of Elections
and the Ohio Secretary of State that resulted in the full scope of the problem being revealed.
On April 4, 2008, the Butler County Board of Elections sent a letter to Premier and copied the Secretary
of State, Jennifer Brunner, notifying Premier of the problem.
38
e Board sent a follow up letter to
Premier on April 9, 2008 notifying them of a recurrence of the problem.
39
On May 16, 2008, in response to Butler Countys complaint, Premier issued a report that blamed the
problem on antivirus software the county had run on their system as well as human error.
40
County Election Director Betty McGary reports that on May 23, she wrote to Dave Byrd, President
of Premier, calling their report “highly speculative,and rejecting their assumptions. She states that she
requested Premier continue to research and diagnose the root source of the discrepancies.
41
Had Butler Countys Board of Elections been less persistent, that might have been the end of the story. Other
election officials using this system around the country might not have learned of the problems experienced in
Butler County, and almost certainly would not have discovered its true cause.
Fortunately, the Butler County Board asked the Ohio Secretary of States office to assist it in its own
investigation of the problem. On August 6-7, 2008, Butler County election officials and the Ohio
Secretary of State conducted a simulation of the vote counting process with Premier observers. ey
conducted eight of these simulations over two days in some cases disabling the antivirus software
Premier had blamed for the malfunction, in other cases enabling it.
42
e testing revealed that the machines dropped votes during multiple memory card uploads from
individual voting machines onto the county server regardless of whether the antivirus software was
enabled.
43
After the testing, Premier conceded that the apparent root cause for the problem was an error with their
server software, which the company determined contains a logic error” that can sometimes result in
dropped votes from a sharing violation when multiple cards from individual machines were uploaded
at the same time.
44
Following its additional investigation, Premier sent a product advisory to all counties using its systems
detailing procedures intended to “mitigate and reveal this issue should it occur.
45
Director McGary supports a mandatory requirement for “voting machine vendors to report all
malfunctions and complaints they receive from election officials to a central and searchable database,
noting that “such reporting should be mandatory.
46
12 | Brennan Center for Justice
2. Humboldt County, California, November 2008
In November 2008, election officials in Humboldt County, California implemented a post-election
“Transparency Project,whereby a separate scanner not manufactured by the voting machine vendor
electronically counted every paper optical scan ballot during the election. e purpose was to verify the
official vote totals and to post ballot images on the internet in order to allow any member of the public
to conduct independent recounts.
47
e Transparency Project turned up a counting error on Humboldt County’s voting machines: they
failed to count approximately two hundred ballots.
48
According to Humboldt County Clerk Carolyn
Crnich, the first batch of absentee ballots scanned into the voting system, known as deck zero,
disappeared from the totals produced by the voting system before officials finished scanning all of
the ballots and certified the vote totals.
49
Upon learning of the problem, Crnich contacted the voting
system vendor.
50
Crnich states that after examining copies of the countys
database, the vendor told her that a programming error
in its election management system, the software used
to aggregate the votes from all of the countys voting
machines, caused the problem.
51
Wired and Computerworld magazines have reported
that the voting system vendor was aware of the deck
zeroproblem for years, but did not notify the Election
Assistance Commission, the National Association of State
Election Directors, or the California Secretary of State,
Californias chief election official.
52
Instead, according to
a report issued by California Secretary of State Bowen
after the Humboldt County incident came to light,
the vendor sent a vague e-mail to election officialsin
California that used the software with the programming
problem, recommending a workaround” procedure
without identifying the problem or the potential
consequences (i.e., lost votes) of failing to implement the
workaround.
53
e voting system vendor has testified that once it first identified the software problem in October
2004, it communicated” its findings, and a simple procedure workaround to mitigate this issue, via
email to all California counties then affected.
54
Carolyn Crnich does not dispute that the vendor may
have informed her predecessor of the problem. She is certain, however, that her predecessor did not
leave any documentation about the problem when she took over, or institute procedures that would
have prevented the problem from causing the voting system to lose votes.
55
Nor did the vendor report the problem to the California Secretary of States office. As the vendor noted in
testimony, at the time there was no “mandate for reporting issues of this nature” to the Secretary of State.
56
wired and computerworld
magazines have reported that the
voting system vendor was aware of
the “deck zero” problem for years,
but did not notify the election
assistance commission, the national
association of state election
directors, or the california
secretary of state, california’s chief
election official.
Brennan Center for Justice | 13
Humboldt County Election Director Crnich has stated that if there were an EAC database with
information detailing problems that other counties had experienced using the same voting system used
in Humboldt County that she could have accessed before the November 2008 election, she almost
certainly would have used it. Such a database would have alerted her to the programming issue with her
countys tally server, as well as the workaround. Crnich stated that she believed it was well within the
EAC’s mandate to provide this kind of information to local election officials and that new officials, who
might have little experience with the systems they are charged with using, would particularly benefit
from this kind of database.
57
3. Orange County, Florida, November 2006
In 2007, the Florida Division of Elections listed Orange County as experiencing the highest undervote
rates in the state on absentee ballots cast in the 2006 general election for both the U.S. Senate race and the
state Governor’s race.
58
Alarmed by the exceptionally high rate of undervoted ballots in a major election
– nearly 5 percent – the Florida Fair Elections Centers Associate Director contacted the Orange County
Elections Administrator, who promised to investigate the issue.
59
According to the Center, Orange County
officials responded to the inquiry by stating that their manual inspection of the ballots confirmed that
some legitimately cast ballots had not been counted. e Center adds that when they questioned the
vendor of the countys OpTech optical scan machines about the problem, the vendors representative
identified the problem as the scannersfailure to read certain types of gel ink used by voters to complete
their ballots.
60
On further investigation, the Center discovered that the same problem seemed to have occurred on
similar scanning equipment in March of 2004 in Napa County, California. In that election, optical
scanners manufactured by Sequoia failed to count some ballots voted with gel ink.
61
is problem was
only discovered during the states legally-mandated hand count of 1 percent of the ballots cast in the
election.
62
Sequoia told Wired magazine that the problem was not with the machines themselves, but
rather with the countys calibration procedures – the machines were calibrated to read only carbon ink,
not dye-based ink found in many gel pens.
63
According to Sequoia, the issue could have been avoided
through more thorough pre-election testing.
64
When the Florida Fair Elections Center delved more deeply into the history of this type of problem,
they learned that in the 2000 election, Orange County’s optical scan machines failed to count more
than 400 votes in the presidential race for no apparent reason.
65
At the time, it was postulated that one
possible explanation for the machines’ failure to count these ballots was “low carbon content in the
ink pens used to mark them.
66
Kitty Garber, Associate Director of the Center, believes that both the
vendor and the state were well aware of this before the time she discovered the issue in 2007 – in part
because the vendor so quickly identified the source of the problem. For some reason, she states, this
was not adequately communicated to the people actually running the elections” in Orange County in
2004 or 2006.
67
Bill Cowles, Supervisor of Elections for Orange County noted in an interview with us that the county
switched to a different model of ES&S scanner after the 2006 general election.
68
Florida has also
implemented a post-election audit law in the intervening years, though a 2008 study by the Brennan
Center and others has been critical of that audit as being insufficiently robust to catch many problems.
69
14 | Brennan Center for Justice
4. Pulaski County, Arkansas, May 2006
During early voting in the May primary, several voters complained of problems with an ES&S touch screen
DRE.
70
According to a local newscast, Pulaski County election officials tested the machine and determined
that the machine was not broken; an optical illusion perceived by voters who were over six feet tall caused
the problem.
71
Officials determined that the angle at which particularly tall voters viewed the screen caused
them to believe that they were voting for the candidate below the one for whom a vote was recorded.
72
is
is a significant problem given that more than 15 percent of American males over the age of 20 are six feet
tall or taller.
73
Pulaski County Director of Elections Susan Inman told the Arkansas Democrat-Gazette that when she
asked ES&S to examine the machine to ensure that there wasnt a problem with the equipment, a company
employee told her that they were already aware of optical illusion problems experienced by tall voters.
74
A review screen that appears before voters finalize their ballots alerted some to the fact that their votes
were not recorded as intended. However, several studies have shown that most voters will not notice
errors on their final review screens, so there is no way to know how many voters in Pulaski County
actually cast their ballots for candidates other than the candidate of their choice.
75
Officials were livid
at the thought that ES&S could have known about the problem and failed to warn them.
76
Pulaski
County Prosecuting Attorney Larry Jegley launched an investigation into the issue, saying, “I cant
understand how in the world a big company like ES&S, with contracts all over the state of Arkansas,
would know about a problem like this and fail to fix it.
77
•••
5. Florida, November 2006
In 2007, Diebold, Inc. conceded that its optical scan readers had a glitch that caused memory card
failures, and told the Daytona Beach News-Journal that it would investigate the “J40 connector” that
attaches memory cards to its optical scan voting machines.
78
is admission came after complaints
about memory card failures from election officials dating as far back as 2000.
79
According to the News-Journal, Volusia County, Florida reported that eleven memory cards in Diebold
optical scan machines failed during the November 2006 general election.
80
Premier told the News-
Journal that the 4.4 percent error rate in Volusia County was unusual,but an investigation by the
paper revealed even higher error rates in other Florida counties using the same equipment.
81
According
to public records obtained by the paper, several other Florida counties experienced failure rates that
were comparable to or higher than those observed in Volusia County.
82
e 2006 incidents were not the first time that memory cards in Diebold machines failed in Volusia and
other Florida counties. According to the News-Journal, a 2004 county report indicates that Volusia had 57
memory card failures, which Diebold stated was more memory card failures than the rest of our customers
in Florida combined.’”
83
e paper also reported that Volusias problems with memory cards dated back to
the 2000 general election, when 300 ballots went uncounted when a memory card failed in the middle of
ballot scanning.
84
e loss of votes was not discovered until a hand recount began as a result of the close
contest.
85
e News-Journal noted that “Volusia Countys most infamous memory card problem . . . when
more than 16,000 negative votes were recorded against Al Gore,had never been determined.At the
time, a county election official wrote an angry e-mail asking the manufacturer to please explain this so
that I have the information to give the auditor instead of standing here looking dumb.
86
Brennan Center for Justice | 15
Despite this long history of failures with the same equipment, state election officials said in 2007 that
they were previously unaware of the problem.
87
By 2007, nearly 25,000 Diebold optical scans machines were in use nationwide. e News-Journal
reported that the manufacturer conducted a survey of its customers to determine the frequency of such
failures, but refused to release results from the study, calling it proprietary information.
88
According to
the News-Journal, officials at the Election Assistance Commission told the paper that they could not
compel distribution of this information unless an official government agency requested the action.
89
Many saw this as an argument for the EAC to bolster its clearinghouse function. “[T]he federal agency
required by law to act as a clearinghouse on voting system problems the U.S. Election Assistance
Commission has been slow to develop a place where such information can be shared,the News-
Journal reported in 2007, “e [election] supervisors are left largely on their own.
90
•••
6. Broward County, Florida, November 2004
Two days after Election Day in November 2004, Broward County election officials double-checked election
results and discovered that tens of thousands of votes on certain state amendments were not counted.
e problem: a software glitchin the system used
to count the countys absentee ballots.
91
According
to the Palm Beach Post, the software started counting
backward after it logged 32,000 votes in a race.
92
Once officials identified the problem and obtained
correct vote totals, the newfound votes contributed to
a changed result for a statewide gambling amendment
and sparked angry calls for a recount.
93
Several newspapers reported that ES&S, the voting
system vendor, claimed to have noticed the problem
in 2002, and said it notified the Secretary of State’s
office of the issue after that election.
94
It isnt clear from news accounts why Broward County did not
adopt procedures to safeguard against this glitch once it was discovered. Broward County officials told
the Palm Beach Post that the manufacturer claimed its upgrades were rejected by the Secretary of States
office in 2002; the state contested this claim.
95
One reason officials in 2004 may have been unaware of
the problem: there was turnover in the offices of chief election officials in both Broward County and
the State of Florida between 2002, when the software glitch was originally discovered, and 2004, when
the unaddressed problem caused Broward County to miscount the votes.
Regardless of who was to blame for Broward Countys failure to address the problem ahead of time, a
centralized database could have prevented it, by allowing Broward County officials in 2004 to review
reported problems for their systems, including necessary workaround procedures, and avoid the
controversy that followed the well-publicized tallying problems.
the news-journal reported that
diebold conducted a survey of its
customers to determine the frequency
of such failures, but refused to release
the results, calling them proprietary.
16 | Brennan Center for Justice
7. Florida, June 2004
According to the Miami Herald, only five months before the 2004 general election, some state officials
learned that touch-screen voting machines used in 11 of the states counties contained a software flaw that
would make it impossible to conduct a manual recount of ballot images in close races.
96
Election officials
in at least one Florida county knew about the problem as early as 2002, but for whatever reason, the
existence of the flaw was not understood by the relevant State election officials for nearly a full year.
97
Miami-Dade County learned of the problem after an election in May 2003. e division director of
the County’s technology department found that the electronic event log of voting activity scrambled
the serial numbers of voting machines.
98
He wrote a letter to the County elections supervisor on June 6,
2003 stating that “I believe there is a serious ‘bugin the program(s) that generate these reports, making
the reports unusable for the purpose that we were considering (audit an election, recount an election
and, if necessary, use these reports to certify an election).
99
e vendor of the machine assured all parties that the software flaw would not affect the counting of
votes. Nevertheless, there was concern that if counties were ordered to produce a record of the votes in
a close race for the purpose of conducting a recount, some of the relevant data could be lost.
100
Press reports indicate that, at least initially, the media attention to the flaw in June 2004 led to a
round of finger-pointing among Florida election officials, with the Florida Secretary of State “blasting
Miami-Dade officials for failing to notify her office when they learned of the problem a year earlier,
and Miami-Dade officials, arguing that other counties that discovered the same problem should have
notified the state, to put more pressure on the vendor to “come up with a so-called work-around to the
problem before the mistake was repeated.
101
Again, a centralized database that listed reports of problems from vendors and election officials would
probably have provided election officials in Florida with much earlier notice of the problem.
•••
8. Alameda and San Diego Counties, California, March 2004
According to the San Diego Union-Tribune, on the morning of the March 2 primary election, more than
700 Diebold precinct control modules that activate the cards used to call up ballots on touch screen
machines displayed the wrong start-up screen.
102
With no way to load ballots onto the voting machines,
hundreds of polling sites had to delay opening their doors, some by as much as three hours.
103
Some
voters told the Union Tribune that they had to leave before getting the opportunity to cast a ballot.
104
Shortly after the primary, a Diebold spokesman acknowledged that the start up screen on precinct
control modules could fail in the event of a problem with the unit’s power supply, calling the glitch “a
possibility […] but it was an improbability.
105
A report released by the company six weeks later revealed
that the problem was caused by faulty power switches that failed to fully turn off the units when placed
in the ‘off’ position, causing power to drain from the machines before election day.
106
e Secretary of State’s Voting Systems and Procedures Panel called a hearing in late April to examine
the problems experienced during the primary. At the hearing, former Diebold technician James Dunn
testified that the problems with batteries losing power were evident before the machines were shipped
to San Diego and Alameda counties.
107
Brennan Center for Justice | 17
e technician, whose job was to assemble voting machines, load their software, and pack them for
shipment, testified that battery problems could lead to incidents like those seen on primary day:
We had a significant amount of problems with the batteries. In fact, one of the things we
were told one of the last things we were to check was before they packed up, was that
they were supposed to have 60 to 70 percent battery load in them due to the problem of
the batteries discharging once they reached anywhere from 20 to 15 percent charge rate,
they would then dump the settings, sometimes dump the software load, and then on initial
startup, would being up a standard Windows CE screen and not the Diebold screen
[is occurred] [f]requently. All the time.
108
Calling the disaster in San Diego and Alameda counties “predictable” and the problem “fully known,
Mr. Dunn testified that he notified supervisors of the problem and was told that the company knew
that machines encountered this problem once the batteries discharged to a certain point, and that their
solution was simply to ensure that the machines were shipped with a sufficient charge.
109
At the hearing, an attorney for Diebold contested the accuracy of Mr. Dunns testimony in vague terms,
but of the battery problems on Election Day, company president Robert Urosevich said, We were
caught. I apologize for that.
110
•••
9. Bernalillo County, New Mexico, November 2002
Ten days after Election Day in 2002, Bernalillo County Commissioners discovered that their electronic
voting system reported approximately 36,000 votes even though nearly 48,000 voters had signed in at
the polls. As reported in the Albuquerque Tribune, the vice president and regional manager of the voting
system vendor stated that the individual touch-screen machines recorded the votes correctly, but the
software program used to [aggregate] all the votes,” did not have the capacity to handle the totals and
was overwhelmed by the data.
111
e result was that nearly 12,000 votes were missing from the totals
produced by the voting system.
In fact, the very same problem occurred weeks earlier in Clark County, Nevada and was fixed for
future elections.
112
Unfortunately, according to the Albuquerque Tribune, the technician in charge of
Bernalillo County’s problems was not told of the Clark County problems, and was not provided with
the patch.
113
James Noel, who served as counsel to one of the candidates on the ballot that day, discovered the
problem several days after the election.
114
According to Mr. Noel, as he reviewed the unofficial results,
he noticed undervote rates of 20 to 25 percent for the early voting period for statewide and federal
offices. is was higher than the undervote rate in down-ticket races, not something that one would
typically expect. He estimated that thousands of votes might not have been counted.
115
Mr. Noel stated that he brought this anomaly to the attention of the County Clerk, and that despite this
fact, she recommended final certification of the results several days later.
116
Mr. Noel objected to certification,
pointing out the unusually high undervote rate in statewide and federal races, and the board voted to delay
certification pending investigation by the County Clerk.
117
When told of the problem, the vendor re-ran the
results “using the software patch this time,” and issued a new report that included the missing ballots.
118
18 | Brennan Center for Justice
Denise Lamb, who currently serves as Chief Deputy Clerk for Elections in Sante Fe County, New Mexico,
believes that a central database that detailed malfunctions for each system, as well as workarounds or
software patches supplied by the vendors, could have prevented the problems that Bernalillo County
encountered with its tally server in 2002, and certainly would have allowed the County to understand
quickly the potential source of the malfunction once it occurred. She noted that, vendors are in the
business of selling machines, and often dont have an incentiveto inform present and future customers
of problems with their systems.
119
•••
10. Wake County, North Carolina, November 2002
According to Wired News, ES&S discovered a glitch in the firmware of its touchscreen voting machines
used during early voting in the 2002 general election in Jackson County, North Carolina.
120
e glitch
made the ES&S machines falsely sense that their memories were full,” a company spokeswoman told
the magazine.
121
e potential result of this error was that memory cards associated with the machines
would not record votes that had been cast. Fortunately,
the problem was fixable.
122
Election officials in neighboring Wake County later
found this same glitch “by chance” during their own
early voting period that year.
123
Election officials told
Wired that at the time, early voters would fill out paper
applications which contained tracking numbers. Each
application had a tracking number, and before the early
voters cast their votes on the touch-screen machines,
poll-workers typed the number into the machines. At
some point, election officials compared the number of
votes on the machines to the applications, and found
that the two figures did not match.
124
As the Brennan
Center and other organizations have documented, even
today many election jurisdictions do not always follow
such reconciliation practices.
125
According to Cherie Poucher, Director of the Wake County Board of Elections, upon learning of the
discrepancy, she immediately contacted ES&S. She says that at that point, she was told that Jackson
County had experienced a similar problem. Poucher stated this was the first time she had been informed
of this problem.
126
Ms. Poucher stated that in all, six touch-screen voting machines used in Wake County had lost 436
ballots as a result of the problem. Because the county had paper applications and a numerical code
associated with the lost votes, they were able to contact voters whose votes had been lost, and provide
them with the opportunity to revote. Many did so.
127
Unfortunately, as Professor David Dill of Stanford
has noted, we cant be sure “that other counties didnt lose votes that they didnt catch.
128
denise lamb, who currently serves as
chief deputy clerk for elections in
sante fe county, new mexico, noted
that, “vendors are in the business of
selling machines, and often don’t
have an incentive” to inform present
and future customers of problems
with their systems.
Brennan Center for Justice | 19
accessible voting systems
e Help America Vote Act (HAVA) requires that every polling place used for federal
elections be equipped with a voting system that is accessible for individuals with disabilities
. . . in a manner that provides the same opportunity for access and participation (including
privacy and independence) as for other voters.
129
Many disability rights advocates rightly
hail this provision of HAVA as a civil rights milestone, providing millions of Americans
with the opportunity to vote privately and independently for the first time in their lives.
Unfortunately, however, reports indicate that accessible systems too often malfunction
on Election Day, frustrating voters with disabilities.
130
As a result, HAVAs mandate has
sometimes gone unfulfilled, and these voters have been forced to either seek assistance and
lose their privacy while voting, or to give up on voting in their polling places altogether.
Over the past several years, there have been several individual
131
and institutional
132
reports
detailing problems that voters with disabilities have experienced using these systems.
133
ey have included audio keypads and output that failed to work, VVPAT printers
on accessible units that malfunction, and accessible machines that cannot read ballots
correctly. Frequently these malfunctions could have been avoided if poll workers or
election officials were aware of procedures to prevent them.
e appendix available in the online version of this report details some of the defects
that have threatened to disenfranchise voters with disabilities, but a comprehensive
collection of reported malfunctions in accessible voting systems does not exist in one
location. ere is no easy way for election officials, disability rights advocates or voters
with disabilities to review a comprehensive list of problems associated with these systems,
or the countermeasures that election officials can implement to avoid them.
A centralized database that allowed users to search these kinds of problems could greatly
improve the voting experience of voters with disabilities. Such a database would not
prevent malfunctions. But it would give election officials significant knowledge, so that
they could take steps to prevent malfunctions, or quickly correct them. It would also give
election officials – and voters – the opportunity to warn their counterparts in other areas
of the country about problems experienced in various polling places with accessible units
and to remedy such problems prior to Election Day.
20 | Brennan Center for Justice
Ms. Poucher has stated that it would be fantastic” if the EAC or other federal agency would establish a
searchable database that would allow election officials to see what kind of problems other jurisdictions
around the country had with the same systems she was using, whether the vendor had provided them
with patches or other fixes, and what procedures they were using to prevent similar problems in the
future.
134
She noted that if she discovered a problem during a small, off-year election, it might draw
little notice in her own county, but an EAC database would provide her with an opportunity to inform
other election officials throughout the country using the same system of the malfunction, and prevent
more damaging problems down the road. “I think this kind of database would get integrity back in the
system,” she added.
135
Additional case studies further suggest that vendors are too often slow to acknowledge problems
with their systems and frequently do not cooperate as fully or as transparently as public officials (and
members of the public) would like when problems are confirmed. At the same time, because the last
few years have seen a number of voting system vendors go out of business or get bought out by rivals,
election officials may not have any vendor to turn to for explanation when a problem occurs (see Case
Study 11 below and e Suboptimal Structure of the Voting System Market at pp. 25 - 26). is is why
we believe it is critical that vendors be required to report problems to the clearinghouse within a certain
time period when certain events occur.
136
•••
11. Fairfax County, Virginia, March 2009
In March of 2009, during the post-election canvass for a closely contested special election to fill a
vacancy for the County Board of Supervisors, election officials and observers noticed that the combined
totals for two AVC WinVote DRE voting machines in a precinct showed a total of 359 votes cast, with
377 votes recorded for the Republican, 328 for the Democrat, and eighteen for other candidates, for
a total of 723 votes – or 364 more votes recorded than cast.
137
Officials in Virginia were lucky to catch
this problem. As the Brennan Center and others have shown, several jurisdictions in the United States
have inadequate ballot accounting and reconciliation practices.
138
In such cases, it is possible to miss
this kind of error.
e post-election canvass showed that while one machine recorded 723 votes, only 707 voters had
signed-in to that precinct on Election Day, and 348 votes had been cast and correctly counted on
another machine in the same precinct. Election officials decided to print the “ballot images(or the
softwares digital representation of the ballots cast by voters), and add those up by hand. When they did
that, the number of votes on the two machines combined equaled the number of voters who checked
in at the polls, and for each machine the number of voters equaled the sum of the number of votes for
each candidate. It was only by checking the number of votes against the poll books and by using an
alternative method of vote counting that officials were able to determine the result with some certainty.
It is important to note, however, that while the ballot images correlated with the number of voters who
used the machine, the WinVote does not produce a paper trail; the ballot images are created by the
voting system itself, not by voters.
While Fairfax County officials were satisfied that ballot accounting redundancies allowed them to
determine accurate election results, they were never able to determine the cause of the problem with
certainty.
139
ey surmised that the problematic machine did not reset properly after pre-election
testing, but they are not sure why this occurred in 2009 after years of using the same machines without
Brennan Center for Justice | 21
incident.
140
Because the manufacturer of the system was no longer in business, the county had to track
down a programmer familiar with the system to corroborate its theory.
141
County Election Manager
Judy Flaig acknowledges that conducting the type of investigation necessary to determine the cause
of the problem to even this limited extent would not have been possible for every county, and that as
much as jurisdictions try to communicate with manufacturers and one another, some information falls
through the cracks.“Most jurisdictions dont have the resources to do this kind of work,Flaig said,
something like [a database of voting machine problems] would really be helpful.
142
Rokey Suleman, Fairfax County Registrar at the time of the incident, says he still does not know
what caused the problem. Although he no longer works in Fairfax County, he believes other officials
using these machines “should certainly want to know what happened, so that they can put the proper
procedures in place to ensure it doesnt happen again.
143
Hinds County, Mississippi also uses the AVS WinVotes. Suleman noted that unless they read a tiny
Vienna, Virginia newspaper where this story was reported, theres no way they would know about this
problem.
144
Indeed, the Brennan Center contacted Hinds County Supervisor Robert Graham, and he
had never heard of the problems in Fairfax County.
Suleman, who is now the Executive Director of the District of Columbias Board of Elections and Ethics,
noted that in addition to providing election officials with notice of potential problems with the machines
they are using, a central database would have other benefits for election officials. Mr. Suleman stated that
Washington, D.C. “just went through a procurement process for new voting machines,and that he
would have been greatly helped by a central database that could serve as a repository to let me know what
issues exist with the machines, rather than having to rely on what the vendors spoon-feed me.
•••
12. District of Columbia, September 2008
A District of Columbia Council investigation after the Districts 2008 primary found that vote totals
originally produced by a Sequoia tally server on election night were obviously inaccurate.
145
e
Board of Elections traced the problem to a cartridge for a Sequoia precinct-count optical scanner in one
precinct, which reported voter turnout nearly twice that of the registered population of the precinct and
showed 1,554 write-in votes in a race without a write-in campaign.
146
In two reports in the Washington
Post, election officials indicated that the malfunctioning cartridge caused other problems with the
preliminary vote totals.
147
According to acting Executive Director of the D.C. Board of Elections Sylvia
Goldsberry-Adams, one defective cartridge caused vote totals to be duplicated into multiple races on
the summary report issued by our office;” on this summary report, 1,542 appeared as the number of
overvotes in five contests.
148
e District Board of Elections asked Sequoia to explain the problems. In an initial response, Sequoia
stated that it found “no anomalies or irregularities in either the data or the internal event logs that can
be identified as having caused or contributed to the issue experienced on election night.
149
In response
to the Board’s request for a more detailed explanation, Sequoia issued a report that attributed the
problem to human error and ruled out “[e]ndemic hardware and software failures […] as the cause.
150
As for the Board’s request for information about past occurrences of this error, Sequoia responded,
“[s]ince our customers conduct the actual elections not Sequoia we do not have any way of keeping
track of such incidents, nor is it our responsibility to do so.
151
22 | Brennan Center for Justice
Sequoia identified four possible causes for the problem, including a transient malfunction of the
memory pack reader, which transmits information from the memory cartridge to the tally database,
but stated that the voting system event logs would not record any of the possible malfunctions, making
it impossible to provide a more definitive answer.
152
Nine days after the September primary, the D.C. Council subpoenaed information about the voting
machines source code from Sequoia so that it could conduct a more thorough investigation.
153
According to the Washington Post, Sequoia objected to this request on the grounds that it constituted
trade secrets or otherwise protected material.
154
e vendor also objected to the Councils request for all
documents related to any irregularities in similar voting systems, stating that it had no documentation
of such incidents.
155
e Post reported that when the Council persisted in its attempts to get the information, the company asked
for a $20 million bond to guarantee confidentiality.
156
According to board officials, the company still had
not responded to the subpoena as of late April 2009.
157
In April, the
Council filed a motion in the District of Columbia Superior Court
to attempt to force the company to comply with the subpoena.
158
On June 5, under a protective order from a Superior Court Judge,
Sequoia agreed to release the source code for the voting system.
159
When news of the agreement broke, Councilwoman Mary Cheh
said, “they fought us tooth and nail until now.
160
e findings of the District of Columbia investigation will soon
be made public. In the interim, the Council passed election
reform legislation that includes a warranty provisionrequiring
any vendor that sells voting systems to the District to “[p]romptly
and fully disclose any flaw, defect, or vulnerability in the voting
system of which the vendor is aware or becomes awareand to
remedy the problem appropriately.
161
While this bill will help D.C. election officials get needed information about voting system defects in
the future, Councilmember Mary Cheh still sees a need for a centralized, national database. “It was
difficult for us to get the information we needed in D.C. If we were in a smaller, more fragmented, or
politically divided jurisdiction, it would have been even harder for us to get necessary information in
an expeditious fashion,said Cheh. As it is now, theres little communication between jurisdictions, so
vendors hold all the cards.
162
•••
13. New Jersey, February 2008
According to a public records request for results obtained by researchers at Princeton University,
thirty-eight Sequoia AVC Advantage DRE voting machines in eight New Jersey counties experienced
anomalies during the February 5, 2008 primary election.
163
A county official initially discovered the
problem by comparing machine counter totals with the paper printouts produced by the machines at
the close of the polls.
164
e county alerted other counties, which do not routinely reconcile these two
totals, to the potential malfunction.
165
e Times of Trenton reported that the problem initially appeared
as it is now, there’s little
communication between
jurisdictions, so vendors
hold all the cards.”
district of columbia
councilmember mary cheh
Brennan Center for Justice | 23
to be with the turnout totals while all votes seemed to be correctly recorded, the total number of
individuals who cast votes for each party appeared to be slightly off.
166
Sequoia inspected the equipment
and concluded that the problem was poll worker error, not equipment malfunction.
167
In early March,
the company issued a technical bulletin advising users of the machine on how to protect against this
error in the future.
168
In March, several counties decided to enlist a team of Princeton computer scientists to conduct an
independent study on the equipment used in the February primary. After the counties’ intent to hand
over their voting machines for assessment became known, one of the researchers who was set to conduct
the analysis reported on his blog that he received an e-mail from Sequoia stating that the company
will “take appropriate steps to protect against any publication of Sequoia software, its behavior, reports
regarding the same, or any other infringement of [its] intellectual property.
169
e Star-Ledger reported
that at least one county which subsequently backed the effort received a letter from Sequoia stating
that conducting an independent investigation would violate the licensing agreement between the
vendor and the county, and threatening to sue if the county proceeded with the inquiry.
170
In addition,
that same month, advocates at the Rutgers Constitutional Litigation Clinic issued a subpoena for the
necessary information to conduct an independent analysis, including the machines’ source code, build
tools, operator manuals, and maintenance manuals.
171
According to the Princeton researchersfinal
report, Sequoia vigorously protested” sharing its source code on grounds of defending its intellectual
property and it took “months of litigation” to negotiate a protective order under which Sequoia would
share the information.
172
In May, a Superior Court
judge issued a protective order permitting the team
of Princeton researchers to examine two of the DREs
used in the February primary but preventing the
disclosure of any conclusions or comments” about
the machines resulting from the investigation.
173
In June, after the plaintiffs who issued the subpoena
and researchers conducting the assessment refused
to sign the protective order on the grounds that it
violated their speech rights and academic freedom,
the judge who issued the initial protective order
reversed her ruling with respect to the non-disclosure
of the researchers findings.
174
e results of the
released independent analysis showed the researchers
concluded that on all but one of the thirty-eight machines that malfunctioned during the primary, the
number of votes for candidates of a certain party exceeded the number of individuals who voted on that
party’s ballot.
175
Some machines logged more votes for Democrats than Democratic voters, and others
logged more votes for Republicans than the number of Republican voters.
176
e researchers concluded
that it would be “easy and natural” for poll workers to make the mistake that triggered the programming
error that produced incorrect vote totals. Some voters were effectively disenfranchised by this error. ose
who received the wrong partys ballot could not choose a candidate of their own party as was their legal
right, and write-in votes for their chosen party were not counted because it is unlawful for a voter to vote
in the primary election of a party to which she does not belong.
177
Furthermore, the researchers identified
serious insecurities in the machines, and stated that the machines could be quickly and imperceptibly
hacked to steal votes by anyone with “only ordinary training” in computer science.
178
officials in montgomery county,
pennsylvania, an avc advantage
county located less than fifty miles
from the new jersey border, told
the philadelphia inquirer that they
were unaware of the problems that
had occurred in new jersey.
24 | Brennan Center for Justice
Had advocates and researchers in New Jersey not been persistent in their efforts to overcome Sequoias
resistance to a thorough and independent investigation of its machines, these flaws may never have come
to light. Indeed, even given the significant media attention that the incident received, other users of AVC
Advantage machines were unaware of the malfunctions that occurred in New Jersey. In March of 2008, after
county election officials in New Jersey had begun clamoring for an investigation of the machinesbehavior
in the February primary, officials in Montgomery County, Pennsylvania, an AVC Advantage county located
less than fifty miles from the New Jersey border, told the Philadelphia Inquirer that they were unaware of the
problems that had occurred New Jersey.
179
Were Sequoia obligated to report any known flaws in its voting
system to a federal oversight agency, these problems may have been revealed more expeditiously.
•••
14. Indiana, May 2006
Less than two weeks before the May 2006 primary election, voting machine manufacturer MicroVote admitted
to election officials that the voting equipment it had sold to dozens of Indiana counties was uncertified, in
violation of state law.
180
e delay in obtaining certification caused a panic amongst county election officials
who faced the threat of legal action by the state if they used uncertified equipment in the primary.
181
According to testimony and reports in the local media, while MicroVote continued to work toward certication,
on April 22, ten days before the primary, it learned from the independent testing authority contracted to complete
its certification that the companys voting machines would allow some voters to cast votes for candidates who
would not represent them. e testing authority found that the companys Infinity DREs, installed in 47
counties across the state, allowed voters casting straight-ticket ballots in split precincts,where voters living in
the same precincts choose from different sets of candidates, to vote for the wrong set of candidates.
182
In order to pass the certification process in time for the May 2 primary election, MicroVote opted to shut down
the machines’ straight ticket functionality altogether, allegedly at the advice of the independent testing lab.
183
According to the Indianapolis Star, MicroVote worked in secret to develop a software update that would
resolve the problem before the general election, when the straight-ticket function would be necessary
in several split precincts.
184
A sales representative for the company testified that MicroVote installed
the update on all Infinity DREs in the state without notifying the Indiana Election Commission,
applying for certification only after the installation was complete.
185
Election officials only learned of
the problem with the straight-ticket function when MicroVote applied for certification of this update,
nearly four months after the company first became aware of the defect.
186
Fortunately, the straight ticket function is not necessary in primary elections, but Indiana Election Code
requires that certified equipment be functional for both primary and general elections.
187
Perhaps more
importantly in the minds of Indiana election officials, MicroVote appeared to have concealed information
from the Indiana Election Commission for months, and it is unclear what the company would have done
had they failed to come up with a solution before the general election.
188
Upon learning of the glitch and of
MicroVotes prior knowledge of the problem, Indiana Election Commission chair Tom Wheeler said he was
disturbed by [the companys] lack of candor.
189
One year later, an administrative law judge fined MicroVote
over $360,000 for 198 violations of state election law occurring between October 2005 and the 2006 general
election.
190
Brennan Center for Justice | 25
the suboptimal structure of
the voting system market
Discussion of the need for regulatory reform in the voting system market is incomplete
without mention of the market’s suboptimal structure. Purchasing a particular voting
system essentially binds election officials to the vendor who sold them their system for
many years to come. Because officials have extremely limited funds, they are unlikely to
turn to a new vendor when problems arise: the systems (from precinct voting machines
to tally servers) are designed for matched components, which makes it impractical for
officials to replace parts with those produced by a different vendor. Instead, they are
effectively forced to buy an entirely new system and new machines for every polling place
(an exceptionally expensive proposition). In addition to this cost, election officials bear the
additional burden of training election workers, poll workers and educating the public on
the new systems.
At the same time, as a result of contractual constraints and because voting system vendors
generally have monopolies over the production of all replacement parts and exclusive
control over the firmware and software in each system, election officials will generally
remain extremely dependent upon the voting system vendor to address problems and
ensure that their systems are working smoothly. is includes programming their machines,
providing them with software patches, diagnosing and fixing malfunctions, and providing
replacements when systems fail.
Vendors often constrain election officials in more explicit ways. Many contracts
explicitly disclaim liability for damages resulting from problems that cause data loss.
191
Furthermore, vendors have in the past threatened to sue election officials and others who
publicize machine flaws or independently investigate and test machines malfunctioning
machines.
192
Election officials say this encourages them to keep quiet about machine malfunctions.
ough understandable, this reluctance to publicize malfunctions contributes to the
possibility that election officials and watchdog groups remain in the dark about known
problems. Consequently, it is essential to provide election officials with protections against
vendor retaliation. One solution is to allow election officials to post information about
known problems on a nationwide database semi-confidentially,” meaning that only other
election officials and/or the agency charged with maintaining the database could view
the official’s contact information. Similarly, through statute, Congress or the states could
provide monetary penalties and perhaps the creation of a private right of action against
vendors that retaliate and/or harass individuals or localities who report problems.
193
Last
years announcement that the largest voting system vendors, Election Systems and Software
(ES&S) and Premier (formerly known as Diebold), planned to merge raised concerns. A
central worry was that the merger would leave election officials in an even weaker position
26 | Brennan Center for Justice
relative to voting system vendors. At the time, the New York Times estimated that the
merger would mean that nearly 70 percent of the nations precincts would use machines
made by a single company.
194
e newspaper noted that this “would make it harder for
jurisdictions to bargain effectively on price and qualityfor new purchases.
195
More to the
point, it would make jurisdictions more dependent on a single vendor for everything
from repairs to future service – and thus less likely to speak publicly about voting system
deficiencies.
After the U.S. Department of Justice and nine state attorneys general filed an antitrust
suit over the merger in March of 2010, the Department of Justice announced that it
had secured an agreement from ES&S to divest itself of many of Premier’s voting system
assets (though not necessarily its service contracts).
196
In May, Dominion Voting Systems
another manufacturer with a significant share of the U.S. market – announced that it had
acquired the assets of Premier from ES&S in accordance with the Department of Justices
proposed settlement.
197
e merger highlights two major concerns for election officials in a market that just one
vendor could eventually dominate. First, election officials already weak leverage with
vendors would further diminish, for if there is essentially just one vendor, there is no
viable alternative vendor available, regardless of how poor the service or function of the
machines.
Second, and perhaps more importantly as relates to this report, having the vast majority
of the nations voting systems manufactured and/or serviced by a single company could
also mean much greater vulnerability nationwide to software bugs or other problems,
particularly if such problems are not immediately publicly reported and corrected
throughout the country.
198
Brennan Center for Justice | 27
As this report shows, the current regulatory scheme for voting systems does not adequately ensure
that problems with these systems are detected and corrected.
199
e Brennan Center proposes a new
regulatory structure to address this inadequacy, one based upon our interviews with election officials
and regulatory experts, our review of analogous regulatory structures in other important industries. e
new regulations and/or statute must include at least four key provisions:
1. A Publicly Available, Searchable Centralized Database: Election officials, in particular, would
benefit from a publicly available, searchable online database that includes official (i.e., election
official-reported or vendor-reported) and unofficial (i.e., voter-reported) data regarding voting
system failures and vulnerabilities, and other reported problems and establishes criteria for the
databases contents and organization.
2. Vendor Reporting Requirements: Vendors must be required to report to the appropriate
government agency via the database and certified mail early warning” data regarding known
and suspected voting system failures and vulnerabilities, and other reported problems,
including when vendors receive a complaint from a customer (an election official), when they
receive a warranty claim and/or take some action to satisfy a warranty, when they conduct an
investigation of a reported problem, and when a customer or other person sues them.
3. A Federal Agency with Investigatory Powers: e best way to ensure that vendors address
potential problems in a timely manner is to empower the appropriate government agency to
investigate all voting system failures and vulnerabilities listed on the database, grant the agency
subpoena power to facilitate its investigations, and require vendors to, among other things,
maintain records that may help the agency determine whether there are indeed voting system
failures or vulnerabilities, and whether the vendor has taken appropriate action to address the
failures or vulnerabilities.
4. Enforcement Mechanisms: e appropriate government agency must have the power to levy
civil penalties on vendors who fail to meet the reporting requirement or to remedy failures or
vulnerabilities with their voting systems.
200
is section discusses in detail how we believe each of these critical provisions should be drafted,
analogous legislation and regulations that contain similar provisions, and the key benefits that the new
provisions would bring to the regulation of voting systems.
Of course, as with any regulation or law, good definitions will be critical to creating an effective
regulatory scheme. We provide suggested definitions for many key terms used in these sections (“voting
systems,” “vendors” “failures,” etc.) in Appendix A.
1. A Publicly Available, Searchable Centralized Database
A robust regulatory system should mandate the creation of a searchable online database. It should
be easily accessible through the appropriate government agency’s home page. And, it should contain
comprehensive information about all reported voting system failures, usability concerns, vulnerabilities,
iv. a better way to track and address voting system problems
28 | Brennan Center for Justice
or potential vulnerabilities by any person, including, among others, machine vendors, election officials,
and voters. Vendor reporting of such problems should be mandatory. Reporting from others should
be permissive. Additionally, while voters, election officials, and others should be able to simply upload
their reports to the database via the Internet (subject to review by the appropriate agency), vendors would
be required to both upload the reports to the database via the Internet and send the reports to the
appropriate government agency via certified mail.
As already discussed (supra p. 25), given their ongoing reliance on voting system vendors to repair and
service their systems, election officials should have the option of filing reports confidentially, meaning
they can request no individuals except other election officials know their identity. is would encourage
more honest and timely reporting by those most likely to observe voting system malfunctions.
Similarly, it makes sense to provide whistleblowers
working for voting machine companies and/or state
and local governments with the option of requesting
that personal information be kept confidential. To
ensure confidentiality, Congress would probably
have to provide a FOIA exemption for reports filed
by election officials.
201
A searchable database would have benefits beyond
the issuance of “advisories to customers upon a
problems discovery. In part because of high turnover
among election officials, such advisories can get lost
from election to election (see, e.g., the Humboldt
County, California case study discussed supra pp.
12 - 13). Moreover, officials looking to purchase or
deploy new systems will not necessarily have easy
access to advisories issued by a vendor or the EAC (in the case of EAC certified systems) in the recent
or distant past. By making it simple for election officials and the public to search for problems and
workarounds associated with voting systems at any time, a well designed database could increase the
likelihood that jurisdictions looking to use new machines would learn about potential problems before
purchase or use.
A. Provision Details
A bill or new regulations addressing this issue should set specific requirements for reports, including:
(1) a description of the make and model of the voting machine involved;
202
(2) the jurisdiction(s) in
which the machine is being used, if applicable; (3) a description of the nature of the problem or concern
with the machine; (4) the date of the discovery of the problem or concern; (5) the name and contact
information for the person submitting the report; (6) a verification by the person submitting the report
that the information submitted is true and accurate and that the person consents to such information
being included in the database; (7) versions of hardware, software, and firmware affected; and (8) any
suggested workarounds and fixes, or instructions for how to retrieve this information when it becomes
available.
a searchable, central database could
“serve as a repository to let me know
what issues exist with the machine,
rather than having to rely on what
the vendors spoon-feed me.”
rokey suleman, executive director of
the district of columbias board of
elections and ethics
Brennan Center for Justice | 29
Regulations should stipulate that the database be searchable by: (1) the date of discovery of the problem
with the voting machine; (2) the make and model of the voting machine involved; (3) the nature of the
problem with the machine; (4) the jurisdiction in which a system is used; and (5) such other categories
as the appropriate government agency deems necessary. Additionally, we propose that the legislation
or regulatory framework include a provision that prohibits the appropriate government agency from
disclosing to anyone other than an election official the contact information of an election official, voter,
or vendor employee who submits a report to the database without the express written consent of the
person submitting this information.
203
e database should also be searchable by and distinguish among reports submitted by (1) election
officials, acting in their official capacities on behalf of their governmental units; (2) vendors; and (3)
all other submitters, including voters, whistleblowers and anyone else. Voters, election officials, and
vendors are likely to discover very different issues with voting systems, and their reports are likely to
carry different weights with different audiences. is division should particularly help election officials,
voting rights groups, and any agency investigating potential problems.
Finally, the agency responsible for creating and maintaining the database should probably be given
some ability and responsibility to review reports before they are posted, to ensure that on their face they
belong on the database. Allowing anyone to report anything without a filter could allow individuals to
overload the database with useless or irrelevant reports, resulting in an essentially useless database.
204
B. Responsible Agency
e EAC could construct and maintain the database. e EAC is already in the business of establishing
voluntaryguidelines to which new voting systems are tested. Section 202 of HAVA gives the EAC
the responsibility of serving as a national clearinghouse and resource for the compilation and review
of informationrelated to the administration of elections. And, as already discussed (supra pp. 8 - 9)
the agency collects anomaly reports provided by vendors and election officials for the few systems it has
certified. e EAC does not currently list the data it receives from vendors and election officials in the
kind of searchable database that other agencies use, and that we believe would be most useful to election
officials, but there is no reason it could not do so in the future (in fact, this may be largely addressed
with the adoption of a new clearinghouse policy in the near future).
More problematically, the EAC has taken the position that its powers to facilitate the understanding
and resolution of problems with non-EAC certified voting equipment is extremely limited, absent
explicit Congressional authority that does not currently exist.
205
As the vast majority of machines
currently in use in the United States have not been certified by the EAC,
206
this could represent a
serious impediment to the creation of a database under the auspices of the EAC, absent a clarification
from the EAC or Congress. As the Government Accountability Office has noted, if Congress explicitly
expands the EAC’s powers in this regard, it should also consider providing the EAC with the additional
resources necessary to take on this additional work.
207
e EAC’s Voting System Testing and Certification
Program currently employs only five staff members.
208
Alternatively, the Comptroller General and the GAO could assume responsibility for the database.
e GAO has a strong reputation of competence and impartiality, and it has a well-established track
record for acquiring and publicizing information.
209
It has also already done a considerable amount of
work related to HAVA and voting machines,
210
and has in fact noted the void that currently exists in
identifying and resolving problems with non-EAC-certified voting systems, in particular.
211
30 | Brennan Center for Justice
Finally, the Department of Justice could create and maintain the database. e Department has similar
responsibilities in other contexts. For instance, the Anti-Car eft Improvements Act of 1996
212
gave
the DOJ responsibility for creating the National Motor Vehicle Title Information System, a database
designed to compile information from a variety of sources on the histories of individual motor vehicles.
e Act also provided the Department with enforcement powers in the event someone required to
submit information to the database failed to do so.
C. Analogous Regimes
Analogous regulatory regimes for other industries demonstrate the power and usefulness of this kind
of regulatory scheme.
e clearinghouse we propose is similar in principle to the database that Congress ordered the Consumer
Product Safety Commission (CPSC) to establish as part of the Consumer Product Safety Improvement
Act (CPSIA) of 2008.
213
One of CPSIAs key features was Congresss mandate that the CPSC create
a publicly available, searchable database that records information from, among others, consumers,
government agencies, and healthcare professionals, regarding the harms related to the use of a consumer
product regulated by the CPSC.
214
A House Committee Report on the legislation states that the “goal
of the CPSC should be to devise a database that can rapidly provide consumers with early warning’
information about specific products that could pose serious safety hazards.
215
e National Highway Traffic Safety Administration (NHTSA)
216
and the Food and Drug
Administration (FDA) maintain similar early warning” databases. NHTSAs Early Warning Reporting
database collects and makes publicly available property damage reports and death and injury reports
provided by manufacturers pursuant to the Transportation Recall Enhancement, Accountability, and
Documentation (TREAD) Act of 2000.
217
Meanwhile, NHTSAs Office of Defects Investigations (ODI)
database, even before the passage of the TREAD Act, has allowed consumers both to make and search
for safety complaints regarding problems with motor vehicles or pieces of motor vehicle equipment.
218
e FDAs Adverse Event Reporting System, moreover, collects and makes publicly available adverse
drug reaction information by healthcare professionals and consumers.
219
ese databases, although not quite as robust and user-friendly as the proposed CPSIA database, have
played important roles in protecting the public. Indeed, it was independent analysis of NHSTAs
complaint database that catalyzed the Bridgestone/Firestone tire recall years ago.
220
And, it was the
Bridgestone/Firestone recall that was the impetus behind the passage of TREAD.
221
In June 2009, researchers mining the NHTSA complaint database discovered the high rate of failure of
Chinese valve stems; this ultimately led to the recall of millions of valve stems in vehicle tires.
222
Later
in the year, NHTSA launched an investigation of Toyota vehicles after receiving over 400 consumer
complaints about acceleration problems with the cars; some of these problems appear to have been
responsible for fatal accidents.
223
is investigation led Toyota to announce on November 25 that it
would repair the accelerator pedals on some 4.26 million vehicles.
224
After the first recall, reports of
problems with Toyota vehicles, which had been steady before acceleration problems began to receive
attention in the fall, surged.
225
As more customers came forward with complaints, federal investigators
expanded the probe to look at other problems with Toyota vehicles, and the company issued additional
recalls for brake and acceleration problems.
226
On April 5, 2010, NHTSA assessed the largest legally
permissible fine against Toyota. In his letter to the company notifying them of the fine, NHTSA Chief
Counsel O. Kevin Vincent identified the manufacturers failure to notify authorities in a timely manner
after becoming aware of the defects as the primary rationale for levying the harsh penalty.
227
Brennan Center for Justice | 31
Results from the FDAs Adverse Event Reporting System have been no less powerful. For example, in 1998
shortly after the Systems launch, reports of liver injuries on the system resulted in the FDAs removal of the
anti-inflammatory drug Duract from the market.
228
More recently, a handful of reports on the system about
a serious condition affecting bone marrow of those taking the antibiotic Zyvox, led to a change to the drug’s
labeling to inform providers about the risk of this reaction and recommendations for monitoring patients.
229
We chose the CPSIA database as a model because it will be the most comprehensive publicly available
early warningdatabase ever created, and the CPSC has been developing it based upon lessons from
the NHTSA and FDA databases with a desire to improve upon them by, among other things, making
them more user-friendly.
230
Indeed, in CPSIAs legislative history, a House Committee urges the CPSC
to “examine these and other Agency efforts, if applicable, when designing its own database.
231
D. Key Benefits
A database that conforms to the specifications we have outlined would allow the appropriate government
agency, vendors, and the public to efficiently access, evaluate, investigate, and share information
regarding voting system failures or vulnerabilities. is, in turn, would result in earlier detection of
system failures and vulnerabilities by vendors and election officials, and ensure more rapid responses to
those issues ultimately making it less likely that voting systems will malfunction on Election Day, when
it matters most. More specifically, however, the database would achieve the following objectives:
Increase Election Official & Public Awareness of
Problems & Solutions
e database would allow election officials to
easily access and share information. For example,
this would help to ensure that election officials in
different states using the same make and model of
machine or other voting system element would be
aware of any problems the other had encountered using that equipment if one official discovers a
vulnerability and posts this information to the database, the other official will be able to check his or
her system to ensure that it does not also have that vulnerability. Moreover, if a vendor provides one
election official with a procedural solution, such as a workaround to address a problem, the database
would provide officials using the same voting equipment in other jurisdictions with that solution.
Further, the database would provide voter protection groups, political parties and other concerned
members of the public with more knowledge about potential voting system problems. Before an
election, they would, among other things, be able to determine whether there have been any issues
with the voting systems in their counties and, if so, to advocate corrective action before the election and
be on the lookout for similar problems during voting.
Improve the Ability of Government and Others to Identify Failures or Vulnerabilities & Respond Quickly
e database would allow the appropriate government agency, voter protection groups, and other
concerned members of the public to mine the information on the database and spot patterns and trends
that may indicate voting system failures or vulnerabilities. is is a key reason for allowing individual
voters to report problems they encounter. All of the databases hosted by federal agencies that we
reviewed allow such reporting from the general public. After collecting and analyzing this information,
the appropriate government agency would be able to more quickly launch an investigation to rectify
these potential vulnerabilities or failures.
a national, searchable database
would allow election officials to
easily access and share information.
32 | Brennan Center for Justice
Provide Timely & Organized Access to Information
e database would include all of the reports uploaded to it in an easily searchable format essentially in
real time. It would help to eliminate the thousands of state and county government silos by centralizing
all the data in one place. Further, placing the database on the Internet via a single home page would
significantly increase the accessibility of the information to all.
Assist Election Officials in Evaluating the Comparative Performance of Voting Systems
By allowing for searches by the make and model of a voting machine or other voting system component
among other datasets, the database would provide election officials with a helpful resource to determine
which equipment has been the most reliable. is information would be particularly useful for election
officials considering the purchase of new voting systems.
Provide Election Officials and Others with the Opportunity to Identify Machine Issues without
Fear of Retribution
A provision that prohibits the government from disclosing the contact information of any election official
or other person who submits a report to the database to persons other than election officials without the
written consent of that person protects the confidentiality of these individuals; this protection reduces
the risk that a voting system vendor attempts to harass or seek retribution against them for posting a
negative report to the database. For similar reasons, there should be monetary penalties and perhaps the
creation of a private right of action against vendors that retaliate and/or harass individuals or localities,
including especially whistleblowers, who report problems.
2. Vendor Reporting Requirements
A. Provision Details
New legislation or regulations should require voting system vendors to provide “early warning” data to
the appropriate government agency regarding voting system failures or vulnerabilities. is would go
well beyond what manufacturers of the few EAC certified systems must currently report to the EAC (see
supra pp. 8 - 9) if they wish to maintain that EAC certification. We propose that the government require
all vendors to provide written notification via certified mail (in addition to uploading the information to
the database) when they determine that a voting system failure or vulnerability may exist, including when
• theyreceiveacomplaintfromacustomer(anelectionocial);
• theyreceiveawarrantyclaimand/ortakesomeactiontosatisfyawarranty;
• theyconductaninvestigationofareportedproblem;and
• acustomerorotherindividualsuesthem.
e legislation or regulations should set specific requirements for the notification, including: (1) the
location of the failure or vulnerability; (2) a description of the failure or vulnerability; (3) the vendor
of the voting machine; (4) the jurisdictions where the machine is used; (5) an evaluation of the risk to
election outcomes related to the failure or vulnerability; (6) the vendor’s intended remedy for it; (7)
versions of hardware, software, and firmware affected; and (8) any suggested workarounds and fixes, or
instructions for how to retrieve this information when it becomes available. It is critical that the EAC
Brennan Center for Justice | 33
or other relevant federal agency has the discretion to require vendors to include other information in
the notice, and to require further reporting related to any corrective action plan, to ensure that (when
necessary) remedial steps are taken and are adequate.
B. Responsible Agency
Just as the EAC, GAO or DOJ would be appropriate agencies to appoint to create and maintain
the database, all three would be appropriate agencies to vest the power to require vendors to provide
additional information in the required notice. Whichever agency is assigned responsibility for
maintaining the database and setting notice requirements for vendors should also probably be the
agency which receives such notices.
C. Analogous Regimes
e same federal agencies that maintain the databases discussed above: NHTSA, the CPSC, and
the FDA, also require manufacturers to report early warningand other data directly to them. Our
proposed reporting requirements for voting machine vendors are similar to these regimes, particularly
in respect to the contents of the required notices.
ere are two types of requirements for reporting problems to NHTSA: reports regarding early
warning” data and reports regarding defects. Regarding early warning data, rules promulgated by
the Secretary of Transportation
232
under the authority of the TREAD Act
233
require manufacturers to
submit information on each make and model of vehicle offered for sale in the United States within the
previous two years that details (1) incidents involving death or injury that were alleged or proven to be
caused by a possible defect, including foreign incidents occurring in substantially similar or identical
vehicles; (2) property damage claims, warranty claims, and consumer complaints; and (3) field reports
identifying defects, fires, or rollovers.
234
Requirements for other types of vehicles and equipment such
as child restraints and tires are substantially similar.
235
After submitting a one-time report of historical
information covered by the regulations,
236
manufacturers must submit the information described above
on a quarterly basis.
237
If a manufacturer identifies a defect
238
and determines “in good faith” that the defect has an impact on
motor vehicle safety or that the vehicle or equipment does not comply with applicable safety standards,
the manufacturer must notify the Secretary of Transportation and all owners, purchasers, and dealers
of the vehicle or equipment in question.
239
is notification must contain: (1) a clear description of the
defect or noncompliance; (2) an evaluation of the risk associated with the defect; (3) the measures to be
taken to obtain a remedy; (4) a statement that the manufacturer will provide the remedy without charge;
(5) the period during which the defect will be remedied without charge; and (6) the procedure for
notifying the Secretary of Transportation of the manufacturers failure to remedy a defect as mandated
by law.
240
Depending on the magnitude of the risk presented by the defect and the cost of providing
public notice relative to number of additional owners the notice is likely to reach, the Secretary of
Transportation may also require manufacturers to provide public notice.
241
Reporting requirements to the CPSC are similar to those for NHTSA. Manufacturers who discover that
a product does not comply with product safety rules or standards or contains a defect that creates the
risk of injury or death are required to inform the CPSC of the problem.
242
If the CPSC determines that
the noncompliance or defect constitutes a substantial product hazard requiring consumer notification,
34 | Brennan Center for Justice
the CPSC may compel the manufacturer to stop distribution; notify those involved in the transport,
distribution, or sale of the product by mail or other means; provide public notice on the Internet, TV,
and radio; and/or mail notice to all known purchasers of the product.
243
e CPSC has the authority to determine the form and substance of any such notice,
244
but the law
provides some guidelines for manufacturers. Unless the CPSC rules otherwise, all notices must contain,
among other things, (1) identifying information such as a model number and photograph of the
product; (2) a description of the action taken to remedy the defect or noncompliance; (3) a description
of the hazard caused by the product; (4) a number and description of injuries and deaths caused by
the product; (5) a description of available remedies and how to avail oneself of them; and (6) retail
information concerning the product.
245
Finally, the FDA requires all manufacturers of all drugs
marketed under an approved FDA application to report to
the FDA all serious, unexpected adverse drug experiences
associated with the use of their drug products.
246
e FDA
mandates that the manufacturer submit to it, within fifteen
days of learning of the adverse drug experience, a form
247
that contains: (1) a description (i.e., sex, age, weight,
height) of the patient that took the drug; (2) the outcomes
attributed to the adverse event; (3) the date of the event;
(4) the date of the report; (5) a description of the event; and (6) various information regarding the
product suspected to be the cause of the event.
248
D. Key Benefits
Many of the benefits of mandatory vendor reporting of machine failures and vulnerabilities are the same
as those of the proposed database (e.g., assisting election officials in identifying and resolving problems,
aggregating information in a timely and organized manner, and allowing election officials to compare
the performance of voting systems); below we detail some additional benefits that are particular to the
proposed vendor reporting requirements.
Incentivizes Vendors to Enhance Internal Controls
Vendors will presumably want to minimize the number of reports that they must make to the appropriate
government agency. One way that they will be able to do this is by enhancing their own testing and
internal standards to avoid any late-stage defects that would trigger a requirement to make a report to
the appropriate government agency.
Ensures Maximum Disclosure of Information by Vendors
As we detail above, Congress passed the TREAD Act and established these early warning” reporting
requirements, in part, because investigations in the wake of the Firestone tire recall revealed that “both
Firestone and Ford knew that there were problems [with the tires] years before they told [NHTSA] or the
American public.
249
e case studies in Part III of this report show that, at the very least, many election
officials and other concerned citizens worry that voting system vendors have sometimes taken too long
to acknowledge and publicize problems with their systems. is provision would require vendors to take
mandatory vendor reporting of
failures to a searchable database
would incentivize those vendors
to enhance internal controls.
Brennan Center for Justice | 35
affirmative steps to increase their transparency and would ensure, among other things, that at the very least,
election officials and the appropriate government agency will have access to problems soon after vendors
discover them.
3. A Federal Agency with Investigatory Powers
A. Provision Details
If this new regulatory structure is going to be effective, a federal agency must have adequate enforcement
authority. e most logical model would allow the appropriate federal agency to initiate an investigation
after reviewing any of the information posted to the database and determining that a machine failure
or vulnerability potentially exists.
In order to facilitate these investigations, new legislation should provide the appropriate federal agency
with the power to issue subpoenas and include a provision that would require vendors of electronic
voting machines to maintain records, reports, and other information to enable the agency to determine
whether there is compliance with other provisions of the legislation.
B. Responsible Agency
e EAC already has some investigatory powers related to its federal certification program. Specifically,
manufacturers who register to have new voting systems federally certified by the EAC must, pursuant to
the EAC’s VSTCPM (discussed previously at page 8),
250
agree to “[c]ooperate with any EAC inquiries and
investigations into a certified systems compliance with VVSG standards and the procedural requirements
of this Manual . . . .
251
While the VSTCPM does not currently require vendors to report all of the types
of problems we have detailed (see supra pp. 8 - 9) to a centralized database (none currently exists, of
course), we could imagine an amendment to the VSTCPM which would require such reporting as part
of the Voting System Testing and Certification Program, and require manufacturers to cooperate with any
investigations into their compliance with such mandates. Of course, as previously noted in this report, the
EAC is currently limited to investigating manufacturers registered under the Voting System Testing and
Certification Program, and who stay registered under that program.
252
For this reason, Congress might
need to explicitly empower the EAC to employ these investigatory powers as applied to problems arising
with non-EAC certified systems. It would also probably need to provide the EAC with extra funding, as
the EAC has previously stated that even if given this power, it does not have the resources to track and
resolve problems related to non-EAC certified systems.
253
Alternatively, the GAO regularly conducts investigations in support of its mission.
254
As discussed in greater
detail below, the GAO is almost certainly constitutionally barred from taking enforcement action against
vendors or others, but as a legislative agency, it should have the power to investigate and gather information.
255
Consequently, if the GAO is given responsibility for creating and maintaining the database, it might well
make sense to also give it explicit investigatory powers necessary to ensure that the database is accurate.
Finally, it may make sense to vest investigatory powers with the Department of Justice, an agency
with a substantial infrastructure to conduct investigations and bring enforcement actions, where
necessary.
256
e Voting Rights Section of the Civil Rights Division at the Department of Justice
has a long history of taking enforcement actions under a variety of federal laws relating to voting,
including voting machines, and also has experience administering complex statutory schemes.
257
e
36 | Brennan Center for Justice
Civil Division of the Department of Justice has experience enforcing an even wider variety of federal
laws and regulations; the Federal Programs Branch of the Civil Division assists federal agencies like the
Department of Energy, Housing and Urban Development, the Department of Health and Human
Services, and others to carry out their regulatory obligations by initiating litigation against those who
violate statutes or regulations.
258
In fact, giving the Department of Justice investigatory and enforcement
powers would be consistent with the structure that already exists in HAVA. While HAVA assigns the
EAC a clearinghouse and advisory role, it also gives the DOJ enforcement authority to bring an action
for declaratory and injunctive relief for failure to comply with HAVAs minimum requirements for
voting systems, provisional voting and voter registration.
259
C. Analogous Regimes
Both NHTSA and the CPSC have broad investigatory powers.
260
e CPSC has the authority to enter
and inspect any factory, establishment, or conveyance used to facilitate placing goods into the stream of
commerce.
261
NHTSA has similar authority.
262
e simple ability to investigate information uploaded
to their databases, of course, falls within the ambit of these agencies’ powers. For example, it is the
responsibility of the Office of Defects Investigations within NHTSA to elicit from every available
source and evaluate on a continuing basis any information suggesting the existence of a safety-related
defect.
263
NHTSAs databases allow the ODI to both elicit and evaluate this information, and as we
have discussed above,
264
investigations initiated after reviewing information in the databases have been
fruitful.
Both NHTSA and the CPSC have subpoena power
265
and a recordkeeping requirement for the entities
that they regulate.
266
We propose that the new provision closely track this language, particularly that of
the Consumer Product Safety Act of 1972.
267
D. Key Benefits
Ensures Reliability of Information on the Database
e information posted to the proposed database must be accurate so that false information does not
mislead election officials, the public, the government, EAC, or unfairly sully vendors’ reputations. e
government will need the tools necessary to ensure that only accurate information is posted to and
remains on the site.
Grants the Government Power to Effectively Respond to Information it Receives
A recordkeeping requirement would allow the appropriate government agency to address reported
failures or vulnerabilities of the machines quickly, without having to wait for vendors to gather this
information. Moreover, granting subpoena power to the appropriate government agency ensures, among
other things, that the agency will be able to gather the information it needs to adequately judge the
seriousness of reported problems and ensure that proper steps have been taken to prevent malfunctions
or other failures from happening in the future.
Brennan Center for Justice | 37
4. Enforcement Mechanisms
A. Provision Details
New legislation or regulations should give the appropriate agency adequate enforcement powers by
authorizing it to seek civil penalties against voting system vendors for failure to comply with their
duty to report any voting system failure or vulnerability or to remedy the issue after learning about
it. It should also cap the penalty amount for a series of related violations at a specific dollar amount;
similarly, there should be a specific dollar cap on the penalty amount for problems for any individual
model and version of a product.
B. Responsible Agency
As already discussed, the Department of Justice and the Voting Rights Section of its Civil Right Division
have a long history of taking enforcement actions under a variety of federal laws related to voting. In fact,
according the Departments website, it has filed a dozen actions to enforce various HAVA requirements,
including requirements related to voting machines.
268
Accordingly, the most logical place to house
enforcement powers related to the database is with the Department.
HAVA already divides authority between the EAC and DOJ, giving the EAC responsibility for providing
information and advising local jurisdictions, and giving the Department of Justice enforcement
powers.
e EAC’s ability to take enforcement actions against
manufacturers is likely to be more limited. As already
discussed, the EAC currently has the power to decertify
systems
269
it previously certified (which is only a small
percentage of systems currently used in the United
States)
270
or suspend the registration of a manufacturer
271
seeking federal certification for new systems for various
infractions. is does not include failure to report to a
central database, but we can imagine an amendment
to the VSTCMP that allows the EAC to take such
actions for failure to comply with database reporting requirements. However, using decertification
and/or suspension of registration would be rather blunt instruments for what could, in many cases be
relatively minor infractions. As the VSTCPM itself notes, “[d]ecertification is a serious matter. Its use
will significantly affect Manufacturers, State and local governments, the public, and the administration
of elections.
272
It is doubtful that the EAC would want to take such a drastic step for all but the most
serious infractions. At the same time, the EAC does not have the power to decertify systems it has not
certified, which represents the vast majority of the systems in use today.
273
While it may be technically possible for the EAC to take additional enforcement action – such as seeking
imposition of monetary penalties – if Congress amended HAVA to vest it with that power – it is not clear
that the agency currently has the infrastructure or institutional knowledge to carry out such tasks. the
GAO is almost certainly constitutionally barred from taking any enforcement action, as it is considered a
creature of Congress, which prevents it from taking actions that amount to executing the law.
274
ensuring that the appropriate
federal agency can enforce
new vendor reporting requirements
would help restore public
confidence in voting systems.
38 | Brennan Center for Justice
C. Analogous Regimes
Civil penalty provisions are not uncommon in analogous regulatory regimes. Recently the trend has
been for Congress to increase such penalties. For example, in 2002, the TREAD Act increased the
penalties originally established in the Vehicle Safety Act of 1966
275
that NHTSA could seek to have
imposed on vehicle manufactures from $1,000 per violation to $5,000 per violation and from a penalty
cap of $800,000 to a cap of $15,000,000.
276
Similarly, in 2008, CPSIA dramatically increased the civil
penalties already provided for in the Consumer Product Safety Act of 1972 (e.g., from $5,000 per
violation to $100,000 per violation) for knowing sale of products that do not conform to an applicable
rule or standard or failure to make necessary records available to the CPSC.
277
e penalty language
legislation should resemble the language of both the Vehicle Safety Act and the Consumer Product
Safety Act.
D. Key Benefits
Increases Vendor Accountability
As Part III (Failures of the Current System: Case Studies) of this report illustrates, in the past, vendors
have not been held accountable for voting machine failures or their inability or refusal to correct them.
Legislation with enforcement mechanisms would provide the EAC or another federal agency with the
necessary enforcement tools to hold vendors accountable for their acts or omissions after an investigation
and hearing shows that this is justified.
Incentivizes Vendors to Quickly Comply with Mandates
Civil penalty provisions would act as a deterrent to wrongdoing by vendors; penalties would be high
enough so as to not simply be regarded by vendors as a cost of doing business.
Helps to Restore Public Confidence in Voting Systems
HAVA was meant to help restore public confidence in the integrity of the electoral process. During the
signing ceremony for HAVA, President Bush stated that “[t]he legislation I sign today will add to the
nations confidence.
278
He further added, “[t]hrough these reforms, the federal government will help
state and local officials to conduct elections that have the confidence of all Americans.
279
Giving an independent government agency the power to investigate problems and take action to remedy
problems should greatly increase public trust in our voting systems and elections.
Brennan Center for Justice | 39
what local governments can do now
While a national, centralized and searchable database along the lines we have suggested in
this report is ultimately the best way to ensure that problems with machines are publicized
and corrected throughout the country, there are important interim steps that county and
state governments, in particular, can begin taking immediately to increase the chances that
election officials learn of problems with their voting systems and can avoid some of the kinds
of problems detailed in this report
Negotiate Better Contracts
As previously detailed in this report, provisions in many voting machine contracts make it
much more difficult for election officials and the public to get detailed information about
system problems reported in other parts of the country, or to hold vendors responsible for
problems when something goes wrong. To increase voting system reliability and maximize
vendorsmotivation to minimize the risk of such problems, counties and states should begin
demanding certain key contract terms, including:
• mandatingreportsfromvendorsonaperoccurrencebasisofanyhardwareor
software system error occurrences resulting from design or manufacturing defects
in any jurisdiction” in which the voting system is being used;
280
• mandatingreportsfromvendorsofanycomplaints(includingusabilityconcerns),
warranty claims and lawsuits about their systems, together with the result of
any vendor investigation, explanations, and actions taken to address these
complaints;
• extendingwarrantyperiodsforthepurchasers;
• mandatingnancialliabilitytovendorsintheeventofamalfunction;
• ensuringthatvotingsystemsaretestedagainstthemostmodernfederalguidelines
instead of older versions that computer scientists have faulted as inadequate; and
• allowinglocalelectionocialstoindependentlytesttheaccuracyofthemachinesas
long as they do not disclose proprietary information or trade secrets.
281
e New Jersey Department of the Public Advocate discusses how jurisdictions might
construct many of these provisions in a document entitled “e Purchase of Voting
Systems in New Jersey: How Government Can Better Protect Taxpayer Rights and Voting
Security.
282
Washington, D.C. has also passed a law setting more stringent requirements
for voting system contracts.
283
We strongly urge election officials entering into contracts
with voting system vendors to review the contents of these documents.
40 | Brennan Center for Justice
Of course, adding these kinds of provisions to contracts will probably only be of benefit
to jurisdictions as they negotiate new contracts. Jurisdictions not purchasing new voting
systems in the near future (i.e., the vast majority of counties and states throughout the
country) are unlikely to secure agreement from vendors to substantially amend their
contracts along these lines.
Furthermore, even if all jurisdictions were able to get these kinds of provisions into their
contracts, it would still be important to mandate many of these provisions in regulation.
For instance, if a vendor violates the notification terms of its contract with a small county
in Arkansas by failing to inform it of a voting system malfunction in New Jersey, the
Arkansas County is unlikely to find out about the violation under current circumstances.
More to the point, even if it does learn of such a failure, the county is unlikely to bring
an action under the contract (for all of the reasons discussed on pages 25 - 26 of this
report) unless the malfunction brings disastrous results in its own election sometime later.
e point of the database and stronger regulation is to minimize the likelihood of such
occurrences by bringing problems to light before they can cause such significant harm.
Implement Stronger State Regulation
e legislature in at least one state, California, has passed legislation requiring vendors
selling systems within its borders to notify the Secretary of State and all local election
officials using its systems of any defect, fault or failure” it discovered, within 30 days
of discovery.
284
As of the writing of this report, the legislation awaits a decision by the
governor, who vetoed an earlier version of the bill in 2009.
285
is bill would empower the
Secretary of State to seek monetary penalties against vendors for failing to comply with
reporting requirements, and requires the Secretary to notify the EAC of the problem.
While the California proposal would not have solved all of the problems identified in this
report, it would have gone a long way toward forcing vendors to begin to publicly report
problems with their systems when alerted to them. It could also have provided many of the
benefits (albeit on a smaller scale) that a national centralized database would provide: in
particular, increasing vendor accountability and incentivizing vendors to enhance internal
controls; benefiting state certification programs by supplying tips for targeted testing and
review of the effectiveness of mitigations proposed by vendors; allowing election officials
(in California, anyway) to get up-to-date information about their systems before deploying
them in elections.
e California model presents the best legislative attempt we have seen, to date, to address
the problems we have discussed in this report. We hope the governor will sign this bill into
law and that other states will follow suit.
Still, the limitations of this proposal as compared to a national clearinghouse are obvious.
Among other things, the California bill only requires reporting for systems used in
Brennan Center for Justice | 41
California; the bill does not mandate a searchable database that new election officials could
review before each election; and because the bill could not provide a method for election
officials and whistleblowers to anonymously report problems that might not technically
fall within the definition of defect, fault or failure (such as the problems in Pulaski
County, Arkansas, described on page 14 of this report), it would not necessarily result
in the sharing of information about some kinds of problems that still result in the loss of
hundreds and thousands of votes.
Create a Voluntary Database
While the EAC has previously indicated that it does not believe it has the authority or resources to
facilitate the understanding and resolution of problems with non-EAC certified voting equipment,
we can think of no legal reason why even absent additional authorization from Congress it
could not create a searchable database to which election officials, voters, and vendors could report.
In the alternative, if the EAC determines it does not have the resources to create such a database,
election officials, through organizations such as the National Association of Secretary of States, the
National Association of State Election Directors, , the National Association of County Recorders,
Election Ocials and Clerks, or the International Association of Clerks, Recorders, Election
Officials and Treasurers could sponsor such a database. ere would be no way to force vendors to
report to this database, or to provide election officials with whistleblower protections for making
voluntary reports – two important suggestions for the mandatory clearinghouse detailed in this
report – but it could still lead to a much better resource for election officials than currently exists.
Election ocials using the same equipment could also create user groups where they could share
information about their systems electronically.
Officials we interviewed wanted to see a database that vendors would be required to report
to, but many said they viewed a voluntary database as a good first step. For instance,
Cuyahoga County, Ohio Board of Elections Director Jane Platten stated that an exchange
of all information about anomalies, malfunctions and failures of voting systems is not only
much needed, but should be a requirement.However, she added that taking the small step
of creating a voluntary exchange of information would in my opinion have a huge impact on
election administration and operations.
Pressure Vendors to Voluntarily Post Information on eir Own Sites is Year
One drawback of the three previous recommendations in addition to the larger
recommendation of the creation of a central, mandatory, and federal database is that each will
take time to implement, and probably cannot be accomplished in time for this falls election. By
contrast, vendors could significantly reduce the risk of a repetition of past problems with their
systems by creating their own databases, and offering election officials who use their machines
access to them. As already noted in this report, because there is such high turnover among
election officials, a new county director might not be aware of product advisories or software
patches sent by a voting system vendor to her predecessor three years earlier. Nor is she likely
42 | Brennan Center for Justice
to be aware of problems reported by election officials in other counties. ose problems could
have been caused by poll worker error, a misuse or misprograming of the system, or a host of
other reasons that are not related to a software or hardware malfunction but that could still
easily be repeated in other counties in future elections.
Ideally, vendors would create for election officials a central, easily accessible and searchable
site where they could review all previously issued product advisories, software patches and
workarounds, and a list of all election official complaints, warranty claims and lawsuits about
their systems (together with the result of any vendor investigation, explanations, and actions
taken to address these complaints). is would provide election officials with an opportunity
to be more fully appraised of potential problems and safeguards that could be taken ahead of
each election.
County and state officials can and should demand this voluntary action from vendors now,
in time to make a difference for Novembers election and reduce the likelihood that we will
see a repetition of previous system failures.
Ultimately, of course, a mandatory reporting system with clear guidelines will be preferable.
As Butler County, Ohio Board of Elections Director Betty McGary put it, “[C]learly when
a vendor is continuing to sell their product in a State, they are not going to be excited
about voluntarily reporting deficiencies in their systems . . . requiring the reporting be
mandatory will be the only way” to get comprehensive cooperation.
Brennan Center for Justice | 43
Voting is the most important of all our federal rights. It is the right that protects all other rights. Despite
this, we have all too often been strangely cavalier about protecting it. e very different way we regulate
voting systems and other commercial products clearly dramatizes this fact.
Given the billions of dollars spent by federal and local government to purchase and maintain new voting
systems over the last decade,
286
the failure to take stronger measures to ensure that we are tracking and
correcting system failures is particularly troubling.
We propose a regulatory scheme that would greatly improve our election systems. It would:
• helpleveltheplayingeldbetweenelectionocialsandvendorsastheynegotiateoverservice
and hardware contracts;
• increasevendoraccountabilityandincentivizevendorstoenhanceinternalcontrols;
• providepublicadvocacyandvotingrightsgroupswithdataonpotentialproblemswithvoting
systems;
• provide the government and concerned citizens with the ability to locate patterns of
problems;
• benetstatecerticationprogramsbysupplying tipsfortargetedtestingandreviewof the
effectiveness of mitigations proposed by vendors;
287
and perhaps most importantly,
• allow election ocials to get the most up-to-date information about their systems before
deploying them in elections.
In light of the importance of safeguarding our democracy through accurate and fair elections, these are
especially worthwhile goals. We encourage policymakers to work with regulatory experts in other fields,
consult database experts, and talk to election officials and voting system vendors to ensure the creation
and quick use of the best and most effective database possible.
. 
44 | Brennan Center for Justice
  :  
As with any new statute or regulation, defining key terms will be critical to the effective implementation
of the proposals in this report. Among other things, a new statute or regulation must clearly define (1)
what kind of equipment is covered; (2) what types of problems must be reported; and (3) who must
report such problems.
What equipment should be covered by this new regulatory scheme?
Voting Systems: the new statute should cover “voting systems” as defined in Section 301(b) of the Help
America Vote Act,
288
and in addition should include electronic poll books.
Electronic Poll Books: electronic poll books are used with increasing frequency around the country.
289
As
with voting machines, electronic poll books rely on software and firmware that can be subject to bugs,
misprogramming and other glitches. And as with voting machines, electronic poll book malfunctions
have caused long lines, and the likely disenfranchisement of many voters.
290
For this reason, we believe
electronic poll books should be covered under a new regulatory scheme. Such poll books might be
defined as an electronic mechanism (including stand-alone software) by which an election official
at a polling place, at the time an individual seeks to vote, may obtain information on the individual’s
eligibility to vote (including whether the individual is registered to vote in an election for Federal office,
the polling place to which the individual is assigned, and whether the individual has already voted in the
election), whether the mechanism is operated by integration with a voting system or independently.
What types of problems must be reported?
To reduce the kinds of problems that cause lost votes, voting system vendors should be required to
report both voting system failures and vulnerabilities they have knowledge of. In the course of testing,
servicing and repairing machines, vendors may become aware of vulnerabilities that have yet to cause
a system failure, but that could be reasonably expected to cause a failure in the future. Vendors should
also be required to make reports to the database when they receive a complaint from a customer (i.e.,
election official), whether or not they agree that their machine was the cause of the alleged problem;
when they receive a warranty claim and/or take some action to satisfy a warranty; when they are notified
by a customer of a usability issue that could lead voters or poll workers to operated the system in a way
that would lead to disenfranchisement or the recording of an unintended vote; when they conduct an
investigation of a reported problem; and when a customer or other person sues them.
System Failures
291
: e term “system failure” should mean any event that results in
(a) loss of one or more voting system functions;
(b) degradation of performance such that the device is unable to perform its intended function
for longer than 10 seconds;
(c) automatic reset, restart or reboot of the voting device, operating system or application
software;
(d) a requirement for an unanticipated intervention by a person in the role of poll worker or
technician before the test or operation of the device can continue;
(e) error messages and/or audit log entries indicating that a failure has occurred; or
(f) failure to tabulate, tally, or report results accurately.
Brennan Center for Justice | 45
System Vulnerabilities: should include any flaw in a voting system which might reasonably lead to a
System Failure.
Who must report problems?
Anyone, including election officials, should be permitted to report voting system problems. For the
reasons detailed in this report, if the new regulatory scheme is going to have maximum effectiveness,
voting system vendors must be required to report both failures and vulnerabilities brought to their
attention.
Voting System Vendor should include any sole proprietor, partnership, LLC, corporation, commercial
entity or non-commercial entity that has contributed to the initial development, building, distribution
or other parts of the supply chain, or maintenance of all or part of the voting system.
Who is entitled to ask that their personal information be kept confidential?
At the very least, election officials and whistleblowers who work for vendors or state and local governments,
should be entitled to request that their name and identifying information be kept confidential.
A provision establishing this confidentiality option might look like this:
(a) If information is submitted for inclusion in [the database] by or on behalf of an election
official who affirmatively requests that his name and identifying information be kept
confidential, [the agency] shall not release to the public the submittersname and identifying
information, notwithstanding the provisions of Title 5 United States Code Section 552 or
other provision of law, unless the [head of agency] determines that such public disclosure
is necessary to advance the purposes of this chapter.
(b) In the case of information submitted for inclusion in [the database] by or on behalf of any
person other than an election official or a voting machine vendor, [the agency] shall not
release to the public the submittersname and identifying information, notwithstanding
the provisions of Title 5 United States Code Section 552 or other provision of law, unless
the submitter affirmatively authorizes such release or the [head of agency] determines that
such public disclosure is necessary to advance the purposes of this chapter.
(c) Except as provided in subsections (a) and (b) of this section, [the agency] shall make all
information submitted for inclusion in [the database] available to the public.
46 | Brennan Center for Justice
!
Machine Type
State
Location/Description
ES&S
Optech III-P
Eagle
Alabama
Lauderdale County, AL
The Times Daily reported that there was an error in
vote tabulations for the 2008 presidential primary
race at the Underwood-Petersville precincts. The
voting machines in these precincts showed that seven
more ballots were cast than were supported by
documentation.
292
ES&S
Optech III-P
Eagle
Alabama
Baldwin County, AL
According to the Associated Press, an optical scanner
used in the general election labeled an unopposed
Republican County Commissioner as a Democrat.
The error only occurred when voters attempted to
cast straight-ticket Republican ballots. Although the
election outcome was not affected, the error
remained undetected until Election Day had passed
and observers noted that the candidate received an
uncommonly low number of votes.
293
ES&S
Optech III-P
Eagle
Alabama
Baldwin County, AL
The Birmingham News and the New York Times
reported that an error in the way officials
downloaded vote data from a computer cartridge led
to an incorrect initial tally of votes in the
gubernatorial election. The initial tally of the votes
showed that the Democratic incumbent had received
19,070 votes in Baldwin County. A reexamination of
the vote tallies showed that the incumbent received
only 12,736 votes, which gave the victory to his
Republican challenger. The incumbent initially called
appendix b : reports of voting system issues
Below is a list of close to 200 reports of inaccurate vote tallies and other problems using voting
systems over the last few years. These reports come from news stories collected by Common Cause,
Voters Unite, ACCURATE and the Brennan Center. News items about voting system troubles often
do not include many details; this makes it hard to identify from these reports the precise cause of a
particular malfunction. Whatever the causes of a particular problem, it is fair to assume that their
occurrence in one jurisdiction will often eventually be repeated in another unless election officials
throughout the country are made aware of both the causes of the problem and how to avoid them. A
central and national database, to which vendors were required to report, would be one way to ensure
that election officials had access to information they might need, so that they could ask voting system
vendors the right questions and take steps necessary to avoid repeating the kinds of problems listed
below.
!
Brennan Center for Justice | 47
for a recount of all counties and questioned the
legitimacy of the election until he finally conceded
two weeks later.
294
ES&S
Model 100
Arizona
Cochise County, AZ
The Douglas Dispatch reported that, in Cochise
County, during the 2008 primary presidential race,
“a computer glitch that kept counting five polling
places over and over againfor five timescaused
[a] reporting error” of the election’s results. County
Election officials reportedly discovered and fixed the
“computer error” well after midnight on election day
and only after they had sent their final report to the
Arizona Secretary of State’s Office. Consequently, the
error resulted in Mitt Romney erroneously being
declared winner of Cochise County over John
McCain in news reports on the day after the election.
Cochise County Election Officer, Tom Schelling
stated that “‘[i]t was a cumulative (computer) error
that just kept adding the results for five polling places
every time new figures were added.’” Moreover, “the
error got worse when the cumulative error went
through five updates. It was then realized that the
total number of ballots cast according to the wrong
report was more than the people registered in the
county, Schelling said.” This realization “led to a
scramble to correct the problem and change the
official count to 100 percent of the 18 polling places
having reported to the state, [Schelling] said. By
then, the newspaper was being run.”
295
ES&S
Optech IV-C
Arizona
Maricopa County, AZ
According to the Arizona Republic, the original totals
in the Republican primary for State House in District
20 showed that one candidate led his closest
competitor by only four votes. The small margin led
election officials to conduct a recount.
296
The
Republic reported that the optical scan recount found
nearly 500 additional votes for the five candidates in
the race, and the initial second place candidate won
the election by 13 votes.
297
ES&S
Model 650
Arkansas
Carroll County, AR
According to the Carroll County News, in the 2008
general election, the inked black block on the upper
left hand corner of some of the ballots did not
contain enough ink to be processed by the counting
machine. After receiving advice from ES&S, the
48 | Brennan Center for Justice
machine’s vendor, workers manually inked the black
block on the ballots. Moreover, they had to have the
ballots initialed by at least three separate individuals
before feeding them back into the counting machine,
a time-consuming process.
298
ES&S
iVotronic
Arkansas
Jefferson County, AR
The residents of the towns of Redfield and Altheimer
who cast early ballots for the aldermen elections in
the towns were supposed to be able to vote for
alderman candidates from any ward. According to
the PineBluff Commercial, however, the electronic
ballots presented to voters in these towns only
allowed them to vote for candidates of the ward in
which the voters were registered. The County
Election Commission discovered and fixed the error
by, among other things, using paper ballots for the
aldermen races. Voters who had already cast early
ballots, however, could not re-vote without a court
order.
299
ES&S
Model 650
Arkansas
Faulkner County, AR
The Log Cabin Democrat reported that the initial
result of the House District 45 election in Faulkner
County was incorrect. The mistake was reportedly
the result of three cartridges not being included in
the vote tally (i.e., human error) and an error in a
voting machine that caused votes from another race
to be included in the House District 45 race. Local
officials were unsure whether human error or a
machine malfunction caused the mistake with the
machine.
300
ES&S
iVotronic
Arkansas
Benton County, AR
The Benton County Daily Record reported that a
number of the Benton County polling sites had
problems with their electronic voting machines,
including failure of the machines to boot up. Many
sites used paper ballots until the machines were fixed.
Results of the election were not compromised.
301
ES&S
iVotronic
Arkansas
Faulkner County, AR
The Log Cabin Democrat reported that when a voter
reviewed the voting machine printout after casting
her ballot, she discovered only a short horizontal line
and a long vertical line. Due to the problem with the
printer, the vote went unverified. It did, however,
still count.
302
Brennan Center for Justice | 49
ES&S
iVotronic
Arkansas
Cleburne County, AR
According to the Heber Springs, Arkansas Sun Times,
some voters who took part in early voting reported
that the mayoral candidate they selected was switched
when they saw their vote on the review screen. The
county clerk confirmed the problem. The vote-
switching was blamed on a calibration error and was
not caught until 252 votes had been cast on the
questionable voting equipment.
303
ES&S
iVotronic
Arkansas
Washington County, AR
The voter-verifiable paper record printed the
incorrect voting district numbers for each state
representative candidate. As reported by the
Northwest Arkansas Times, a candidate in the race
caught the error as he tried to cast his own ballot.
ES&S re-programmed the machines and the county
election coordinator insisted that the error “never
affected the tallying of votes,” just the district listed
on the audit trail.
304
ES&S iVotronic
Arkansas
Pulaski County, AR
The Arkansas Democrat-Gazette reported that after
several voters complained of problems with the
machine used in early voting, officials determined
that an optical illusion experienced by voters over six
feet tall caused many to select the candidate above the
one for whom they intended to vote.
305
According to local television station THV, election
officials who contacted the manufacturer to ensure
that there wasn’t a problem with the machine were
told that the company was aware of an issue with
optical illusions causing voters to select the
unintended candidate.
306
For more on this incident, see case study 4 on page
14 of this report.
ES&S
M115
Arkansas
Carroll County, AR
According to the Carroll County Star Tribune, an
incorrectly programmed chip from an optical scan
system skewed results from the race for Justice of the
Peace in District 2. Election officials fortunately
discovered the glitch when they met to certify the
elections. As a result, the ballots were recounted.
307
ES&S
M150
Arkansas
Craighead County, AR
50 | Brennan Center for Justice
The Jonesboro Sun reported that the initial results of a
constable race in District 13 showed that one
candidate received all 158 votes cast in one precinct.
When the opponent questioned the results of the
elections, the machine was inspected and an error was
found in a computer chip’s code. A recount showed
that both candidates received votes, though the
outcome of the election was unchanged.
308
ES&S
M150
Arkansas
Fulton County, AR
According to the South Missourian, a malfunction in
a ballot scanner caused county election officials to
recount ballots for the primary election by hand.
County officials blamed the machine manufacturer
for incorrectly programming the machine. The
company blamed the county officials for not sending
all of the sample ballots needed for the company to
program the machines accurately.
309
Premier
Accu-Vote ES
2000
California
Humboldt County, CA
According to the Times Standard, “a glitch in the
[Humboldt] county election's software . . . resulted
in almost 200 ballots not being included in initial
vote totals and the county certifying inaccurate
election results.”
Carolyn Crnich, the Humboldt County Registrar of
Voters, told the paper that she realized there was a
problem when “she discovered a deck of 197 vote-by-
mail ballots for the precinct that had been run
through the ballot counting optical scanner, but did
not seem to appear in the final vote tallies.”
310
Wired and Computerworld magazines reported that
the voting system manufacturer, Premier, was aware
of the software flaw that caused the problem for
years, but failed to notify federal or state authorities,
opting instead to send an e-mail to county officials.
Crnich states that the information in this e-mail was
not passed onto her.
311
For more on this incident, see case study 2 on pages 12 -
13 of this report.
Sequoia
AVC Edge II
California
Santa Clara County, CA
In Santa Clara County, the San Jose Mercury News
reports, “fifty-seven electronic voting machines for
the disabled malfunctioned Tuesday and could not
Brennan Center for Justice | 51
be replaced for hours, despite requirements that they
be available in each precinct.” Indeed, “backup
machines took three hours to prepare, with some
arriving as late as 2 p.m.” It was unclear what caused
the malfunction of the machines.
312
Sequoia
Optech 400C
California
San Bernardino County, CA
According to the Press Enterprise, a “computer glitch
in the San Bernardino County registrar of voters'
ballot-counting system sent some losing candidates in
Tuesday night's election to bed thinking more
precincts were left to be heard from when, in fact, all
the votes were in and counted.
San Bernardino County voter Registrar Kari Verjil
told the paper, “the glitch is embedded in computer
vote-counting software provided by Sequoia Voting
Systems, the Denver-based contractor that serves
many counties in California.”
A staffer in the registrar’s office told the paper that
San Bernardino County officials discovered the glitch
last year and told the manufacturer. A Sequoia
spokeswoman acknowledged that San Bernadino
reported problems and stressed that the issue “has
absolutely nothing to do with vote totals or
tabulations.”
313
ES&S
Model 650
California
Sacramento County, CA
The Sacramento Bee reports that “improper
maintenance of some of Sacramento County's voting
machinesand the tint of the Feb. 5 ballotswere
to blame for malfunctions that sidelined vote-
counting scanners and delayed results of last month's
presidential primary, according to the county's top
election official.”
The county, during its investigation, “said that the
vendor that supplies and maintains the scanners,
Elections Systems & Software, conducted improper
recalibration and preventive maintenance on the
machines in December.” Moreover, according to the
Bee, “the [county] report said that ballots printed by
Consolidated Printers were too dark to allow the
ballot to be correctly read by the faulty scanners.
314
Sequoia
Optech 400C
California
Santa Clara County, CA
The San Francisco Chronicle reported that, at one
polling station, “more than a dozen people trying to
52 | Brennan Center for Justice
vote in the Democratic primary at that polling
station were almost turned away without casting
ballots after paper ballots ran out and the only
electronic voting machine at the polling place
malfunctioned.” An “independent election observer
and . . . a software engineer waiting to vote -- who
fixed the touch-screen machine -- saved the day for
the 13 people waiting to cast their ballots.” The
Chronicle contained no additional information
regarding the “malfunction” or what caused it.
315
Sequoia
Optech 400C
California
Riverside County, CA
“The $500,000 ballot counting system bought by the
Riverside County Registrar three months ago
malfunctioned on election night,” the Desert Sun
reported, “delaying results from [the] presidential
primary.” Specifically, “three of the six counters had
problems, including one that did not work for four
hours.” According to the paper, the ballot scanners
“ran at a tenth of the advertised speed.” Although a
Sequoia spokeswoman stated that “‘[t]he slowness of
the machines is due to the printing problems
Riverside experienced with the vendor,’” the County
Registrar of Voters did not comment on what caused
the delays.
316
ES&S Optech
IV-C
California
San Francisco County, CA
According to Information Week, the conversion
function that translates ballot images into computer-
readable data shut down after the amount of data
exceeded the limit set by the vendor, apparently as a
result of the city’s switch to a ranked-choice voting
system.
317
Sequoia
Edge
California
Sacramento, CA
Wired magazine and the Tri-Valley Herald reported
that when Sequoia demonstrated its voting machines
to the state officials, its touch-screen machine
outfitted with a paper trail failed to report votes on
Spanish language ballots. “The paper trail itself
seemed to work fine but what it revealed was when
[the Sequoia representative] demonstrated voting in
Spanish, the machine itself did not record his vote,”
Darren Chesin, staff director for the Senate Elections
and Reapportionment Committee, told the Herald.
“Programming errors can occur and the paper trail
was the way we caught it.”
318
Brennan Center for Justice | 53
Diebold
AccuVote-OS
California
San Diego, CA
Weeks after Election Day, local officials discovered
an error in eight Diebold scanners that had been used
on 208,446 absentee ballots. According to the North
County Times, votes were miscounted in both the
Democratic presidential primary race and the
primary race for the Republican U.S. Senate seat.
319
A recount was conducted, revealing that “2,821
absentee ballots cast for Democratic presidential
hopeful John Kerry were actually counted for Dick
Gephardt.” Similarly, in the Senate race, 68 votes for
one candidate and six votes for another were credited
to a third candidate. The Union Tribune reported
that multiple scanners caused the error, feeding data
into the tabulation system at once.
320
Diebold
PCM500
California
Alameda and San Diego Counties, CA
According to the San Diego Union Tribune, more
than 700 precinct control modules that activate the
cards used to call up ballots on touchscreen machines
displayed the wrong start-up screen on primary day,
causing delays of up to three hours at the polls.
321
At a hearing called by the Secretary of State’s office to
investigate the problems during the primary, a
Diebold technician called the problem – which
occurred when the batteries in the equipment
discharged beyond a certain point – “fully known.”
322
For more on this incident, see case study 8 on pages 16 –
17 of this report.
Premier
AccuVote-TSX
Colorado
Adams County, CO
The Colorado Independent reported that “[a]n
electronic voting machine in Adams County
repeatedly failed to accept [one voter’s] vote for a
Democratic state Senate candidate—instead
registering the vote for her Republican opponent—at
an early voting site . . . and [that the machine] has
been removed from service.” “The voter reported the
problem to election judges, who canceled her ballot
and allowed her to vote—this time accuratelyon
another machine, . . . .” After quarantining the
machine, Adams County Clerk and Recorder Karen
Long “said that election officials hadn’t received any
other complaints about the particular machine and
that no other voting errors have been reported.”
323
54 | Brennan Center for Justice
Hart InterCivic
Kodak i840
Colorado
Douglas County, CO
According to the Denver Post, “ballot scanning
machines from Hart InterCivic continue to suffer
from the same problems that led them to be
decertified last year.” The Post states that the “optical
scanners too often read stray marks as votes, even if
they are just the tiny dots from somebody resting a
pen on the ballot before marking a box, according to
the reports [by the Colorado voting machine testing
board].
324
Sequoia
Insight 400 C
Colorado
Denver, CO
The Rocky Mountain News reported that vote
counting in an election for several local offices took
nearly a week because 70,000 absentee ballots had
been mailed out with barcode misprints. The Sequoia
scanners could not sort ballots correctly by district, so
election officials had to do so by hand. Additionally,
election officials reported that many voters had used
incorrect ink pens or marked their ballots in a
manner that made it unlikely that the scanners could
record their votes. As a result, at least 5% of absentee
ballots had to be transcribed onto unused ballots by
poll workers in order to be scanned.
325
Diebold
AccuVote-OS
Colorado
Pitkin County, CO
According to the Aspen Times, almost 1,200 phantom
votes were reported in one precinct. Election officials
initially announced that 1,560 people voted in
Precinct 5, but later official results showed that 374
people had voted in that precinct. Apparently, the
mistake was not caused by an electronic machine
error, but instead by faulty spreadsheet formulas used
to tally votes from each precinct.
326
Premier
AccuVote-OS
Connecticut
Litchfield County, CT
The News-Times reported that the “new optical
scanning machine at [New Milford]’s Lanesville
district malfunctioned, which necessitated the hand
counting of all 426 votes there.” At the time, it was
unclear what caused the problem.
327
Premier
AccuVote-OS
Connecticut
Hartford County, CT
According to the Boston Globe, “[a] new optical
scanning machine did not work and voters had to put
their ballots in an auxiliary slot to be counted later.”
The Globe reported no further information regarding
Brennan Center for Justice | 55
the machine or why it did not work.
328
Premier
AccuVote-OS
Connecticut
Hartford County, CT
Newsday reported that “[s]ix out of the 825 [optical
scan voting machines] used across the state had to be
replaced with back-ups. Two jammed and three were
not properly programmed, [Windham Democratic
Registrar Paulann Lescoe] said.” It was unclear why
the last of the six machines had to be replaced.
329
Sequoia
Optech III-P
Eagle
District of
Columbia
Washington, DC
The Washington Post reported that a malfunctioning
cartridge for a Sequoia precinct-count optical scanner
caused vote totals to be duplicated into multiple races
and reported an implausibly high overvote rate of
1,542 in five contests.
330
According to a report on the incident that Sequoia
sent the Board of Elections shortly after the incident,
the manufacturer could find no irregularities in the
system and attributed the problem to human error.
331
For more on this incident, see case study 12 on pages 21
- 22 of this report.
Sequoia
Optech III-P
Eagle
District of
Columbia
Washington, DC
According to the Washington Post, “[s]ome polling
sites . . . [dealt] with jammed machines, which were
caused when ballot clerks failed to properly tear the
stubs off paper ballots fed to the machines to be
counted.” Moreover, “[a]t Metropolitan AME
Church on M Street NW, the ballot-counting
machine broke down. Edgar R.M. Frazier, the
precinct captain, said Board of Elections officials
brought another one ‘within 40 minutes.’” It was
unclear at the time what caused the breakdown.
332
Premier
AccuVote-OSX
Florida
Hillsborough County, FL
The Tampa Tribune reported that, according to
Hillsborough County election supervisor Buddy
Johnson, a delay of election results was caused by
“problems with the . . .elections system purchased
from Premier Election Solutions. Among the
problems: Election officials had to break up early
voting results into smaller pieces to prevent crashing
the computer system, and about 150 optical scanners
had trouble transmitting results and had to be
manually taken to the elections office.”
333
56 | Brennan Center for Justice
Sequoia
Optech Insight
Florida
Palm Beach County, FL
The Palm Beach Post reported that delays in the
tabulation of election results were caused “when
computers were unable to read cartridges from
optical scanners at six polling sites. . . . Exactly what
caused the cartridges to malfunction was unclear, said
Assistant County Administrator Brad Merriman.”
334
Sequoia
Optech Insight
Florida
Palm Beach County, FL
An optical scanner jammed at a local precinct
according to local NBC affiliate WPTV. Voters thus
had to place their votes in the “Emergency Ballot
Box” to be scanned later. As one voter put it: “‘I
voted. I went to put my vote in the machine and the
machine was clogged, so they said ‘stick it in this side’
which is the emergency bin and I couldn't put it in
there because the emergency bin was full,’ says Liz
Huhn of West Palm Beach.” Thus, “[m]any ballots
ended up in cardboard boxes . . . .” According to
WPTV, “[e]lections officials sa[id] ballots that did
not go through the machines, w[ould] be run
through the machines . . . after everyone in line has
voted.”
335
Premier
AccuVote-OS
Florida
Manatee County, FL
According to the Herald Tribune, “a software
problem . . . prevent[ed] Manatee elections officials
from counting nine of the county's voting precincts
and two-thirds of the ballots cast during early voting.
When the ballots [were] attempted to be uploaded
into the county voting system, [election officials]
g[o]t an error message that prevent[ed] the tallying of
those votes.”
336
A later Herald Tribune report indicated that “[a]
simple change in one Manatee County race, where a
candidate dropped out . . . , caused the Premier
software program known as GEMS to lock up. That
prevented the counting of 46,000 ballots on Tuesday
night, Supervisor of Elections Bob Sweat said.”
Manatee officials eventually “fixed the glitch; results
were released at noon and no election outcomes
changed.”
337
ES&S
DS-200
Florida
Thirteen Florida Counties
A study from the Florida Fair Elections Center shows
that counties using the ES&S DS200, which in the
Brennan Center for Justice | 57
event of an overvote displayed a confusing message
and did not automatically reject a ballot, had an
overvote rate on Election Day 2008 that was as much
as 18 times that of systems used in other Florida
counties.
338
After being alerted to the results of the Florida study,
the State of Wisconsin completed its purchase of the
same system on the condition that overvotes would
be automatically rejected. New York, which was
unaware of the Florida study at time of purchase,
opted to set up machines to work in the same way
they had in Florida in 2008.
339
Sequoia
Optech Insight
Florida
Palm Beach County, FL
The Palm Beach Post reported that creases in some
absentee ballots appeared to cause the high-speed
optical scanner to read the creases as votes: “When
elections officials this week tried to figure out why
good ballots were being rejected, they began
suspecting that the machines are incorrectly reading
the crease as a vote. That would mean they are
incorrectly reporting that a person voted for more
than one candidate in a race that lines up near the
fold.” This reportedly caused the machines to read
the creased ballots as overvotes. “Since so-called
overvotes aren't allowed, the machines are
programmed to reject ballots where they occur.
Those ballots, along with ones that voters didn't
mark properly, are then given to the canvassing board
to review.” Consequently, election officials “[r]ather
than risk miscounting votes in any race . . . are asking
staff to duplicate all questionable ballots. They will
then review those ballots again to ensure they are
duplicated properly.”
340
Premier
AccuVote-OS
Florida
Duval County, FL
First Coast News reported that nearly half of the
voting machines in Jacksonville were malfunctioning,
indicating the ballot was too long. Officials insisted
that they had measured the ballots and they were
regulation length.
341
A report from a local television affiliate stated, that
officials eventually learned that “the real problem was
with the ballot printers.” “If the paper is not loaded
58 | Brennan Center for Justice
properly, it can be skewed just enough that the
timing marks on the side will produce a ballot that
will not be read,” an official told the station. “Due to
the tolerance of the machines, some machines may
accept it and some may not.”
342
Premier
AccuVote-OS
Florida
Flagler County, FL
The Flagler News-Journal reported that “Flagler
County elections officials switched voting machines
for early voting Tuesday to eliminate a problem that
had kept them from scanning 96 ballots during early
voting hours Monday.” Election Supervisor Peggy
Rae Border told the paper that “the county's vendor
thinks slightly shorter ballots1/32 of an inch in
some casesmay have caused the problem. She said
she suspected the new machines were more sensitive
to the variation in ballot length.” The 96 ballots
were eventually successfully scanned.
343
Permier
AccuVote-OS
Florida
Brevard County, FL
According to Florida Today, “[a] glitch involving a
mere 20 votes tallied on touch-screen voting
machines Tuesday caused a two-hour delay in
election results . . . . About a third of the 140
machines malfunctioned.”
Interim Supervisor of Elections Bert Childress told
the paper that the problem had been identified the
previous year, and that the manufacturer was unable
to get necessary replacement parts to the county in
time to remedy the defect. “‘They told us it was our
phone lines,’ Childress said. ‘They finally admitted in
July that it was a problem with their machines. About
a third of our machines failed to download
properly.’” According to the paper, Premier agreed to
replace 150 modules at no cost, but told Brevard
officials that the units would not be ready in time for
the primary.
344
ES&S
DS200
Florida
Pinellas County, FL
The St. Petersburg Times reported that “[i]n Pinellas,
12 scanners had to be replaced at voting precincts,
according to county elections spokeswoman Nancy
Whitlock. Some experienced paper jams, and on
some the screens froze, but they were traded out with
functioning ones in a matter of minutes, she said.”
No further information regarding the problem was
provided.
345
Brennan Center for Justice | 59
ES&S
DS200
Florida
Pasco County, FL
According to the St. Petersburg Times, eighteen of the
147 machines in Pasco County malfunctioned when
election officials tried to transmit election results via
modem. Results had to be transmitted via a thumb
drive. Five machines experienced other problems.
346
Premier
AccuVote-OSX
Florida
Hillsborough County, FL
According to WMNF Radio, “[r]esults from
yesterday’s primary in Hillsborough County were
slow to be released. Supervisor of Elections Buddy
Johnson blames a software issue with the new optical
scan machines manufactured by Premier Election
Solutions.”
Director of Government Affairs for Premier Kathy
Rogers stated, in response, that “her company did not
provide adequate documentation for their GEMS
software to Hillsborough County to avoid the
problem. The problem was minor, Rogers said, and
only affected how fast the results were made public,
not the actual results.
347
ES&S
AutoMARK
Florida
Sarasota County, FL
The local ABC affiliate WWSB reported that on
election night, absentee ballots had to be hand-
counted when the optical scanning machines would
not communicate with the server. More than 10,000
absentee ballots had to be hand counted and then re-
scanned.
“‘We could not get the absentee ballots totals to
upload into the main server to combine all of the
totals together for absentee early voting,’ says
Supervisor of Elections Kathy Dent. So workers had
to count them individually. ‘We had to manually
add those together, and obviously when you're doing
something manually it takes a lot more time.’” At
the time, it was unclear what caused the problems
with the server.
348
Sequoia
Optech Insight
Florida
Palm Beach County, FL
The Palm Beach Post reported that, after the special
city commission election in Palm Beach County,
“[n]early 700 votes from three precincts14 percent
of the total castwere added into the final results
released by the supervisor of elections office after the
60 | Brennan Center for Justice
standard post-election audit Wednesday and
Thursday.”
During the audit, “three cartridges containing vote
totals were labeled ‘suspended,’ meaning their votes
hadn't been counted on election night when all the
cartridges were brought to a tabulation center to be
‘read’ by vote-counting machines, said elections office
spokeswoman Kathy Adams.
Adams “said the office didn't know why the
cartridges weren't read properly the first time. She
said it was possible that one reader wasn't working
properly and that all three cartridges were read by
that reader.”
349
ES&S
DS200
Florida
Pinellas County, FL
According to the St. Petersburg Times, two software
viruses attacked the center tabulation system in
Pinellas County. The worms were eventually safely
removed and did not corrupt the software or affect
any elections.
350
Premier
AccuVote-TSX
Florida
St. Lucie and Okeechobee Counties, FL
TC Palm reported that poll workers at three precincts
in both counties were unable to send the voting data
electronically to the central election office. “Votes
were delayed from two precincts at Harbour Place, an
assisted living facility in Port St. Lucie, as poll
workers tried unsuccessfully to electronically send
them over, said Supervisor of Elections Gertrude
Walker.” It was unclear why the poll workers could
not send the voting data electronically.
351
ES&S
iVotronic
Florida
Broward County, FL
According to the Sun-Sentinel, new electronic voter
check-in machines caused confusion and long waits
for primary voters as some staffers were unfamiliar
with the new equipment. Moreover, a few voters
complained of problems with party registration.
352
Sequoia
Optech Insight
Florida
Palm Beach County, FL
Supervisor of Elections Arthur Anderson told the
Palm Beach Post that a “defective early-voting
cartridge” slowed the production of vote results.
“Although a backup tape allowed elections staff to
recoup the results, Anderson said the problem was so
Brennan Center for Justice | 61
significant it may lead to the elections office having
to reprogram all of its voting machines.
353
Premier
AccuVote-OS
Florida
Volusia County, FL
After scanning 3,986 ballots and noticing a four
ballot discrepancy, election officials planned to scan
the ballots a second time according to the Daytona
Beach News Journal. The Supervisor of Elections,
Ann McFall, blamed the error on the voting
machine: “McFall said she received an advisory from
the equipment's vendor last week saying similar
problems had been encountered elsewhere. ‘The
other sites balanced perfectly,’ McFall said. ‘I think
it's the machine.’
354
ES&S
AutoMARK
Florida
Sarasota County, FL
The Herald Tribune reported that the voting
machines were not working and had to be replaced.
“‘They were all tested before and for various reasons
they just went bad,’ said Kathy Dent, Sarasota
County supervisor of elections.
“Some machines had problems with the memory
card, while others had a faulty scanner. Dent said
backup machines are located downtown and in south
Sarasota County, and crews are on standby to rush
the new machines to precincts where problems are
reported. ‘They are replacing the scanners as soon as
we get the call,’ Dent said.”
355
ES&S
DS200
Florida
Miami Dade County, FL
The Miami Herald reported that for reasons that are
unclear, a ballot for the presidential primary
candidates did not appear on at least one voter’s
machine.
356
Premier
AccuVote-OSX
Florida
Hillsborough County, FL
10Connects.com reported that some voters had trouble
casting their votes on the touch-screen machines. A
machine’s screen went blank. Moreover, some of the
machines were reading votes for one candidate as
having been for the candidate that the voters had not
chosen.
“Supervisor of Elections Buddy Johnson says the
current touch screen system will be replaced this
spring in time for November’s General Election. ‘It
counts correctly, but it’s been like steering a car
62 | Brennan Center for Justice
where you can’t get the alignment right sometimes,’
Johnson says. ‘It’s just that it is difficult to
manage.’”
357
Sequoia
AVC Edge
Florida
Palm Beach County, FL
Voters told the South Florida Sun-Sentinel that vote-
switching had occurred as they cast their ballots in
the 2006 House race. Additionally, others reported
problems when the wrong ballots appeared on their
touch screens while voting.
358
ES&S
iVotronic
Florida
Sarasota County, FL
According to the Sarasota Herald Tribune, there were
over 18,000 undervotes in the 13th Congressional
race between Vern Buchanan (R) and Christine
Jennings (D).
359
Buchanan defeated Democrat
Christine Jennings by a slim 369 vote margin.
360
The
Tribune reported on numerous complaints that the
13th Congressional District race was not appearing
or recording properly on the touchscreen voting
machines in use in the county. Some voters told the
paper that votes for Christine Jennings failed to
appear on the review screens at several polling
locations.
361
ES&S Optech
Eagle
Florida
Orange County, FL
The Florida Division of Elections reported that
Orange County experienced the highest undervote
rate in the state for absentee ballots cast in the U.S.
Senate and Governor’s races.
362
According to the Flordia Fair Elections Center,
which contacted Orange County about the undervote
rate, the machines’ vendor told county officials that
the issue appeared to be a calibration problem that
caused the machines to fail to read certain kinds of
gel ink. Further investigation by the Center revealed
that the same problem had occurred on ES&S and
Sequoia scanners in previous elections.
363
For more on this incident, see case study 3 on page 13 of
this report.
ES&S
iVotronic;
M650
Florida
Broward County, FL
The Miami Herald reported numerous complaints
that votes cast for the Democratic gubernatorial
candidate were given to his Republican opponent
during early voting. A spokeswoman for the
Brennan Center for Justice | 63
Supervisor of Elections told the paper that the
machines likely needed to be recalibrated.
364
ES&S
M650
Florida
Broward County, FL
According to the South Florida Sun-Sentinel, a
programming error in an optical scan
tabulator would have changed the outcome of a
ballot measure had it not been caught by alert
election officials. The optical scan tabulator was used
to count absentee ballots. The glitch occurred
because the machines in use were programmed to
accept only 32,000 votes per discrete ballot item,
after which it started counting backward.
Once the machines were reprogrammed to allow for
counting of all of the votes, the measure received
more than 64,000 “yes” votes and it passed.
365
For more on this incident, see case study 6 on page 15 of
this report.
Diebold
AccuVote OS
Florida
Volusia County, FL
According to local media, seven Diebold optical scan
memory cards failed, causing votes to disappear.
366
One card’s tally was missing 13,244 votes.
ES&S iVotronic
Florida
Eleven Florida Counties
Five months before the 2004 general election, Miami
Herald reported that state election officials had
learned of a flaw in the touchscreen voting machines
in use in eleven of the state’s counties that would
prevent the conduct of a manual recount in the event
of a close race.
367
It later came to light that Miami-Dade county
discovered the problem more than a year earlier,
when the director of the county’s technology
department found during an audit that the electronic
log of voting activity produced by the system
scrambled the machines’ serial numbers.
368
For more on this incident, see case study 7 on page 16 of
this report.
ES&S
iVotronic
Florida
Broward County, FL
CNN reported that a software error caused 103,222
votes, cast with ES&S iVotronic paperless machines,
to be left uncounted in the original tally. The error
64 | Brennan Center for Justice
was discovered the morning after Election Day.
When the missing votes were added, voter turnout
for the county was adjusted from 35% to 45%.
369
ES&S
M100
Florida
Union County, FL
Black Box Voting reported that a programming error
in Union County, Florida caused machines to read
2,642 Democratic and Republican votes as entirely
Republican in the September 2002 election. The
vendor, ES&S, accepted responsibility for the
programming error and paid for a hand recount.
370
ES&S
iVotronic
(M650
Unity Election
Reporting
Management
System used to
combine totals)
Florida
Miami-Dade County, FL
According to the Miami Herald, the vote tallies for
two city council races in Medley, Florida were
miscounted when the results of the absentee votes
from optical scan machines were combined with the
results of the electronic ballots. The initial count
showed victories for two candidates who had actually
lost the election. The Miami-Dade elections
supervisor told the paper that all software had been
tested before the election without a hitch, but poll
workers noticed the problem as they fed results into
the computers. Evidently, a technician from the
voting machine manufacturer inadvertently bumped
the first candidate to the last position when he
opened the ballot program on the memory cards to
change the heading of the ballot. When the results of
the two systems were combined, they didn’t match
properly.
371
Premier
AccuVote TS
R6
Georgia
Fulton County, GA
In North Fulton County, according to the Atlanta
Journal-Constitution, a group of voters had to vote
using provisional ballots because the election database
reported that they already voted; they had not. Also,
in south Fulton County, all voting machines in a
precinct were down for about an hour. While the
machines were down, the polling place ran out of
printed ballots and paper provisional ballots. An
hour later, however, the machines began to work
again.
372
Premier
AccuVote TS
R6
Georgia
Fulton and DeKalb Counties, GA
The Atlanta Journal-Constitution reported that people
in line to vote had to, at times, wait anywhere from
45 minutes to 8 hours. The delay was caused by
problems with the computers used for check in. The
Brennan Center for Justice | 65
computers were sluggish and sometimes crashed. It
was unclear why the machines crashed.
373
Premier
AccuVote TS
R6
Georgia
Fulton County, GA
The local CBS affiliate reported that “[c]omputers at
the North Fulton Government Center and other
county polling places lost their connection to the
state's database, officials told CBS 46’s Joanna
Massee. That glitch created a big headache for
election workers and long waits for voters.” It was
unclear why the computers lost their connection to
the database.
374
Premier
AccuVote TS
R6
Georgia
Muscogee County, GA
In Muscogee County, according to the Ledger-
Enquirer, the express poll machine was not
“functioning properly” so 16 voters had to cast
provisional votes. “‘It's not that they didn't work,’
said Nancy Boren, executive director of the
Muscogee County Office of Elections and
Registrations. ‘It was that they were slow in coming
up.’
“Once the machines were functioning correctly,
voting returned to normal. The 16 provisional ballots
will be counted as regular votes, Boren said.
375
Premier
AccuVote TS
R6
Georgia
DeKalb, Fulton and Cobb counties, GA
According to the Atlanta Journal-Constitution, four of
five voting machines went offline and became
inoperable.
376
Premier
AccuVote TS
R6
Georgia
McIntosh County, GA
The Florida Times-Union reported that the electronic
voting machines did not work when the polls first
opened. “‘The touchscreen voting machines had
worked well when tested, but didn't when the
precinct first opened,’ [election official Phillip
Kempton] said.” Poll workers recorded the names of
the individuals they had to turn away from voting.
They later contacted them when the machines were
working, and got them to come and vote. It was
unclear why the machines did not work initially.
377
Diebold
TS-R6; TS-RS
Georgia
DeKalb, Fulton, and Cobb Counties, GA
According to the Atlanta Journal-Constitution, touch-
screen voting machines in three counties switched
votes for Democratic candidates to Republicans.
66 | Brennan Center for Justice
Technicians were dispatched to recalibrate machines.
The vote-switching problems significantly slowed
down the voting process for those who went to the
polls. By mid afternoon, at least 30-40 voters
reported experiencing the vote-switching problem.
378
Hart
eSlate 3000
Hawaii
Honolulu, HI
A local TV news station reported that voting
machines gave voters the option of selecting a Green
Party ballot, although there were no Green Party
candidates. As a result, 22 voters wasted their votes
and were essentially disenfranchised.
379
ES&S
M550
Idaho
Bannock County, ID
According to the Casper Star Tribune, ES&S scanners
failed to recognize the ink that the vendor instructed
county election officials to use for ballot marking.
Once the uncounted ballots were identified, poll
workers marked each ballot with ink that would be
recognized by the scanner and fed them into the
machine for a second time.
380
Premier
AccuVote-OS
Illinois
St. Clair County, IL
The Bellville News-Democrat reported that a mayoral
candidate contested the results of a mayoral race in
St. Clair County after “a voting machine malfunction
caused several hundred ballots to go uncounted.” No
further information was given regarding the
malfunction.
381
Premier
AccuVote-OS
Illinois
St. Clair County, IL
In the village president race in St. Clair County,
according to the Bellville News-Democrat, 400
uncounted ballots affected the results of the race. The
origin of the uncounted ballots: “‘There were two
machines at [a certain] polling place, one with the
touch screens for people with disabilities and optical
scan everybody else votes in,’ [County Clerk Bob]
Delaney said. ‘When voting was done, we uploaded
both cards to be counted. But for some reason (the
computer) did not count optical scan.’” It was
unclear why.
382
Premier
AcuVote-TSX
Illinois
DuPage County, IL
The Chicago Daily Herald reported that during a
special election, the Diebold (Premier) touch-screen
election machine had the word “Republican
flashing at the beginning of each ballot as though the
Brennan Center for Justice | 67
election was a primary election. To solve the
problem, some election officials placed tape on the
screen over where the word flashed.
383
ES&S
Model 100
Illinois
Lake County, IL
The Chicago Daily Herald reported that the optical
scan vote-tabulation machine was unable to transmit
electronically the voting results. Thus, election
officials were forced to drive the machines to the
nearest transfer station for tabulation. Election
officials launched an investigation to determine the
cause of the problem.
384
Sequoia
Optech Insight
Illinois
Cook County, IL
According to the Daily Northwestern, votes were
miscounted at one precinct in Cook County during a
local race. The results “were missing 247 of the 540
ballots.”
385
Sequoia
AVC Edge II Pl
Illinois
Chicago and Cook County, IL
According to the Chicago Tribune, Chicago and
Cook County officials said they had received a few
complaints from early voters about pressing once
candidate name on the touch screen and having the
machine register another. This type of event can
occur if the machine is not adequately calibrated.
386
ES&S
M100
Illinois
Lake County, IL
The Chicago Tribune reported that because of a
programming error that failed to account for the
option of “no candidate” on the ballot, election
results were placed next to the names of the wrong
candidates in four different races. The problem was
corrected by 10pm on the evening of Election Day.
387
ES&S
iVotronic
Indiana
Wayne County, IN
Officials in Wayne County ran into problems,
according to the Palladium-Item, when officials tried
to get a final tally of the vote. “When local officials
tried to tally the vote they received a ‘system error’
message. They stopped the count, secured the votes
and then had to wait for an answer to the problem
from company officials at Election Systems &
Software, which provided the computer system for
the local election.” The call to ESS alone took over
an hour but issue was resolved on the call. It was
unclear why the “system error’ message appeared.
388
ES&S
Indiana
Madison County, IN
68 | Brennan Center for Justice
iVotronic
The Herald Bulletin reported “that as many as 7,400
of the 12,000-some ballots used for early voting
could not be counted by the machines. As it turns
out, the coding on that portion of the early ballots
was in the wrong position on the paper, tripping up
the machines.”
According to an editorial in the paper, “an official
from Omaha-based Election Systems & Software,
which provided the counting system, seemed to
acknowledge that the company had sent the county
ballots that wouldn’t work. But the county should
take some blame too for not taking the precaution of
testing the new set of ballots when they arrived.”
389
ES&S
iVotronic
Indiana
Marion County, IN
According to WISH-TV, workers were unable to
retrieve votes from the 520 touch-screen machines
used by disabled voters when the polls closed at
6:00pm on Election Day. The problem occurred
because ES&S had programmed the machines for
Pennsylvania's polling hours, which stayed open until
8:00pm. Disabled voters were concerned that their
votes would not be counted. In fact, when the
incident was reported three days after the election,
those same votes had not yet been recovered or
counted.
390
MicroVote
Infinity
Indiana
47 Indiana Counties
The Star Press reported that ten days before the 2006
primary election, voting system manufacturer
MicroVote discovered a defect in the function that
allows voters to cast ballots in “split precincts,” where
voters in the same precincts choose from different
sets of candidates.
391
According to the Indianapolis Star, the manufacturer
shut down the split precinct function on the
machines and worked in secret to develop a software
upgrade that would restore the function, all without
notifying the Indiana Election Commission. The
paper noted that the Commission only learned of the
defect when MicroVote applied for certification for
the software upgrade.
392
For more on this incident, see case study 14 on page 24
of this report.
Fidlar
Indiana
Franklin County, IN
Brennan Center for Justice | 69
AccuVote 2000
ES
The Associated Press reported that a glitch caused
optical scanners to count Democratic straight-line
votes as Libertarian votes. After the error was found,
the recount changed the results of the county
commissioner’s race in favor of the Democratic
candidate. The glitch was suspected when a
Libertarian congressional candidate was receiving
four times the vote in Franklin County than in the
rest of the district. The voting machine
manufacturers called the glitch an “isolated
incident.”
393
Premier
AccuVote-OSX
Iowa
Blackhawk County, IA
In a recount in Blackhawk County, according to the
WCF Courier, out of the 19 ballots rejected by an
optical scanner, 18 of them were found to be valid
votes; the other was a no vote. “Officials don't yet
know what happened, and said they will try to
determine the mistake's origin.”
394
Premier
AccuVote-OS
Iowa
Scott County, IA
The Quad-City Times reported that, in Scott County,
vote tabulations were delayed due to two memory
card failures. “The problem started with two memory
card failures before 11 p.m., as early voting results
were being added to Election Day results. The
memory cards held the results of 11,627 ballots cast
in early voting. Election officials turned to printout
tapes of the summary reports to be manually entered
into the computer system, but the software required
more detailed and time-consuming precinct-by-
precinct information to be entered.” Poll workers
“tried to find a different way to deliver the vote totals
but didn’t complete the task until shortly before 1
a.m.,” delaying the results for hours. It was unclear
what caused the failures.
395
ES&S
M100
Iowa
Pottawattamie County, IA
According to the Daily Nonpareil, a ballot
programming error caused the new optical scan
system to tabulate votes incorrectly. When absentee
ballots were tabulated for a county recorder’s race in
the Republican primary, results showed one
candidate, a University of Nebraska at Omaha
student, had 99 votes, while his opponent, the
county recorder since 1983, had only 79. The
surprising outcome led to a recount of absentee
ballots, revealing that the incumbent actually received
70 | Brennan Center for Justice
153 votes while his challenger received 25.
396
ES&S
Optech III
Iowa
Scott County, IA
The Des Moines Register reported that optical scan
machines malfunctioned when tabulating absentee
ballots. As a result, poll workers had to manually
feed about 23,000 ballots one by one.
397
ES&S
Model 100
Kansas
Crawford County, KS
In Crawford County, the Morning Sun reported that
a programming glitch failed to follow sequencing
numbers that differentiate between ballots with
candidates in different positions. Under Kansas law,
the order in which candidates appear on a ballot must
alternate. Because the machines counted all ballots as
if they were configured the same, it misrecorded the
results for the race. “For instance, in the Crawford
County attorney’s race on ballot 15, John Gutierrez
is listed first, and Michael Gayoso Jr. second. But on
ballots 16 and 17, Gayoso is the top name, with
Gutierrez underneath. With the error then, a vote for
Gayoso on ballot 17 was counted for Gutierrez, and
vice versa. At the same time, there are some races
where a vote on ballot 16 would be the same as a vote
on ballot 17.”
County Clerk Don Pyle stated that “the glitch could
affect four races,” but that the precise outcome would
be unknown until testing was complete.
398
ES&S
iVotronic
Kansas
Saline County, KS
In Saline County, according to the Salina Journal,
during the Salina City Commission race, some
machines were flipping votes”When a voter
pressed a certain candidate's bar on the voting
machine's screen, the candidate above the selected
candidate instead received the checkmark.
“The error, [County Clerk Don] Merriman said,
most likely was caused by a calibration problem in
the machine caused during transit to the polling
places.”
399
ES&S
Optical Scan
(model
unknown)
Kansas
Clay County, KS
According to the Lawrence Journal World & News, a
computer glitch in an optical scan voting system
showed that a challenger in a primary race for county
commissioner had won, but a hand recount showed
Brennan Center for Justice | 71
that the incumbent commissioner won by a landslide:
540 votes to 175. The computer had mistakenly
reversed the totals for the candidates in one ward.
400
Hart InterCivic
eScan
Kentucky
Madison County, KY
The Richmond Register reported that results were
delayed because a data disc from a Berea precinct
voting machine was generating an error message
“[b]ecause its data could not be downloaded, not
even a preliminary [vote] total could be generated.” A
technician eventually fixed the problem.
401
Hart InterCivic
eScan
Kentucky
Kenton County, KY
In Kenton County, according to the Cincinnati
Enquirer, election officials were forced to eventually
halt voting on the handicap accessible machines. The
voting machines were not recording straight ticket
voting.
In a hearing regarding the problems, “Harp Elections
Services President Roger Baird took responsibility for
the machine's malfunctions.
“Baird blamed a proof-reading error made while the
E-slate machines were being programmed by his
staff.”
402
Danaher
Controls
Shouptronic
1242
Kentucky
Bullitt County, KY
The Louisville Courier-Journal reported that election
results were delayed. County Clerk Kevin Mooney
stated “that the delay in the tallying was mainly due
to some newer machines for special-needs voters that
had trouble reading memory cards.” It is not clear
why the machines had trouble reading the cards.
403
Hart
eSlate
Kentucky
Calloway County, KY
According to VoteTrustUSA, vote-switching
appeared to have occurred on the review screens of
Hart InterCivic eSlate machines. Straight-ticket
Democratic votes were switched to Republican
straight-ticket votes in all contested races.
404
Sequoia
AVC Advantage
Louisiana
Vermilion Parish, LA
According to the Daily Iberian, one of the candidates
in an alderman’s race “said a glitch in the Vermilion
Parish Clerk of Court’s electronic voting system
prevented District 3 voters in Delcambre from voting
for him or for [his opponent], both Democrats, from
72 | Brennan Center for Justice
the time the polls opened at 6 a.m. to 1 p.m
“Secretary of State spokesman Jacques Berry
described what occurred as a ‘reverse lockout. There
was no machine malfunction at all,’ Berry said. ‘It
was strictly a programming error that was discovered
by a voter. We determined that 60 voters had cast
votes in that precinct prior to 1 p.m. and those are
the voters who should have had the opportunity to
cast a vote in that election but did not.’”
405
ES&S
3P Eagle
Maine
Waterville, ME
The Journal & Morning Sentinel reported that
unidentified machine malfunctions caused election
results to indicate that one Senate candidate for
District 25 received 27,000 votes on one of the three
machines used by the town. 27,000 is about 16,000
more votes than the number of registered voters in
the whole city.
406
Premier
AccuVote-TS
Maryland
Anne Arundel County, MD
According to the Capital, three voting machines went
“down” as the election day began. In response, local
officials got “two machines going again by basically
unplugging and plugging them back in.” A local
official told the paper that the problems were
“beyond her ability to fix, and apparently beyond the
skills of the Diebold [Premier] technical assistant
posted at Cape St. Claire Elementary.”
407
ES&S
AutoMARK
Massachusetts
Bristol County, MA
The Standard-Times reported that in New Bedford,
one week before the state primary election, “the
Automark voting machines won't work with the
city's Republican ballots.” Elections Commissioner
Maria Tomasia “said technicians from ES&S have
about 48 hours to get the machines set up properly,
and that the problem ‘seems to be the ink,’ but that
nobody is certain.”
408
Premier
AccuVote-OS
Massachusetts
Plymouth County, MA
After a recount of a voter proposition, the Boston
Globe reported, local officials discovered that the
scanner had counted four votes wrong. Local officials
blamed the problem on the machine getting jammed
while counting the votes: “‘The machines jammed a
few times when counting the votes,’ [Town Clerk
Linda] Adams said.”
409
Brennan Center for Justice | 73
Premier
AccuVote-OS
Massachusetts
Bristol County, MA
The Taunton Daily Gazette reported that votes from
two precincts had to be counted separately due to a
“glitch” with a memory card. “The reason for the
extra work, [Taunton’s City Clerk Rose Blackwell]
said, was a result of a problem she detected last
Friday with regard to the memory card the city
received from LHS Associates of Methuen—the
company that provided the voting machines.” Local
officials said this separate count did not compromise
the outcome of the election. It was unclear what
caused the “glitch” wi
th the card.
410
Premier
AccuVote-OS
Michigan
Barry County, MI
According to the Grand Rapids Press, flawed ballot
programming caused an optical scan system to tally
votes incorrectly. The problem was discovered when
a county clerk received the voting results from the
precinct where he had voted and saw that one
candidate received no votes, despite the fact that he
himself had voted for the candidate. He also thought
it peculiar that 90 out of 127 votes cast in one
precinct selected the option to write in a candidate
on the ballot. Because there were widespread
problems with machines incorrectly tallying votes,
county workers hand counted votes for the school
board election.
411
Sequoia
Optech III-P
Eagle
Michigan
Muskegon, MI
According to the Muskegon Chronicle, Optech scan
machines failed to detect 2% of votes in the race for
Township Clerk. Initially, the machines indicated
that the incumbent had been defeated by a five vote
margin. After a recount, 39 additional votes were
factored into election totals and the challenger won
by two votes.
412
ES&S
Optical Scan
(model
unknown)
Mississippi
Forrest County, MS
The Hattiesburg American reported that the primary
election for a city council race was not certified for
days after the election because of discrepancies in vote
tallies. The number of ballots counted by optical
scanners reportedly did not correspond to the vote
totals on the voting machines.
413
ES&S
M150, M550
Montana
Yellowstone County, MT
74 | Brennan Center for Justice
According to the Billings Gazette, a programming
error caused voting machines to inaccurately record
all votes. An election official explained that he
suspected the problem had occurred because he
forgot to hit the “zero out” button required between
entering absentee and regular ballots on the machine.
Consequently, as many as 3,000 absentee ballots may
have been counted a second time when the regular
ballots were being run through the machines.
Officials decided to conduct a full recount to ensure
that the election outcome was accurate.
414
ES&S
M550
Nebraska
Lancaster County, NE
The Lincoln Star Journal reported that a number of
malfunctions occurred on optical scan machines in
use during the general election. According to the
paper, some machines shut down completely. When
the problems began, election officials stopped to test
the six machines, revealing that two were not
producing correct vote tallies. Those two machines
were shut down, but later in the day, the remaining
machines began to have similar mechanical
problems.
415
ES&S
M650
Nebraska
Sarpy County, NE
Local TV news station WOWT 6 reported that after
ballots were counted in a race for city council,
election officials realized that there were more votes
than voters. According to the officials, the error
affected 32 of 80 precincts and as many as 10,000
votes. It was believed that the glitch affected the
candidates equally and did not alter the outcome of
the elections. The evidence and explanation for the
miscount were inconclusive at the time.
416
ES&S
M100
Nebraska
Sarpy County, NE
According to the Omaha World-Herald, the optical
scan machines failed to tally “yes” votes on the
Gretna school-bond issue, giving the false impression
that the measure was defeated. The measure had
actually passed by a 2-1 margin.
417
Sequoia:
Edge
Nevada
Clark County, NV
According to the Albuquerque Tribune, the software
used to aggregate the vote totals on a Sequoia voting
system was overwhelmed by the volume of ballots
cast and omitted some votes from total reports.
Brennan Center for Justice | 75
Although a technician for the vendor fixed the
problem in Clark County and provided officials there
with a software patch, the same problem occurred a
few weeks later in Bernalillo County, New Mexico.
418
Premier
AccuVote-OS
New
Hampshire
Rockingham County, NH
The Eagle Tribune reported that “a single key on a
faulty computer keyboard” caused the election results
from optical scanners to be mis-entered by election
officials. Consequently, the initial election result was
erroneous.
419
Sequoia
AVC Advantage
New Jersey
Monmouth County, NJ
According to the Asbury Park Press, “results from a
half-dozen towns were delayed because of problems
in transmitting data from machine cartridges, County
Clerk M. Claire French said on Wednesday.” The
cartridges thus had to be physically taken to the
central election offices. "We're studying what caused
these problems," French said.
420
Sequoia
AVC Advantage
New Jersey
Essex County, NJ
The Star-Ledger reported that “[v]oters arriving at
Conklin Hall near the Rutgers-Newark campus were
thwarted this morning by two malfunctioning voting
machines. Voters said they were told the machine
would have to be replaced and to come back later.
“Carmen Cassiano, Essex County's commissioner of
registration, said about half of the machines
malfunctioned because they were turned on
incorrectly, which caused them to jam.
“They could not be reset, and had to be replaced,
Cassiano said. All the jammed machines were
replaced by mid-morning, but there were sporadic
problems throughout the day.”
421
Sequoia
AVC Advantage
New Jersey
Hudson County, NJ
According to the Jersey Journal, “the system of
counting votes at Hudson County Clerk’s office has
gone haywire.”
“According to County Clerk Barbara Netchert, the
way things were supposed to work is that local town
clerks were supposed to be able put the cartridges
from the electronic voting machines into a reader and
the information would be sent via computer to her
76 | Brennan Center for Justice
office.
“But the readers in several municipalities, including
Bayonne, Union City, Secaucus, and West New York
aren't working, Netchert said about 9 p.m.”
Results were therefore delayed by a few hours. It was
unclear what caused the problems.
422
Sequoia
AVC Advantage
New Jersey
Monmouth County, NJ
The Asbury Park Press reported that “a computer
being used by the municipal clerk malfunctioned.”
Local election officials could not fix the error, so the
cartridges were physically taken to the central
tabulation office. Results were therefore delayed by a
few hours. It was unclear what caused the
“malfunction.”
423
Sequoia
AVC Advantage
New Jersey
Eight New Jersey Counties
The Star-Ledger reported,, “[t]he numbers from the
cartridges that print out vote tallies and the paper-
tape backup within the machine didn't match” when
officials in Union County, New Jersey attempted to
obtain vote totals after the 2008 presidential primary.
According to the paper, several other counties,
including Bergen, Gloucester, Middlesex and Ocean,
uncovered similar problems when Union County
officials encouraged them to double-check their
results.
424
According to a report produced by a team of
Princeton researchers who conducted an independent
analysis of the voting equipment used in the primary,
the malfunction, which caused some voters to receive
the wrong party’s ballot, was triggered by human
error that would be “easy and natural” for many
pollworkers to make.
425
For more on this incident, see case study 13 on pages 22
- 24 of this report.
Sequoia
AVC Advantage
New Jersey
Essex County, NJ
The Montclair Times reported that “[i]n Essex
County, Superintendent of Elections Carmine
Casciano said that 12 out of about 650 Sequoia
Advantage electronic voting machines were replaced
on Tuesday. Of that amount, four machines were
replaced due to Board of Election workers mistakenly
Brennan Center for Justice | 77
turning off the machines and thereby deactivating
them.” The other eight had to be replaced due to a
“software malfunctionCasciano said ““[b]y
checking further into it, we have found no evidence
that we think triggered it. So we’re blaming it on the
system. A software malfunction.’”
426
Sequoia
AVC Advantage
New Jersey
Hudson County, NJ
The Star-Ledger reported that some election workers
failed to take cartridges from individual balloting
machines before election officials officially sealed the
machines. The election results were therefore
delayed.
427
Sequoia:
AVC Advantage
New Jersey
Passaic, Paterson, Scotch Plains, and North
Bergen Counties, NJ
WABC-TV New Jersey reported that the US
Attorney for New Jersey dispatched investigators to
address complaints of voting machines that were
preventing voters from casting ballots for the
Republican Senate candidate. Some reported that
their ballot was pre-voted in favor of the Democratic
incumbent, while others said the machine would not
register their selections.
428
Avante
Vote-Trakker
New Jersey
Warren County, NJ
According to the Express-Times, various electronic
voting machines stopped working in Warren County,
apparently because of an election official error
regarding the programming of the machine. 900
voters were forced to use emergency ballots.
429
ES&S
Model 100
New Mexico
Curry County, NM
According to the Clovis News Journal, “county
officials reported to the [Curry County Canvass]
board that memory cards — used to transfer tallies
from the ballot machines to a computer —
experienced corruption issues that rendered them
unusable.
“However, staff were able to bypass the problem
because those same voting machines generated paper
tape printouts, which they were able to manually
count and tally.”
430
ES&S
Model 100
New Mexico
Various Counties, NM
Various counties in New Mexico, the Associated Press
reported, experienced problems with the memory
78 | Brennan Center for Justice
cards used for the ES&S e-voting machines. The
“[p]roblem-prone memory cards used in voting
machines across New Mexico are being recalled to
make certain they're working properly in preparation
for the June primary election. . . .
“Secretary of State Mary Herrera ordered a recall of
the cards in an e-mail sent Thursday to the 33 county
clerks.”
431
Sequoia: Edge;
AVC Advantage
ES&S:
iVotronic
Danaher:
Shouptronic
New Mexico
Bernalillo County and other counties, NM
Throughout the state of New Mexico, significantly
high phantom vote rates and undervote rates led
voting analysts to question the outcome of the
elections in New Mexico. New Mexico led the
nation in the highest undervote rates. Also, certified
results showed a remarkably high number of
phantom votes.
432
Sequoia:
Edge
New Mexico
Bernalillo County, NM
According to the Albuquerque Tribune, local officials
discovered that about 12,000 early votes were not
included in the totals produced by the voting system
after Election Day. Apparently, the software program
used to aggregate vote totals was “overwhelmed by
the data.” Reportedly, the error was confined to early
voting machines, on which approximately 48,000
people cast their ballots.
The Tribune reported that the same problem had
occurred a few weeks earlier in Clark County,
Nevada. Apparently, the technician assigned to
address the problem in Clark County never notified
officials in Bernalillo County of the problem or
provided them with the software patch necessary to
fix it.
433
For more on this incident, see case study 9 on pages 17 -
18 of this report.
Sequoia
ImageCast
New York
Essex County, NY
Local television station WPTZ reported that the
Essex County commissioner complained of the
unreliability of the vote scanning machines. “Essex
County Elections Commissioner Lew Sanders said
big problems with the new scanning voter machines
for the disabled may translate into problems for all
voters come next election.
434
Brennan Center for Justice | 79
Sequoia
ImageCast
New York
Tompkins County, NY
According to the Ithaca Journal, a ballot-marking
machine at a polling place froze during the
instruction phase and could not be used for the rest
of the voting period.
“Democratic Tompkins County Board of Elections
Commissioner Stephen DeWitt said the problem lies
in a software glitch that causes the machine to freeze
up during the instructions segment. The machine at
Titus Towers seems to be the only one with that
problem, he said.”
435
ES&S
Model 100
North Carolina
Craven County, NC
According to the Sun Journal, local officials
discovered that at least 3,643 more votes were cast
than originally reported in unofficial election results.
“The additional votes surfaced when state elections
technician Rosemary Blizzard worked with the
Craven County elections director . . . to ‘read the
media cards back into the tabulation software,’
Blizzard said. . . .
“‘There was an error on a flash card at the main one-
stop sight. Partial results downloaded election night
from that site. I believe there were also some data
entry errors which unfortunately happen,’ Blizzard
said.
‘That's why the results are unofficial until after the
canvass.
Blizzard “said there will probably be at least 54 more
votes to add to results once the audit is complete.”
The results of the race were not affected by the
additional votes.
436
ES&S
Model 100
North Carolina
Forsyth County, NC
The Winston-Salem Journal reported that
“a mistake caused the software to omit results for the
74th N.C. House District in . . . two precincts.”
Officials eventually discovered the error and fixed it.
It was not clear what caused the “mistake.”
437
ES&S
Model 100
North Carolina
Rockingham County, NC
According to GoDanRiver.com, after votes had been
entered into the central tabulator, votes from only
one of the five precincts registered. It was unclear
80 | Brennan Center for Justice
what caused the problem.
438
ES&S
iVotronic
North Carolina
Polk County, NC
The Tryon Daily Bulletin reported that the County
Board of Elections left the public confused as the
results continued to flip-flop between candidates.
This was caused by a malfunction in the computer
reporting system. Eventually, local officials resolved
the problem and were able to announce a winner of
the race.
439
ES&S
Model 100
North Carolina
Onslow County, NC
The Jacksonville Daily News reported that “there were
a significant number of votes missing
approximately 4,000” after a “tabulator broke” on
election night. Rose Whitehurst, director of the
county board of elections, said ““[w]hen we started
looking over the report, we found that the votes
didn't all go into the software,’ . . . . ‘We started
doing an audit and ran through the information (and
found that) both one-stops weren't counted.’” No
further information regarding the missing votes was
provided.
440
ES&S
iVotronic –
Mecklenberg
Co.
Model 100 –
Wake Co.
North Carolina
Mecklenburg & Wake Counties, NC
Local news station, News 14, reported that due to a
“problem” with an optical scanner, certain votes in
the primary election were counted twice. The error
was corrected, and the errors did not change the
outcome of the races but the number of votes
involved was large. “For example, when Democratic
presidential candidate Barack Obama declared
victory Tuesday night in North Carolina’s primary,
he was given approximately 15,000 votes more than
he should have been. That’s 1 percent of all ballots
cast.
“In Wake County, early and absentee ballots were
counted as part of the individual voters precincts and
again as a whole. ‘So in essence, those voters were
recorded double,’ said Cherie Poucher of the Wake
Co. Board of Elections. It was around 37,000 votes.
It was unclear what caused the problem with the
scanner.
441
Sequoia
AVC Advantage
North Carolina
Buncombe County, NC
According to the Mountain XPress, touch-screen
Brennan Center for Justice | 81
voting machines in at least two precincts did not
display one of the races on the ballot. One election
official estimated that the error affected at least 600
voters. Because there was no paper record, it was
impossible to determine how many votes were lost.
442
Unilect
Patriot
North Carolina
Carteret County, NC
The Associated Press reported that about 4,400 votes
were lost in the election for state agricultural
commissioner. On election night, 3.3 million ballots
were cast and the Republican candidate led his
Democratic opponent by 2,287 votes. The touch-
screen DREs that had caused the problems did not
have a backup system, making the lost votes
irrecoverable. With almost twice as many votes
permanently erased as were needed to win the
election, a contentious legal battle ensued, and only
ended three months later when the contesting
candidate decided to concede the election.
443
Diebold (Model
unknown)
North Carolina
Gaston County, NC
According to local news media, 12,000 early votes
were unrecorded in the initial tally due to an
“interrupted download error” that occurred during
the transfer of election data to the tally server.
444
A later review by the Charlotte Observer found that
the number of voters who signed in at the polls
differed from the number of votes tallied in over 60
percent of the county’s precincts.
445
Fidlar AccuVote
2000 ES
North Carolina
Franklin County, North Carolina
According to the Associated Press, Fidlar optical
scanners recorded democratic straight-ticket votes as
straight-ticket votes for the Libertarian party.
Correcting the error changed the outcome of a
congressional race.
446
ES&S:
Optech III-P
North Carolina
Chatham County, NC
According to the News and Observer, machine
programming error caused all straight-ticket
Republican votes to go to the Libertarian candidate
in N.C. House District 54.
447
ES&S:
Optech III-P
Eagle
North Carolina
Wayne County, NC
The News and Observer reported that a programming
error caused machines to skip several thousand party-
line votes, both Republican and Democrat.
82 | Brennan Center for Justice
Correcting the error turned up 5,500 more votes and
reversed the outcome of the race for State House
District 11.
448
ES&S iVotronic
North Carolina
Jackson and Wake Counties, NC
Wired News reported that voting machine vendor
ES&S discovered a glitch in its voting system
firmware that caused the company’s touchscreen
machines to falsely sense that their memories were
full.
According to the magazine, officials in neighboring
Wake County discovered the same problem by
chance later that year. There, six touchscreen voting
machines lost a total of 436 ballots, and the voters
who cast those ballots had to be given the
opportunity to revote.
449
For more on this incident, see case study 10 on page 18
of this report.
ES&S:
M100
North Carolina
Robeson County, NC
According to VotersUnite!, ballot-tabulating
machines malfunctioned in 31 of 41 precincts during
a Senate primary. Local election officials stated that it
was due to a software glitch related to faulty memory
card programming and recounted the ballots.
450
Premier
AccuVote-TSX
Ohio
Montgomery County, OH
The Dayton Daily News reported that because of a
problem at a polling place in Trotwood where five
votes were lost on a machine, the county completed a
state ordered audit. During the audit, “[w]hen the
card was put back into its original machine, it re-
synced and showed the votes, a procedure that would
never occur during a typical count, [Elections
Director Steve] Harsman said. The manufacturer was
unable to determine the cause of this anomaly. The
audit also revealed additional problems with the
voting system; on three occasions, the paper record
failed to track when a user rejected a ballot to make
corrections, and the scanners failed to read some
clearly-marked ballots as valid.
451
ES&S
iVotronic
Ohio
Franklin County, OH
According to the Columbus Dispatch, some voters
reported that electronic touch-screen voting machines
were flipping votes. No further information was
Brennan Center for Justice | 83
provided.
452
ES&S
iVotronic
Ohio
Knox County, OH
The Columbus Dispatch reported that one touch-
screen electronic voting machine in Knox County
omitted Obama and McCain from the electronic
ballots. An election worker eventually solved the
problem by resetting the machine. It was not clear
what caused the issue.
453
Premier
AccuVote-TSX
Ohio
Stark County, OH
Some of the voting machines in Stark County,
according to the Canton Repository, experienced
difficulties. One of the machines “failed to print out
a paper ballot.” And voters had difficulties with
other machines. Jeanette Mullane, Stark County
Board of Elections director, said, “‘The machine issue
is more of a paper jam issue,’ . . . . ‘It's not that the
machines are down. I think it's a problem with the
set-up and our elections technicians are going out to
handle those calls.’
454
Premier
AccuVote-TSX
Ohio
Butler County, OH
In a letter to voting system manufacturer Premier,
Butler Count officials stated that they noticed votes
being dropped when they transferred totals from
individual machines to the system’s central tally
server after the March 2008 primary election.
455
The
Dayton Daily News reported that a subsequent
investigation by election officials revealed that at least
1000 ballots were undercounted in 44 counties.
456
Though Premier initially blamed the problem on
antivirus software installed on the machines, the
president of the company later conceded in a letter to
the Secretary of State that the server software
contained a “logic error” that caused the problem.
457
For more on this incident, see case study 1 on pages 10 -
11 of this report.
ES&S
iVotronic
Ohio
Franklin County, OH
According to the Columbus Dispatch, an e-voting
machine reportedly removed the name of a
Democratic candidate from some electronic ballots
and replaced it with a gray bar that said “candidate
withdrawn.” Secretary of State Jennifer Brunner was
one of the voters who received an erroneous ballot.
84 | Brennan Center for Justice
Investigators from SysTest Labs told Brunner,
“That's exactly what you'd see if someone masked a
name,” and noted that voters in other precincts
reported seeing a similar grey bar with the words
“candidate withdrawn.”
During its investigation of the incident, SysTest
investigators also found election officials had not
tested each voting machine to be used during the
election.
458
Premier
AccuVote-TSX
Ohio
Lucas County, OH
The Columbus Dispatch reported that “[t]here was a
report from one precinct in Lucas County where
ballot issues were not appearing on touch-screen
voting machines because of a problem with the
encoder for the devices.”
According to the paper, “[a] technician was
dispatched to the precinct to fix the problem…[and]
[o]fficials do not think anyone lost the chance to vote
as a result.”
459
ES&S:
iVotronic
Ohio
Mahoning County, OH
According to the Vindicator and the Roanoke Times,
sixteen of the 312 precincts experienced problems
with voting machines on Election Day, resulting in
delays when tabulating the results. The problems
were attributed to a number of issues, including
machine malfunctions, problems with cartridges, and
human error.
460
ES&S:
M550
Ohio
Sandusky County, OH
The News-Messenger reported that an election turnout
of 131% in one town indicated a tabulation error.
Officials concluded that some ballots, in nine
precincts, had been counted twice. They speculated
that some ballots had been fed through the machine
more than once.
461
Danaher/Guard
ian
ELECTronic
Ohio
Franklin County, OH
USA Today reported that an unexplained
malfunction caused the voting system to record
nearly 4,000 extra votes for George Bush when a
voting machine cartridge was plugged into a laptop
computer.
462
Premier
AccuVote-TSX
Pennsylvania
Northumberland County, PA
Brennan Center for Justice | 85
In Northumberland County, the Daily Item reported,
“poll workers discovered that voters who cast a
straight-party ticket could not see a summary of the
candidates they voted for. Mike Anderson, deputy
director of elections, said many of the county’s 94
precincts were notified immediately and told to
inform voters not to cast straight-party tickets.” This
apparently did not happen, and later that evening, a
judge ordered the machines impounded.
463
Danaher
Controls
Shouptronic
1242
Pennsylvania
Bucks County, PA
The Bucks County Courier Times reported that voting
machines in five municipalities malfunctioned. The
cause of the problems was unknown. County
spokeswoman Stacey Hajdak said, “I can't tell you
the exact problems of each machine until a technician
examines it.”
464
Danaher
Controls
Shouptronic
1242
Pennsylvania
Philadelphia County, PA
According to local television station KDKA, at one
polling location, an electronic voting machine
“wouldn’t work” and voters were forced to use paper
ballots. A poll worker speculated that the problems
were caused by a morning power outage in the
precinct: “‘We did have a problem with the
electricity for a minute. It must have gone off and on
– and maybe that triggered what the problem ended
up being.’”
465
ES&S:
iVotronic
Pennsylvania
Allegheny County, PA
The Pittsburgh Post-Gazette reported that twenty
voting machines were removed from polling stations
across Allegheny County due to technical glitches.
The faulty machines failed to “zero out,” or reset the
vote tally at zero before the voting started.
466
ES&S:
Optical Scan
(model
unknown)
Pennsylvania
Luzerne County, PA
According to the Times Leader, unofficial tallies
differed by nearly 6,000 votes from official ballot
counts. Some candidates’ vote totals were hundreds
of votes short of the initial count. Officials attributed
the tabulation errors to the voting machine vender,
who failed to reset a counter on a ballot scanner.
467
Danaher:
1242
Pennsylvania
Berks County, PA
The Reading Eagle reported that 111 votes were lost
during the primaries for the county election board.
The error occurred when the cartridges used to
86 | Brennan Center for Justice
record votes were accidentally programmed as
training cartridges. Election results showed that three
races were determined by less than 111 votes.
468
ES&S:
M550
Pennsylvania
Cumberland County, PA
A ballot programming error for straight-line ticket
votes gave the office of magisterial district judge to
the wrong candidate, according to the Sentinel.
Straight-ticket Democratic votes were given to the
Republican candidate. Straight-ticket Republican
votes were not counted at all. Initial vote totals
showed the Republican candidate had won by a
1,650 to 1,468 margin. Ultimately, however, the
Democratic candidate won by a two-vote margin
1,703 to 1,701 – in the recount.
469
ES&S
iVotronic
South Carolina
Charleston County, SC
According to local television station WSCS, local
officials experienced “an error with the machine that
automatically reads the electronic votes. The
commission instead decided to enter the information
into the system manually, a process that took more
than 6 hours.” It was unclear what caused the
error.
470
ES&S
iVotronic
South Carolina
Beaufort County, SC
Island Packet reported that the votes of some voters
for Bluffton Town Council did not appear on the
review screen of the electronic voting machines,
leading voters to believe that the votes were not
recorded correctly. Voters who recognized the
problem were able to cast their votes with paper
ballots.
471
ES&S
iVotronic
South Carolina
Florence and Horry Counties, SC
According to the Sun News, election officials
incorrectly programmed the voting machines to close
on the wrong date: “In both counties, the voting
machines were incorrectly set to close on Jan. 26, the
date of the Democratic presidential primary, instead
of Jan. 19, the date the Republican primary was
held.” Results were thus delayed as officials could
not access the data until technicians manually closed
each machine. During polling, voters in Horry
County also complained of malfunctioning
machines; according to the paper, supplies of
emergency paper ballots “were running out.
472
ES&S:
South Carolina
Charleston County, SC
Brennan Center for Justice | 87
iVotronic
The Post & Courier reported that iVotronic machines
failed to allow voters to review their choices in
contests in which they voted for more than one
candidate.
“If voters can vote for only one candidate, the review
screen shows who they voted for, but if they can vote
for two or more candidates, as is the case in a
Charleston County School Board race and the
Charleston County Soil and Water Commission race,
then the review screen indicates only whether they
have voted for the maximum number allowed. It
does not state for whom they voted.”
473
ES&S:
iVotronic
South Carolina
Kershaw County, SC
According to The State, initial vote totals in the
Republican and Democratic primary races for a
County Council seat, showed that 3,208 votes had
been cast in District 2. A manual count discovered
that only 768 votes had been cast. Election officials
suspected that the error had occurred because
machine cartridges were incorrectly programmed to
record some votes more than once. A state election
official apparently did not check a box that would
have prevented multiple readings.
474
ES&S
Model 650
South Dakota
Pennington County, SD
KOTA Radio reported that a “glitch” with the
electronic scanning machine caused additional votes
to be added to the total number of votes from the
county. The problem was discovered and corrected,
but the cause was unclear.
475
Premier
AccuVote-TSX
Tennessee
Shelby County, TN
The Memphis Commercial Appeal reported that the
Bartlett municipal election “ballot did not load onto
the voting machines” in Shelby County.
Consequently, some voters used paper ballots instead
while others simply did not vote in that race at all.
According to the paper, “Election Commission
officials said part of the problem was poll workers not
putting the proper code in for the Bartlett ballot in
precincts, such as Bartlett Elementary, where
residents and non-residents are registered.” Precincts
where all registrants were Bartlett residents also had
problems, which candidates said they understood to
88 | Brennan Center for Justice
be the result of a “computer glitch.
476
ES&S
iVotronic
Tennessee
Davidson County, TN
According to the Tennessean, a voting “machine
malfunctioned” in Davidson County, when it would
not let a voter see her ballot. Election officials
nevertheless told her that she had already voted, and
asked her to “move on.” At another site in the
county, at least one voter had trouble using the
touch-screen machine.
477
Hart InterCivic
eSlate
Tennessee
Knox County, TN
The Knoxville News Sentinel reported that an
“anomaly in the voting machines” caused only the
first three letters of the candidates’ first names to
appear on the review screen. This caused some voters
to be confused about for whom their votes were
recorded.
478
ES&S
iVotronic
Tennessee
Decatur County, TN
According to the Decatur County Chronicle, several
voters complained that voting machines “switched”
their votes in the presidential race, checking a
candidate for whom they did not intend to vote.
Voters told the paper that Election Commissioner
Rick Box told them that the problem could occur
when an individual touched the screen close to the
border of the button containing a candidate’s name.
“[Box] blames the problem in part on poor design by
software programmers, and adds that there may be
sensitivity issues with the screen itself. Election
Registrar Irene Campbell disagrees. She said there
was no problem with the machines, though at the
urging of state officials machine technicians were
called in late Monday afternoon to investigate the
problem.”
479
ES&S
iVotronic
Texas
Jefferson County, TX
According to local television station KBMT, Jefferson
County had some problems with electronic voting
machines flipping votes. “[County Clerk Carolyn]
Guidry says she has received less than a dozen
complaints about the machines. Guidry said it is
important for voters to make sure they are completely
within the lines on the screen when selecting a
candidate.”
480
Premier
Texas
Collin & Galveston Counties, TX
Brennan Center for Justice | 89
AccuVote TS
R6; Hart
InterCivic
eScan; Premier
AccuVote TSX
According to Wired, “[a] number of voters in several
Texas counties have been complaining that voting
machines they used to cast early votes flipped their
votes from Democratic choices to Republican ones”
when they tried to use the straight-party option on
the machines.
The problem was reported in at least seven counties
which use several different voting systems
manufactured by Hart InterCivic, ES&S, and
Premier. “A Hart InterCivic spokesman said that at
least one of the scenarios that a voter described isn’t
possible on the company’s machines. A spokesman
for ES&S said very few voters had complained and as
far as he knew poll workers were never able to
replicate the situation they described.” According to
the magazine, voters in jurisdictions West Virginia
and Tennessee that use the same ES&S machines
have also complained of vote flipping, though not
when using the straight-party function.
481
ES&S
iVotronic
Texas
Jefferson County, TX
According to CNN, voters in Jefferson County had
problems with the electronic voting machines
flipping votes. A man reported that his mother
“went to punch the selection for Obama and it
flipped to McCain.”
Jefferson County Clerk Carolyn Guidry told CNN
that she “received about a half a dozen calls about
touch screen machines that were not recording voters’
preferences accurately…. ‘We go out immediately
and re-calibrate machines that [polling workers] say
they are having problems with,’ Guidry said.”
Calibration, the process of identifying for the
machine the point on the screen that is being touch,
generally resolves the problem until repeated use of
the machine requires another re-calibration.
“But Guidry concedes that the problem . . . is
something that needs fixing. ‘I wish we could more
or less double the space between each box [with the
candidates’ names on the touch screen],’ said Guidry.
‘But the program is not designed to do that right
now.’ Guidry also told CNN that Jefferson County
has asked Election Systems & Software, the
manufacturer of the county’s machines, to look into
making that change in the next iteration of the
90 | Brennan Center for Justice
software.”
482
ES&S
iVotronic
Texas
Palo Pinto County, TX
According to the Mineral Wells Index, “[a]t least two
Palo Pinto County residents say they experienced
early voting problems when the touch-screen voting
machines they used kept switching their straight-
party vote from Democratic to Republican.
483
Hart InterCivic
eSlate
Texas
Cass County, TX
The Atlanta Citizens Journal reported that an “error”
allowed voters who did not live in District 1 to vote
for the District’s city council. City Manager Mike
Ahrens told the paper that the mistake was made by
the manufacturer, which failed to program the
machines to restrict voting in the District 1 election
to voters who live in the district. According to
Ahrens, the company admitted fault.
484
Hart InterCivic
eSlate
Texas
Travis County, TX
There were delays in voting at one polling location,
the Daily Texan reported, after “[a] technician
discovered that the machine responsible for printing
out access codes for voters was broken. Voting
resumed about 10 a.m. after the machine was shut
down and replaced with a new one.” Travis County
Clerk Dana DeBeauvoir told the paper that her office
had received reports of broken machines from other
precincts, as well.
485
ES&S
Model 100
Texas
Angelina County, TX
According to the Lufkin Daily News, a “programming
error” caused the central tabulator to be unable to
read the cards containing the count data. “‘Each
machine will only take 10 precincts,’ [Acting Tax
Assessor Thelma] Sherman said. ‘The precinct
numbers labeled on the cards did not match the
information inside the card, and the machines would
not accept them. We had to go back and start from
scratch, physically counting ballots at times.’”
486
Diebold:
TS-R6
Texas
El Paso County, TX
El Paso County Attorney Jose Rodriguez told KFOX-
TV News that 16 people complained that a vote cast
on their touch-screen ballot registered to the wrong
candidate. Five of the people called Rodriguez’s office
to complain. Eleven others called a local radio show
to complain.
487
Brennan Center for Justice | 91
ES&S:
iVotronic
Texas
Hidalgo County, TX
According to the Dallas Morning News, election
officials had to recount votes for U.S. House District
28 manually after discovering that the device used to
download election totals from its touch-screen
machine malfunctioned.
“The director of elections for the Texas Secretary of
State's office, Ann McGeehan, said votes would be
counted using a printout generated by each voting
machine.”
488
ES&S:
iVotronic;
M650
Texas
Bexar County, TX
The San Antonio Express News reported that the
tabulation of election results was delayed for one and
a half hours because the tabulation computers had
not been programmed with updated data in order to
count “mail-in” paper ballots. The computer system
was taken off-line and updated with the information
needed to process the 3,000 paper ballots, which
were then tabulated using high-speed scanners.
489
ES&S:
iVotronic
Texas
Jefferson County, TX
During early voting, voters complained that when
they selected a particular candidate, another
candidate’s name would light up. KDFM reported
that the vote switching occurred with voters who had
cast a straight Democratic ticket as well as individual
votes.
490
Hart:
eSlate; eScan
Texas
Tarrant County, TX
According to the Fort Worth Star-Telegram, a
programming error during the Tarrant County
primary election caused an extra 100,000 votes to be
recorded than had been cast. Initial tallies indicated
that 158,000 people had voted when actually only
approximately 58,000 had voted. The problems
stemmed from a programming error created by the
vendor, Hart InterCivic. The error caused the
computer to compound vote totals each time the
election totals were updated throughout the night,
rather than simply keeping a running total.
491
ES&S:
M315
Texas
Lubbock County, TX
According to the Lubbock Avalanche Journal,
machines failed to count votes in the race for Precinct
8 Democratic chairman. The ballots had to be
recounted with alternate software, provided by
92 | Brennan Center for Justice
ES&S.
492
ES&S:
M650
Texas
Scurry County, TX
An unanticipated landslide victory for two
Republican commissioner candidates caused poll
workers to question the results. According to the
Houston Chronicle, a chip in the ES&S M650
contained an incorrect ballot program. After ES&S
sent a new chip and the county officials counted the
votes by hand, the opposing Democratic candidates
actually won by a large margin.
493
Premier
AccuVote-TSX
Utah
Washington County, UT
According to the Spectrum, “Washington County
Clerk Cal Robison said a technical glitch slowed the
voting early in the morning….The machines were
fixed within a couple of hours.” No further
information was given.
494
Premier
AccuVote-OS
Virginia
Fairfax County, VA
The Washington Post reported that, in Fairfax
County, an electronic “voting machine broke down”
at the end of the election day while “officials were
tallying” votes.
495
According to the Reston Connection, the machine
totals initially showed 364 more ballots recorded than
were actually cast. Election officials had to run the
“ballot images” generated by the machine and hand-
count them in order to obtain accurate vote totals. It
was unclear what caused the problem.
496
For more on this incident, see case study 11 on pages 20
- 21 of this report.
Advanced
Voting Systems
WinVote
Virginia
Fairfax County, VA
According to CNN, “General Registrar Rokey
Suleman said a handful of memory cards did not
work when machines were started this morning ….
He said a ‘very, very small number of machines’ were
affected.” No further information was given.
497
AVS:
WinVote
Virginia
Fairfax County, VA
According to Connection Newspapers, there were
reports that if voters touched the screen around the
U.S. Senate box, the wrong candidate would light up.
Election officers told voters to make sure to use their
fingertips and to notify them of any malfunctions
they encountered. Fairfax’s general registrar called the
Brennan Center for Justice | 93
WinVote manufacturer to report the glitch and was
told it was a calibration issue and couldn’t be
immediately fixed.
498
AVS:
WinVote
Virginia
Fairfax County, VA
The Washington Post reported voter complaints that
the machines were failing to register their votes for
incumbent school board member Rita S. Thompson
(R). Local election officials told the paper that testing
later indicated that for every 100 votes cast for
Thompson, the machines subtracted approximately
one vote for her. According to voters, the machine
would initially display an “x” aside Thompson’s
name, but the “x” would disappear seconds later.
One voter said it took him about 4 or 5 attempts
before he successfully voted for Thompson. It was
impossible to determine whether lost votes were
intended for Thompson or whether other candidates
also lost votes.
499
Sequoia
Optech Insight
Washington
Pierce County, WA
The News Tribune reported that new vote tabulation
software ran so slowly in Pierce Co. that technicians
had to add memory to the computer system.
Election results were delayed.
500
Sequoia
AVC Edge
Washington
Snohomish County, WA
According to KING 5 News, voters in at least four
polling precincts reported that they experienced vote-
switching errors. They explained that the review
screen showed they had chosen the opposing
candidate. It took several attempts for each voter to
correct the mistakes. Review screens showed that the
correct candidates had been selected. Snohomish
County elections official Bob Terwilliger reported
that the problem only occurred in 15 out of 950
electronic voting machines throughout the county.
501
ES&S
iVotronic
West Virginia
Marion County, WV
The Times West Virginian reported that initially
“4,600 early votes could not be counted because of a
glitch in the PEB reading software.” An ES&S
technician was on hand but some were frustrated by
the technician’s performance: “‘I’m very frustrated
with the fact that we had an ES&S technician here
and he couldn’t get the machine to tally the early
votes,’ said Commission [sic] Randy Elliott.”
Eventually the problem was resolved, and election
94 | Brennan Center for Justice
officials managed to get the votes uploaded and
tallied.
502
ES&S
iVotronic
West Virginia
Marion County, WV
According to local television station WHSV, “[t]he
machines that scan the ballots quit tabulating votes
around 7:30 p.m. when clerks tried to feed in the
results of early voting. [County Commissioner
Wayne] Stutler says that caused a glitch in the
software and stopped all vote-counting.” The cause of
the problem was unclear.
503
ES&S
iVotronic
West Virginia
Jackson, Putnam, Ohio, Monongalia, and
Greenbrier Counties, WV
According to the Charleston Gazette, at least 14 voters
from these counties reported that the electronic
voting machines switched their votes from
Democratic to Republican candidates. In each case,
election officials helped the voters to correct the
ballots.
504
ES&S
Model 650
West Virginia
Berkeley County, WV
The Charleston Gazette reported that in Berkeley
County some voters experienced problemstheir
votes for Obama repeatedly to votes for McCain. In
response to voter complaints, the WV Secretary of
State advised county clerks to recalibrate machines if
necessary.
505
ES&S
iVotronic
West Virginia
Ohio County: WV
According to the Wheeling News Register, a touch-
screen voting machines flipped a voter’s ballot. The
voter tried multiple times to cast her vote but kept
running into the vote- flipping problem. She was,
however, eventually able to cast her vote for her
candidate of choice. The paper noted that similar
incidents had been reported in at least two other
West Virginia counties as well.
West Virginia Secretary of State Betty Ireland
identified calibration problems as the likely culprit,
saying that the machines can become miscalibrated
when jostled, but she also stated that touching near,
but not directly on, a candidate’s name “confuses the
machine.” Ireland ordered Ohio county to recalibrate
the machines in use during early voting; according to
the News Register, other counties in the state also took
this precaution.
506
Brennan Center for Justice | 95
ES&S
iVotronic
West Virginia
Jackson County, WV
According to the Charleston Gazette, voters in Jackson
County complained of their votes being flipped from
Democratic to Republican candidates. In response,
election officials were directed by the Deputy
Secretary of State to recalibrate the voting
machines.
507
ES&S
iVotronic
West Virginia
Putnam County, WV
The Charleston Gazette reported that several voters in
Putnam County complained of their votes being
flipped from Democratic to Republican candidates.
508
ES&S
iVotronic
West Virginia
Wayne County, WV
According to the Huntington Herald-Dispatch, eight
to ten voting machines in Wayne County
malfunctioned during the primary election. Wayne
County Clerk Bob Pasley told the paper that the
problem was with the printers attached to the
machines that are intended to create a paper record of
voters’ actions:
“Pasley said several of the printers jammed, causing
certain votes to be counted and digitally saved, but
not displayed on the paper receipt.
Pasley told the paper that voters also had trouble
selecting their intended candidate on touch screen
machines in use in the county due to calibration
issues.
509
Premier Accu-
Vote ES 2000
Wisconsin
Walworth County, WI
According to the Janesville Gazette, memory card
problems caused delays in election results in the
village of Darien. Walworth County Clerk Kim
Bushey told the paper that the card failed
midmorning and needed to be ‘redownloaded’ for
voting to continue. “‘Any votes cast while the card
was down were placed in a locked box in the voting
machine,’ Bushey said.”
A second memory card malfunction prevented poll
workers from submitting results electronically.
Bushey asked village election officials to fax her the
results, which she then entered into the county totals
manually.
510
ES&S:
Wisconsin
Taylor County, WI
96 | Brennan Center for Justice
M150
According to the Marshfield News Herald, four and a
half months after the election, a consulting firm
discovered that the optical scanners were
programmed incorrectly. All straight-party votes
were lost, affecting approximately 600 ballots.
511
ES&S:
Optech III-P
Eagle
Wyoming
Natrona County, WY
According to the Casper Star Tribune, the Unity
Election Management System, used to tally votes
from both optical scan machines and paperless
electronic voting machines, failed to tally votes
correctly in several municipal races. Noticing that the
ballot totals in the city of Evansville seemed low,
election officials checked the printouts from the
precinct voting machines. They found that the totals
didn't match the totals computed by the Unity
software. Election officials conducted a recount
using printouts from the voting machines and the
paper absentee ballots.
512
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
Brennan Center for Justice | 97
  :    *
*Sections of this document have been deleted by the Brennan Center to include only those paragraphs relevant to this report.
A complete copy of the document is on file with the Brennan Center.
98 | Brennan Center for Justice
endnotes
1. Definitions for key terms such as ‘voting systems’ used in the report can be found in Appendix A.
2. Telephone Interview with Denise Lamb, Chief Deputy Clerk for Elections, Santa Fe County, New Mexico (Sept. 30,
2009).
3. United States Election Assistance Commission, Election Administration and Voting Survey Frequently Asked Questions,
http://www.eac.gov/research/election_administration_and_voting_survey_faqs.aspx (last visited Aug. 9, 2010);
Spencer Overton, Stealing Democracy, http://www.stealingdemocracy.com/facts.cfm (last visited Aug. 9, 2010).
4. e Proposed Final Judgment and Competitive Impact Statement are reprinted in 75 Fed. Reg. 12256-12270 (Mar.
15, 2010).
5. S.B. 1404, 2009-10 Reg. Sess. (Cal. 2010).
6. Specifically, S.B. 223 2005-06 Reg. Sess. (N.C. 2005) states that “[t]he vendor shall promptly notify the State Board
of Elections and the county board of elections of any county using its voting system of any decertification of the
same system in any state, of any defect in the same system known to have occurred anywhere, and of any relevant
defect known to have occurred in similar systems.
7. See Michael Traugott, et. al., e Impact of Voting Systems on Residual Votes, Incomplete Ballots, and Other Mea-
sures of Voting Behavior (conference paper presented at the Midwest Political Science Association, Chicago, IL, Apr.
7-10, 2005), Charles Stewart III, Residual Vote in the 2004 Election (Caltech/MIT Voting Technology Project, VTP
Working Paper No. 2.3, 2005).
8. 42 U.S.C. §§ 15322, 15371(a) (2009).
9. 42 U.S.C. § 15322 (2009).
10. Sarah F. Liebschutz and Daniel J. Palazzolo, HAVA and the States, 35 P: J.  F  (), available
at http://publius.oxfordjournals.org/cgi/reprint/35/4/497.pdf; Press Release, Representative Rush Holt, Holt Urges
Election Assistance Commission to Address Voting Machine Failures, Accessibility in 2010 Election Survey (Nov. 19,
2009), available at http://www.house.gov/apps/list/press/nj12_holt/111909.html; VU, P  
EAC: A P I-E M R F V S (May 2008), available at
http://www.votersunite.org/info/EACInfoClearinghouseProposal.pdf.
11. E-mail from Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, to Lawrence Norden, Senior Counsel, Brennan Center for Justice (Sept. 10, 2010, 15:01 EST) (on file
with the Brennan Center).
12. Testimony of Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, available at http://www.eac.gov/News/docs/04-08-10-eac-public-mtg-testimony-layson-final.pdf/at-
tachment_download/file.
13. Testimony of Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, available at http://www.eac.gov/assets/1/AssetManager/05-27-10%20EAC%20Public%20Mtg%20
Testimony%20Layson%20FINAL.pdf.
14. E-mail from Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, to Lawrence Norden, Senior Counsel, Brennan Center for Justice (May 19, 2010, 13:58 EST) (on file
with the Brennan Center).
15. U S G A O, R   C, C  H
A, H  R: F P  C V S N  B
F D, F I,  E 4-5, 31 (September 2008), available at http://www.gao.
gov/new.items/d08814.pdf?source-ra [hereinafter GAO Voting Systems Report]. (“e scope of EAC’s efforts to track
Brennan Center for Justice | 99
and resolve problems with certified voting systems does not extend to those systems that were either qualified by
NASED or were not endorsed by any national authority. According to program officials, the commission does not
have the authority or the resources needed to undertake such a responsibility … As a result, the commissions efforts
to track and resolve problems with voting systems do not include most of the voting systems that will be used in the
2008 elections.”).
16. United States Election Assistance Commission, Testing and Certification Program Manual Version 1.0 (2007)
[hereinafter VSTCPM].
17. E-mail from Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, to Lawrence Norden, Senior Counsel, Brennan Center for Justice (May 14, 2010, 17:09 EST) (on file
with the Brennan Center).
18. Id.
19. VSTCPM, supra note 16 at 2.3.2.7.
20. Telephone Interview with Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election
Assistance Commission (Aug. 25, 2010).
21. VSTCPM, supra note 16 at 8.7.3.
22. U S E A C, V S T  C P V-
 S T A I F/S  EAC C ESS U ...
S (2010), http://www.eac.gov/assets/1/AssetManager/Product_Advisory-ES&S-06.25.10%20FINAL.pdf.
23. Id.
24. U S E A C, V S T  C P
V S T A C M I  EAC 
MV EMS 4.0B (M) (2010), http://www.eac.gov/assets/1/AssetManager/
MicroVoteEMS40B_Product_Advisory-final_august_23_2010.pdf.
25. E-mail from Matt Masterson Deputy Director of the EAC’s Testing and Certification Program to Lawrence Norden,
Senior Counsel, Brennan Center for Justice (July 1, 2010, 11:20 EST and Aug. 26, 2010, 15:32 EST) (on file
with the Brennan Center) (Confirming that two counties in Ohio, two counties and twenty-two towns, cities, and
villages in Wisconsin, twenty-two counties in Florida, and nine counties in Iowa, use certified equipment); E-mail
from Matt Masterson Deputy Director of the EAC’s Testing and Certification Program to Lawrence Norden, Senior
Counsel, Brennan Center for Justice (Aug 31, 2010, 11:16 EST) (on file with the Brennan Center) (Confirming
that Delaware uses EAC certified central count machines to count absentee ballots. Its polling place machines are
not EAC certified).
26. Counties may “upgrade” existing systems to versions that will be EAC certified. But for the most part, given the
current economic climate, counties and states around the country are unlikely to receive big grants to purchase
entirely new equipment.
27. U S E A C, S R   F V S
T  C P, 5 (2009), http://www.eac.gov/assets/1/Page/State%20
Requirements%20and%20the%20Federal%20Voting%20System%20Testing%20and%20Certification%20
Program.pdf.
28. VSTCPM, supra note 16 at 8.7.4.
29. Letter from Douglas A. Kellner, Co-Chair of the New York State Board of Elections to the Honorable omas R.
Wilkey, Director of the EAC (June 30, 2010).
100 | Brennan Center for Justice
30.
Compare County of San Diego Registrar of Voters Contract No. 46619 between County of San Diego and Diebold
Election Systems, Inc. and Diebold Incorporated at 20-21 (2003), available at http://accurate-voting.org/contracts/
CA/San_Diego/CA_sandiego_2003.pdf (demonstrating an contractual obligation for the county to inform the ven-
dor of defects in the voting system with no similar obligation on the part of the vendor), with Contract No. 08455,
Voting Equipment Agreement between Election Systems and Software, Inc. and Kansas Secretary of State at 7 (Nov.
16, 2005) (stating that the contractor will notify the customer of any defects or problems that arise).
31. E-mail from Jane Platten, Director, Cuyahoga County Board of Elections, to Lawrence Norden, May 27, 2010.
32.
For examples of a lag in vendor acknowledgement of voting system problems, see the case studies in this report from
Butler County, Ohio (at pp. 10 - 11) and Humboldt County, California (at pp. 12 - 13).
33. By usability concerns we mean flaws in the machines programming, software or hardware that make poll worker or
voter error significantly more likely, and which lead to significant disenfranchisement.
34. is appears to be precisely what occurred in Humboldt County, California in 2008. is case is detailed on pages
12 - 13.
35. E-mail from Betty McGary, Executive Director, Butler County Board of Elections, to Lawrence Norden, Senior
Counsel, Brennan Center for Justice (May 17, 2010 11:58 EST) (on file with the Brennan Center); see also Letter
from Betty McGary, Executive Director, and Lynn Kinkaid, Deputy Director, Butler County Board of Elections,
to Dave Byrd, President, Premier Election Solutions (Apr. 4, 2008) (on file with the Brennan Center) [hereinafter
McGary Letter 1].
36. Lynn Hulsey, Voting Machine Maker Says Error May Cause Votes to be Uncounted, D D N, Aug. 23,
2008, at A6.
37. Editorial, ats a Pretty Big Glitch, N.Y. T, Oct. 9, 2008, at 36. e paper refers to the manufacturer in this
incident as Diebold.
38. McGary Letter 1, supra note 35.
39. Letter from Betty McGary, Executive Director, and Lynn Kinkaid, Deputy Director, Butler County Board of Elections,
to Dave Byrd, President, Premier Election Solutions (Apr. 9, 2008) (on file with the Brennan Center).
40. See P E S, R  B C M  P E I (May
2008).
41. E-mail from Betty McGary to Lawrence Norden, supra note 35.
42. Office of the Ohio Sec’y of State, Premier Memo (Aug. 21, 2008) (on file with the Brennan Center).
43. Id.
44. Id.; Letter from Dave Byrd, President, Premier Election Solutions, to Jennifer Brunner, Ohio Sec’y of State (Aug. 19,
2008) (on file with the Brennan Center).
45. Id.
46. E-mail from Betty McGary to Lawrence Norden, supra note 35.
47. Telephone Interview with Carolyn Crnich, Registrar of Voters, Humboldt County, California (Sept. 29, 2009)
[hereinafter Crnich Interview].
48. addeus Greenson, Software Glitch Yields Inaccurate Election Results, T T-S, Dec. 12, 2008, http://
www.times-standard.com/localnews/ci_11145349; Kim Zetter, Serious Error in Diebold Voting Software Caused Lost
Ballots in California County, T L (Blog of W M), Dec. 8, 2008, http://www.wired.com/
threatlevel/2008/12/unique-election [hereinafter Zetter 2008].
Brennan Center for Justice | 101
49. E-mail from Carolyn Crnich, Clerk, Humboldt County, California, to Lawrence Norden, Senior Counsel, Bren-
nan Center for Justice at New York University School of Law (May 1, 2010, 12:00 EST) (on file with the Brennan
Center).
50. Id.
51. Crnich Inteview, supra note 47.
52. Zetter 2008, supra note 48; Jaikumar Vijayan, California Finds E-Voting Software Had Errors, Data Deletion Func-
tions, C, Mar. 4, 2009.
53. D B, C S’  S, R   E A C C E-
  D  D/P GEMS V .. 2, 4 (2009).
54. Errors and Deficiencies in Diebold/Premier GEMS version 1.18.19, Hearing before California Secretary of State Panel
19 (Mar. 17, 2009) (statement of Justin Bales, General Manager for Western States, Premier Elections Solutions)
[hereinafter Bales Testimony].
55. Crnich Interview, supra note 47.
56. Bales Testimony, supra note 54, at 20.
57. Crnich Interview, supra note 47.
58. See F D E, A R  O U   G
E, General Over Undervote Table, available at http://election.dos.state.fl.us/reports/index.shtml.
59. Mary K. Garber, Voting System Performance Problems – Conclusions and Recommendations, in Assessing Election
Accuracy in Florida’s 2006 General Election: A Comparison of Voting System Performance Using Undervote Rates in State-
wide Races in Florida’s 2006 General Election (Dec. 2008) (unpublished manuscript, on file with the Brennan Center)
[hereinafter Garber manuscript].
60. Id.
61. Id.
62. Kim Zetter, Lost E-Votes Could Flip Napa Race, W, Mar. 15, 2004, available at http://www.wired.com/politics/
security/news/2004/03/62655.
63. Id.
64. Id.
65. Garber manuscript, supra note 59.
66. Id.
67. E-mail from Mary K. “Kitty” Garber, Associate Director, Florida Fair Elections Center, to Lawrence Norden, Senior
Counsel, Brennan Center for Justice at NYU School of Law (Apr. 12, 1010, 15:17 EST) (on file with the Brennan
Center).
68. E-mail from Bill Cowles, Supervisor of Elections, Orange County, Florida, to Laura Seago, Research Associates,
Brennan Center for Justice at NYU School of Law (May 5, 2010, 9:34 EST) (on file with the Brennan Center).
69. L N, . ., I A R  V S P  V M P
  (Brennan Center 2008), available at http://www.brennancenter.org/content/resource/is_america_ready_to_
vote [hereinafter Is America Ready to Vote].
102 | Brennan Center for Justice
70. Daniel Nasaw, Voting Machine Claim to be Studied by County, A D-G, May 20, 2006 [here-
inafter Nasaw].
71. Nasaw, supra note 70; Height Could Make a Difference in this Year’s Elections (THV Broadcast, May 20, 2006)
(transcript available at http://www.todaysthv.com/news/news.aspx?storyid=28836) [hereinafter THV];
72. THV, supra note 71.
73. See M MD, C F, C O,  K F, N C 
H S, N H S R N. , A R D 
C  A: U S, – 16 (2008).
74. Nasaw, supra note 70.
75. See, e.g., Sarah Everett, e Usability of Electronic Voting Machines and How Votes Can Be Changed Without De-
tection (May 2007) (Unpublished PhD dissertation, Rice University) available at http://chil.rice.edu/research/pdf/
EverettDissertation.pdf, Bryan Campbell and Michael Bryne, Do Voters Notice Review Screen Anomalies? A Look
at Voting System Usability (2009) (paper presented at 2009 Electronic Voting Technology Workshop/Workshop on
Trustworthy Elections) available at http://www.usenix.org/event/evtwote09/tech/full_papers/campbell.pdf.
76. Nasaw, supra note 70.
77. Id.
78. M.C. Moewe, Voting Machine Maker Admits Problem, D B N-J, Nov. 3, 2007, at 1A. In the
article, the paper refers to the manufacturer as Diebold.
79. Id.
80. Id.
81. Id.
82. Id.
83. Id.
84. Id.
85. Id.
86. Id.
87. Id.
88. M.C. Moewe, Info on Voting Flaws Not Shared, D B N-J, Nov. 12, 2007, at 1A.
89. Id.
90. Id.
91. Jeremy Milarsky and Rafael A. Olmeda, Software Tweak Restores Votes, S-S, Nov. 5, 2005, at 23A.
92. Eliot Kleinberg, Broward Machines Count Backward, P B P, Nov. 5, 2004, at 29A [hereinafter Klein-
berg]; see also Mark Leon Goldberg, Don’t Count on It: Why We Need Paper Trails to Back Up Compromised and Fal-
lible Voting Machines – and Why We’re Not Getting em, T A P, Jan. 2005, at A20.
93. Milarsky and Olmeda, supra note 91.
94. Milarsky and Olmeda, supra note 91, Kleinberg, supra note 92.
Brennan Center for Justice | 103
95. Kleinberg, supra note 92.
96. Mary Ellen Klas and Gary Fineout, Touchscreen Voting Machines Have Software Flaw, Florida Officials Learn, M
H, Jun. 12, 2004, at 1B [hereinafter Klas and Feinout].
97. Mary Ellen Klas, Flaw in Florida Touch-Screen Voting Machines Was Known in 2002, M H, Jun. 12, 2004,
at 1B.
98. Klas and Fineout, supra note 96.
99. Id.
100. Id.
101. Klas, supra note 97.
102. Helen Gao, Faulty Switches Blamed for Voting Woes, S D U T, Apr. 14, 2004, at B-3 [hereinafter
Gao].
103. Luis Monteagudo Jr. and Helen Gao, Glitches in Voting Machines Examined, S D U T, Mar. 4,
2004, at B-1 [hereinafter Monteagudo and Gao].
104. Id.
105. Id.
106. Gao, supra note 102.
107. Bill Ainsworth, Tech Says Diebold Knew of Problem, S D U-T, Apr. 22, 2004, at B-1 [hereinafter
Ainsworth]; Hearing Before the Cal. Sec’y of State Voting Systems and Procedures Panel (Apr. 21, 2004) (statement
of James Dunn) [hereinafter Dunn Testimony]; Touch-Screen Trouble, ABC News, Oct. 27, 2004, available at http://
abcnews.go.com/Politics/Vote2004/story?id=203866&page=1.
108. Dunn Testimony, supra note 107, at 73.
109. Id. at 77-78.
110. Ainsworth, supra note 107; Hearing Before the Cal. Sec’y of State Voting Systems and Procedures Panel (Apr. 21,
2004) (statement of Robert Urosevich) at 59.
111. Frank Zoretich, Election Results Certified After Software Blamed, A T, Nov. 19, 2002 (on file with
the Brennan Center) [hereinafter Zoretich].
112. Dan McKay, County Certifies Vote Tally, A J, Nov. 19, 2002, at D1; Zoretich, supra note
106.
113. Zoretich, supra note 111.
114. Dan McKay, Voting Machines Criticized, A J, Sept. 26, 2004, at A1.
115. Supplemental Brief for the petitioner at 10 (citing James A. Noel Aff. ¶¶ 3-8) Lopategui v. Vigil-Giron ex. rel., No.
CV 2005-0433 (N.M. 2d Dist. Ct. Dec. 20, 2005).
116. Id.
117. Id.
118. Zoretich, supra note 111.
104 | Brennan Center for Justice
119. Telephone Interview with Denise Lamb, Chief Deputy Clerk for Elections, Santa Fe County, New Mexico (Sept. 30,
2009).
120. Kim Zetter, E-Vote Machines Drop More Ballots, W, Feb. 9, 2004, available at http://www.wired.com/politics/
security/news/2004/02/62206 [hereinafter Zetter 2004].
121. Id.
122. Telephone Interview with Cherie Poucher, Director, Wake County Board of Elections, in Raleigh, N.C. (Oct. 5,
2009) [hereinafter Cherie Poucher interview].
123. Zetter 2004, supra note 120.
124. Id.
125. Is America Ready to Vote, supra note 69.
126. Cherie Poucher interview, supra note 122.
127. Zetter 2004, supra note 120. According to Poucher, all but 78 recast their votes.
128. Zetter 2004, supra note 120.
129. 42 U.S.C. § 15481(a)(3)(A) (2010).
130. N R, C   E V E   S E II (Verified Voting
Foundation 2009), available at https://www.verifiedvotingfoundation.org/article.php?id=6717; See, e.g., N R-
  J T, A R R  C, T--B V S R
(Office of the California Secretary of State 2007), available at http://www.sos.ca.gov/elections/voting_systems/ttbr/
accessibility_review_report_california_ttb_absolute_final_version16.pdf.
131. R, supra note 130.
132. See, e.g., R  T, supra note 130.
133. In 2008 alone, there were 480 complaints characterized as “accessibility problems” logged on the Election Protection
866-OUR-VOTE system.
134. Cherie Poucher interview, supra note 122.
135. Id.
136. We suggest triggers for mandatory voter reporting at p. 32.
137. Julia O’Donoghue, Big Questions about Voting Machines, R C, Mar. 18, 2009, available at http://
www.connectionnewspapers.com/article.asp?article=326929&paper=71&cat=104.
138. See generally Is America Ready to Vote, supra note 69.
139. Telephone Interview with July Flaig, Election Manager, Fairfax County, Virginia, Office of Elections
(April 13, 2010).
140. Id.
141. Id.
142. Id.
Brennan Center for Justice | 105
143. Telephone Interview with Rokey Suleman, Executive Director, Washington, DC Board of Elections and Ethics (Apr.
1, 2010).
144. Id.
145. M M. C, P M,  H T, C   D  C, R-
 C   D’ E: P R  R   C
B  E  E I S C 1 (Oct. 8, 2008) [hereinafter D.C. Council
Report].
146. Id.
147. Nikita Stewart and Elissa Silverman, Primary Vote Still Doesn’t Add Up, W. P, Sept. 22, 2008 at B1; Nikita
Stewart, District’s Primary Results Certified, W. P, Sept. 26, 2008 at B1.
148. Id.
149. D.C. Council Report, supra note 145, at 4; S V S, R   D  C
B  E  E 1 (Sept. 22, 2008).
150. S V S, supra note 149.
151. Id. at 3 (emphasis added).
152. Id. at 1 (Sept. 22, 2008).
153. Committee Protective Order between Board of Elections and Ethics Investigation Special Committee of the District
of Columbia Council and Sequoia Voting Systems, Inc. (Jun. 5, 2009), available at http://www.voteraction.org/files/
Exhibit%20I%20-%20DC%20protective%20order.pdf [hereinafter D.C. protective order].
154. Nikita Stewart, Cheh to Seek Court Order in Elections Code, W. P, Apr. 23, 2009.
155. Letter from Mary Ellen B. Offer to Mary M. Cheh, Re: Subpoena Duces Tecum to Sequoia Voting Systems, Inc., (Sept.
26, 2008) (“[T]here are no documents that are responsive to this request with regard to the voting system used by
the DCBOEE - a combination of the Eagle III-P precinct-based optical scan units, Edge I non-VVPAT touchscreen
units and WinEDS 3.1.012.”).
156. D.C. protective order, supra note 153; Tim Craig, Firm to Give D.C. Information about Its Voting Devices, W.
P, Jun. 6, 2009.
157. Stewart, supra note 154.
158. Id.
159. D.C. protective order, supra note 153; Craig, supra note 156.
160. D.C. protective order, supra note 153; Craig, supra note 156.
161. City Council R. 18A18-0238, 2009, Council Period 18 (D.C. 2009), 2009 D.C. Law L18-0103.
162. E-mail from David Zvenyach, Chief of Staff, Office of Councilmember Mary Cheh, to Lawrence Norden, Senior
Counsel, Brennan Center for Justice (Apr. 12, 2010, 18:17 EST) (on file with the Brennan Center).
163. Andrew W. Appel, et. al., Princeton University, I  I   S AVC A
9.00H DRE V M 117 (October 17, 2008) [hereinafter Princeton Study].
106 | Brennan Center for Justice
164. E-mail from Penny Venetis, Co-Director, Rutgers Constitutional Litigation Clinic, to Lawrence Norden, Senior
Counsel, Brennan Center for Justice (Apr. 8, 2010, 12:38 EST) (on file with the Brennan Center) [hereinafter Vene-
tis E-mail].
165. Id.
166. Robert Stern, Makers Defend Voting Booths; Tally Errors Blamed on Polling Workers, T  T, Mar. 1, 2008,
at A1.
167. Id.
168. Joe McIntyre, Senior Project/Account Manager, Sequoia, WinEDS Technical Product Bulletin – AVC Advantage Party
Turnout Issue/Operator Panel Usage March 4, 2008.
169. Posting of Ed Felten to Freedom to Tinker, Interesting E-mail from Sequoia, http://freedom-to-tinker.com/blog/
felten/interesting-email-sequoia (Mar. 17, 2008, 14:25 EST).
170. Carly Rothman, Plan for Voting Machine Probe Dropped After Lawsuit reat, S-L, Mar. 18, 2008, available
at http://www.nj.com/news/index.ssf/2008/03/voting_machine_maker_threatens.html.
171. Princeton study, supra note 163 at 113-114.
172. Id. at 1-2.
173. Joshua Brockman, N.J. Voting Machines May Be Tested for Accuracy, N’ P R, http://www.npr.org/tem-
plates/story/story.php?storyId=90727541.
174. Venetis E-mail, supra note 164; Joe Ryan, Judge Rules Public Can See Voting Machine Test Results, S-L, Jun.
20, 2008, http://www.nj.com/news/index.ssf/2008/06/judge_lifts_gag_order_on_votin.html.
175. Princeton study, supra note 163 at 117-118.
176. Id. at 118.
177. Venetis E-mail, supra note 164.
178. Princeton study, supra note 163 at 8-10.
179. Derrick Nunnally, Voter Interest Surges in Pennsylvania Suburbs, P I, Mar. 6, 2008, http://www.
philly.com/philly/news/politics/suburban_pa/16327086.html?viewAll=y.
180. IC § 3-11-7.5-4(d); Rick Yencer, Election Software Not Certified, S P, Apr. 20, 2006, at 3A [hereinafter Yencer
1]; Cindy Larson, Uncertified Voting Machines to Be Used, N-S, Apr. 21, 2006 [hereinafter Larson]; Robert
Annis, Some Voting Results Questioned, I S, Jan. 8, 2009, at 4Y [hereinafter Annis].
181. Yencer 1, supra note 180; Larson, supra note 180.
182. MicroVote General Corp, Admin Cause No. 06-003-ED (eResolution May 21, 2007), www.in.gov/sos/elections/
files/SOS_MicroVote_Order.pdf at 21 [hereinafter Indiana administrative order]; Rick Yencer, Voting Machines to be
Repaired, S P, Sept. 26, 2006, at 3A [hereinafter Yencer 2].
183. Indiana Administrative order, supra note 182, at 21; Yencer 2, supra note 182; Jason omas, Poll Machine Flaw
Hidden, State Says, I S, Sept. 25, 2006, at 1 [hereinafter omas].
184. omas, supra note 183.
185. Indiana administrative order, supra note 182, at 39.
186. Id. at 38.
Brennan Center for Justice | 107
187. Id. at 22.
188. See omas, supra note 183.
189. omas, supra note 183; Yencer 2, supra note 182.
190. Annis, supra note 180.
191. N J D   P A, T P  V S  N J: H
G C B P T R  V S 3 (2009), available at http://www.
state.nj.us/publicadvocate/public/pdf/e_Purchase_of_Voting_Systems_in_NJ_11-24-09.pdf [hereinafter New
Jersey Public Advocate Report].
192. Id. at 5; Online Policy Group v. Diebold, 337 F. Supp. 2d 1195 (Cal. Dis. Ct. 2004); Cory Doctorow, Sequoia
Voting Systems reatens Felten’s Princeton Security Research Team, B, Mar. 17, 2008, http://boingboing.
net/2008/03/17/sequoia-voting-syste.html.
193. For a detailed discussion of how this kind of system could work, see pages 27 - 38 of this report.
194. Editorial, e Voters Will Pay, N.Y. T, Feb. 25, 2010, at A26 [hereinafter Voters Will Pay].
195. Id.
196. Kim Zetter, Feds Move to Break Voting-Machine Monopoly, W, Mar. 8, 2010, available at http://www.wired.com/
threatlevel/2010/03/ess-sued-in-antitrust-cas.
197. Press Release, Dominion Voting Systems, Dominion Voting Systems, Inc. Acquires Premier Election Solutions As-
sets from ES&S (May 19, 2010), available at http://www.dominionvoting.com/images/pdfs/DominionAcquiresPre-
mierReleaseFinal4.pdf.
198. See Voters Will Pay, supra note 194.
199. See Help America Vote Act of 2002, Pub. L. No. 107-252, 116 Stat. 1666 (codified at 42 U.S.C. §§ 15301–15545
(2008)).
200. ACCURATE, the multi-institution voting research center funded by the National Science Foundation (NSF), sug-
gests a fifth provision that would greatly improve regulation of voting systems. Specifically, that “there should be a
rich feedback loop – from problems to investigation to testing – that uses actual problems to inform future testing
procedures.” D M  J L H, P A  E-V P
H N  H S  T, NRC W 23 (2004). ey note that cur-
rent voting system guidelines “lack a process to incorporate suspected system failures or to address changing technol-
ogy. In particular, [they] fail to establish standards that ensure performance data from the used to improve systems
so that the same problems do not contaminate future elections. Problems need to be investigated, understood, and
then fedback into the process of recertifying (at times recalling) existing systems and establishing the next set of
[certification criteria].” E B, C W, J L H  D K. M, P
C    V V S G 30 (2005), available at http://accurate-voting.org/
accurate/docs/2005_vvsg_comment.pdf.
201. ere is precedent for creation of such FOIA exemptions. e Homeland Security Act (6 USC § 133) granted an
exemption for voluntarily submitted critical infrastructure information. Federal whistleblower protections similarly
work to keep individuals who report problems from suffering reprisals. e Whistleblower Protection Act of 1989
states that the Special Counsel tasked with investigating allegations made under the act may not disclose the identity
of the individual who filed the complaint without that persons consent except in cases where imminent danger to
public health or safety makes such disclosure necessary. 5 U.S.C. § 1213(h). See generally 5 U.S.C. § 552(b)(3),
which authorizes legislative carve-outs of FOIA, so long as such carve-outs are specific enough and leave no agency
discretion.
108 | Brennan Center for Justice
202. For voters who may not know the specific make and model of the machine involved, a simple description of the
type of machine will suffice as the EAC will be able to determine a machines make and model from the other data
submitted by the voter in the report.
203. New legislation should also authorize the relevant agency to seek penalties against voting machine vendors who take
any retaliatory action against election officials who post to the site.
204. is type of screening occurs with similar reporting databases. For example, some states screen reports for inclusion
in their highway accident databases. See, e.g., A D  T  P F,
 A T A (May 2003), available at http://www.dot.state.ak.us/stwdplng/highwaydata/pub/
accidents/2001aktraffix.pdf.
205. See GAO Voting Systems Report, supra note 15 at 31; Email from Jeannie Layson, Director of Communications and
Congressional Affairs, U.S. Election Assistance Commission to Susan Greenhalgh, Spokeswoman, Voter Action,
Sept. 22, 2008 (on file with the Brennan Center).
206. GAO Voting Systems Report, supra note 15 at 4-5.
207. Id. at 32-33.
208. E-mail from Jeannie Layson, Director of Communications and Congressional Affairs, U.S. Election Assistance
Commission, to Lawrence Norden, Senior Counsel, Brennan Center for Justice (Sept. 10, 2010, 15:01 EST) (on file
with the Brennan Center).
209. See generally I  I A, T G A O: G A
S  I  I A: I  P F: A
C; F M. K, GAO: G A O  G A
O, CRS R  C (Sept. 10, 2008) .
210. See, e.g., G A O, E R: N S’ E I
F R  C S V R L (Feb. 7, 2006); G
A O, E: R  GAO’ T  V S U  S C
 F  C D (Feb. 8, 2008); G A O, E:
S, T,   D A T  R  I S  M T V
V S E (Sept. 2008).
211. e GAO has asked Congress to consider expanding the EAC’s role under HAVA to address this problem. GAO
Voting Systems Report, supra note 15 at 34.
212. Pub L. 104-152, 49 U.S.C. §§ 30501-30505 (2010).
213. Consumer Product Safety Improvement Act, Pub. L. No. 110-314, 122 Stat. 3016 (2008) (codified in scattered sec-
tions of 15 U.S.C.).
214. See § 212, 122 Stat. at 3048–49 (codified at 15 U.S.C. § 2055a(a), (b)). ere is currently a March 2011 deadline
for implementation of the database, which is tentatively named Saferproducts.org. See, e.g., Consumer Product Safety
Commission Oversight: Current Issues and a Vision for the Future: Hearing before the House Energy and Commerce
Subcommittee on Commerce, Trade and Consumer Protection, 111th Congress (2009) (statement of Inez Tenenbaum,
Chairman, Consumer Product Safety Commission).
215. H.R. R. N. 110–501, at 43–44 (2007).
216. Congress created NHTSA in 1970. See Highway Safety Act, Pub. L. No. 91 –605 § 201, 84 Stat. 1739 (1970).
217. See Transportation Recall Enhancement, Accountability, and Documentation (TREAD) Act, Pub. L. No. 106–414,
114 Stat. 1800 (2000) (codified in scattered sections of 49 U.S.C.). Congress mandated that NHTSA conduct
rulemaking to determine the manner and extent of the early warning data and its collection. § 3(b), 114 Stat. at
Brennan Center for Justice | 109
1801 (“[T]he Secretary shall initiate a rulemaking proceeding to establish early warning reporting requirements for
manufacturers of motor vehicles and motor vehicle equipment to enhance the Secretary’s ability to carry out the
provisions of this chapter.”). e regulations resulting from this mandate ultimately form the structure of the “early
warning” database. See, e.g., C.F.R. § 579.29 (a)(1) (“[E]ach report . . . must be submitted to NHTSAs early warn-
ing data repository identified on NHTSAs Internet homepage . . . . A manufacturer must use templates provided at
the early warning website, also identified on NHTSAs homepage, for submitting reports.”).
218. See Safercar.gov, http://www-odi.nhtsa.dot.gov/index.cfm (last visited May 28, 2010).
219. See Adverse Event Reporting System, http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Sur-
veillance/AdverseDrugEffects/default.htm (last visited May 28, 2010).
220. See, e.g., Keith Bradsher, Firestone Engineers Offer a List of Causes for Faulty Tires, N.Y. T, Dec. 19, 2000,
at C1.
221. See, e.g., 146 C. R.  () (statement of Rep. Markey) (“is legislation was initially prompted by the
Firestone recall of some over 6 million tires used primarily by the Ford Explorer.”).
222 See, e.g., David Shepardson, Agency Ends Probe of Ford Tire Valves, D N, Apr. 30, 2009, at 6B.
223. See, e.g., Losing its Shine, E, Dec. 12, 2009, at 76.
224. Id.
225. Alexandra Burzon, Toyota Complaints Surged After First Recall, W S J, Mar. 11, 2010, at B7.
226. Posting of David Bailey and Chang-Ran Kim to Reuters.com, Toyota Faces New Probe on Corolla Steering, http://
www.reuters.com/article/idUSTRE61D2TS20100217 (Feb 17, 2010, 18:56 EST).
227. Posting of Peter Valdez-Dapena to CNNMoney.com, NHTSA Seeks Maximum Fine Against Toyota, http://money.
cnn.com/2010/04/05/autos/toyota_nhtsa_fine/index.htm?cnn=yes&hpt=Sbin (April 6, 2010, 14:01 EST)
228. See, e.g,, Editorial, Caution Flags at the F.D.A., N.Y. T, July 2, 1998, at A20.
229. See United States Food and Drug Administration, e Public’s Stake in Adverse Event Reporting, http://www.fda.
gov/Drugs/GuidanceComplianceRegulatoryInformation/Surveillance/AdverseDrugEffects/ucm179586.htm (last
visited May 28, 2010).
230. For example, it is slightly onerous for a person to make a complaint on the NHTSA Complaint Reporting Form. It
spans a minimum of six pages, and consumers are, among other things, asked to code portions of their complaints. See
Defects & Recalls, File a Safety Complaint, http://www-odi.nhtsa.dot.gov/ivoq/index.cfm (last visited May 28, 2010).
231. H.R. R. N. 110–501, at 44 (2007).
232. e Secretary of Transportation has delegated this authority, among others, to NHTSA.
See 49 C.F.R. § 1.50 (2009).
233. See 49 U.S.C. § 30166(m) (2008) (“[T]he Secretary shall initiate a rulemaking proceeding to establish early warning
reporting requirements for manufacturers of motor vehicles and motor vehicle equipment to enhance the Secretary’s
ability to carry out the provisions of this chapter.”).
234. 49 C.F.R. § 579.21(a)–(c) (2009).
235. 49 C.F.R. §§ 579.21–579.26.
236. 49 C.F.R. § 579.28(c).
110 | Brennan Center for Justice
237. 49 C.F.R. § 573.1(b)(3) (2009).
238. A defect is defined by law to mean “any defect in performance, construction, a component, or material of a motor
vehicle or motor vehicle equipment.” 49 U.S.C. § 30102(a)(2) (2008).
239. 49 U.S.C. § 30118(c). Manufacturers must send notification via first class mail to all registered owners of the ve-
hicle or equipment in question. If the manufacturer is unable to notify the registered owner, it must notify the most
recent known purchaser. 49 U.S.C. § 30119(d).
240. 49 U.S.C. § 30119(a)(1)-(6).
241. 49 U.S.C. § 30119(3).
242. 15 U.S.C. § 2064(b) (2008).
243. 15 U.S.C. § 2064(c)(1).
244. 15 U.S.C. § 2064(c)(1).
245. 15 U.S.C. § 2064(i)(2).
246. See 21 C.F.R. § 314.80, § 314.98 (2009). Moreover, due to the Dietary Supplement and Nonprescription Drug
Consumer Protection Act of 2006, similar requirements also apply to OTC drugs that are marketed without an
approved FDA application. See Dietary Supplement and Nonprescription Drug Consumer Protection Act, Pub. L.
No. 109–462, 120 Stat. 3469 (2006) (codified as amended at 21 U.S.C. §§ 379aa–381 (2008)).
247. e Form is FDA Form 3500A. 21 C.F.R. 314.80(f) (2009).
248. See, e.g., Medwatch, FDA Form 3500A, http://www.fda.gov/downloads/Safety/MedWatch/HowToReport/Down-
loadForms/UCM082728.pdf (last visited May 28, 2010).
249. 146 C. R.  () (statement of Rep. Markey). Similarly, Congress initially established the reporting
requirements with the non-“early warning” defect in the Motor Vehicle and Schoolbus Safety Amendments of 1974.
See
Motor Vehicle and Schoolbus Safety Amendments, Pub. L. No. 93 –492, 88 Stat. 1470 (1974) (codified as
amended in scattered sections of 49 U.S.C.). Congress passed this legislation, in part, because some manufacturers
refused to notify the public of defects; or, if the manufacturers acknowledged that there were defects, failed to timely
notify consumers of them. See, e.g., H.R. R. N. 93–1191, at 5 (1974).
250. VSTCPM, supra note 16.
251. VSTCPM, supra note 16 at 2.3.2.6.
252. at currently includes most manufacturers of voting systems used in the United States today. See United States
Election Assistance Commission, Registered Manufacturers, http://www.eac.gov/voting%20systems/voting-system-
certification/registered-manufacturers/ (last visited May 28, 2010).
253. GAO Voting Systems Report, supra note 15 at 31.
254. See generally Frederick M. Kaiser, supra note 209.
255. See Bowsher v. Merck & Co., 460 U.S. 824 (1983) (enforcing statutory requirement that government contractors
provide data to the GAO).
256. See 42 U.S.C. § 1971(c).
257. United States v. New York State Board of Elections, et. al, 2006 U.S. Dist. LEXIS 27664 (N.D.N.Y 2006); 28
C.F.R. § 51 (2009).
Brennan Center for Justice | 111
258. See United States Department of Justice, Federal Programs Branch, http://www.justice.gov/civil/Federal%20Pro-
grams.htm (last visited May 28, 2010).
259. 42 U.S.C. § 15511 (2009).
260. See 49 U.S.C. § 30166 (2009); 15 U.S.C. § 2065(b) (2009).
261. 15 U.S.C. § 2065(a) (2009).
262. See, e.g., 49 U.S.C. § 30166(b) (2009) (“e Secretary of Transportation may conduct an inspection or investigation
that may be necessary to enforce this chapter.”).
263. 49 C.F.R. § 554.5 (2009).
264. See Part IV.1.C, supra.
265. See 49 U.S.C. § 30166(g)(1)(B) (2009) (“[T]he Secretary . . . may conduct hearings, administer oaths, take testi-
mony, and require (by subpena [sic] or otherwise) the appearance and testimony of witnesses and the production of
records the Secretary considers advisable.”); 15 U.S.C. § 2076(b)(3) (2009) (“e Commission shall also have the
power . . . to require by subpena [sic] the attendance and testimony of witnesses and the production of all documen-
tary evidence relating to the execution of its duties.”).
266. See 49 U.S.C. § 30166(e) (2009) (e Secretary . . . reasonably may require a manufacturer of a motor vehicle or
motor vehicle equipment to keep records, and a manufacturer, distributor, or dealer to make reports, to enable the
Secretary to decide whether [it] has complied or is complying with this chapter or a regulation prescribed or order
issued under this chapter.); 15 U.S.C. § 2065(b) (2009) (“Every person who is a manufacturer, private labeler, or
distributor of a consumer product shall establish and maintain such records, make such reports, and provide such
information as the Commission may, by rule, reasonably require for the purposes of implementing this Act, or to
determine compliance with rules or orders prescribed under this Act.”).
267. Consumer Product Safety Act, Pub. L. No. 92–573, 15 U.S.C. §§ 2051−2089 (2009).
268. See United States Department of Justice, Voting Section Litigation, Cases Raising Claims Under the Help America
Vote Act (HAVA), http://www.justice.gov/crt/voting/litigation/caselist.php#hava (last visited May 2010).
269. VSTCPM, supra note 16 at 7.
270. See GAO Voting Systems Report, supra note 15 at 4-5.
271. VSTCPM, supra note 16 at 2.6.
272. Id. at 7.1.
273. See e.g., GAO Voting Systems Report, supra note 15 at 31, Hearing on the Election Assistance Commission: Hearing
Before the Subcomm. on Elections of the House Committee on House Administration, 110
th
Cong. (2008) (statement of
Caroline Hunter, Vice Chair, Election Assistance Commission), available at http://cha.house.gov/UserFiles/82_testi-
mony.pdf (stating that the EAC’s power to decertify machines is limited to machines it has certified).
274. Bowsher v. Synar, 478 U.S. 714 (1986).
275. National Traffic and Motor Vehicle Safety Act of 1966, Pub. L. No. 89–564, 80 Stat. 731 (1966) (codified as
amended at 49 U.S.C. §§ 30101–70 (2009)).
276. 49 U.S.C. § 30165(a)(1) (2009).
112 | Brennan Center for Justice
277. See 15 U.S.C. 2069 (2009).
278. Presidential Statement on Signing the Help America Vote Act of 2002, P. P 1926 (Oct. 29, 2002), available
at http://www.america.gov/st/washfile-english/2002/October/[email protected].
279. Id.
280. See Contract No. 071B4200234 between the State of Michigan and Election Systems and Software, Inc. at 70 (May
24, 2004), available at http://accurate-voting.org/contracts/MI/MI_ess_2004.pdf; Contract Award No. NEG-21231,
Group 22300 – Voting Systems and Related Services and Ballot Marking or Other Voting Devices Accessible to Indi-
viduals with Disabilities (Rev. May 9, 2008), available at http://www.panix.com/~burstein/NY_state_2008.pdf.
281. Much of this was adapted from the model contract in the New Jersey Public Advocate Report, supra note 191.
282. See generally New Jersey Public Advocate Report, supra note 191.
283. City Council R. 18A18-0238, 2009, Council Period 18 (D.C. 2009), 2009 D.C. Law L18-0103.
284. S.B. 1404, 2009-10 Reg. Sess. (Cal. 2010).
285. S.B. 541, 2009-10 Reg. Sess. (Cal. 2009).
286. See N A  S  S, H S  U F F  C
O  H A V A (HAVA) (2010), available at http://www.nass.org/index.php?option=com_
docman&task=doc_download&gid=945.
287. is would be particularly important for states that do not require federal certification and rely on their own testing programs.
288. 42 U.S.C. § 15481 (a) (2009) defines a voting system as “(1) the total combination of mechanical, electromechanical,
or electronic equipment (including the software, firmware, and documentation required to program, control, and sup-
port the equipment) that is used— (A) to define ballots; (B) to cast and count votes; (C) to report or display election
results; and (D) to maintain and produce any audit trail information; and (2) the practices and associated documenta-
tion used— (A) to identify system components and versions of such components; (B) to test the system during its
development and maintenance; (C) to maintain records of system errors and defects; (D) to determine specific system
changes to be made to a system after the initial qualification of the system; and (E) to make available any materials to
the voter (such as notices, instructions, forms, or paper ballots).
289. See Brennan Center for Justice, VRM in the States: Electronic Poll-Books, http://www.brennancenter.org/content/
pages/vrm_electronic_poll-books (last visited Apr. 26, 2010).
290. See, e.g., Mike Saewitz, Chesapeake’s electronic pollbooks were voting-delay villains, T V-P, Nov. 6, 2008,
available at http://hamptonroads.com/2008/11/chesapeakes-electronic-poll-books-were-votingdelay-villains, Mary
Lou Pickel, Countdown 2008: Computer Woes Slow Georgians at Polls, T A J-C, Oct.
28, 2008, available at http://www.ajc.com/services/content/printedition/2008/10/28/advancevote.html. T.J. Aulds,
No ‘Super’ Glitches Reported for New Voting System, T G C D N, Nov. 4, 2009, available at
http://www.galvnews.com/story.lasso?ewcd=2dfb9d86700788db, Posting of Mike McPhee to denverpost.com, Ritter
Victorious, Voters Struggle, http://www.denverpost.com/outdoors/ci_4616285 (last updated Jul. 9, 2009, 17:37 EST).
291. Definition from T G D C, V V S G R-
   U S E A C, A A 6 (2007).
292. Trevor Stokes & Bernie Delinski, Northwest Alabama Counties Pick Huckabee and Clinton, T D, Feb.
6, 2008, available at http://www.timesdaily.com/article/20080206/NEWS/802060333?Title=Northwest -Alabama-
counties-pick-Huckabee-and-Clinton.
293. Another Voting Glitch in Baldwin County, A P (Nov. 14, 2006).
Brennan Center for Justice | 113
294. John Archibald & Brett J. Blackenridge, Early Printout Was Only Source of Wrong Tally, B N (Nov. 10,
2002); Jeffrey Gettleman, Alabama Governor Ends Challenge of Vote Count, N Y T (Nov. 19, 2002).
295. Bill Hess, Computer Glitch Causes Hiccup in Cochise County Tally, D D, Feb. 7, 2008, available at
http://www.douglasdispatch.com/articles/2008/02/07/news/doc47ab6bc4d0b70550988123.txt.
296. McComish GOP Winner in District 20 by 13 Votes, Recount Shows, T A R (Sept. 21, 2004).
297. Nedra Lindsey, McComish’s District 20 2nd-place Finish Confirmed, T A R (Sept. 24, 2004).
298. Anna Mathews, Williams, Link Ousted in General Election, C C N, Nov. 6, 2008, available at
http://www.carrollconews.com/stoy/1475931.html.
299. Wes Clement, Error On Redfield, Altheimer Ballots Addressed at Emergency Meet, P C., Oct. 26, 2008,
available at http://www.votersunite.org/article.asp?id=8058.
300. Rachel Paul Dickerson, Recount Reverses Results in District 45 Race, T., May 24, 2008,
http://www.thecabin.net/stories/052408/loc_0524080001.shtml.
301. Gary Lookadoo & Tom Treweek, Slow Going: County Results Trickle In: Unofficial Numbers Show Clinton, Huckabee in
Lead, B C D R, Feb. 6, 2008, available at http://www.votersunite.org/article.asp?id=7502.
302. Joe Lamb, Haggard Notes Problems with Voting Machines, T., Feb. 6, 2008, available at http://thecabin.net/
stories/020608/loc_0206080006.shtml.
303. Louis Short, Vote Machine Problems Reported, (Heber Springs, AR) S-T (Nov. 27, 2006).
304. Trish Hollenbeck, Voting Machines To Get Fix After Audit Trail Prints Right Vote For e Wrong District, N
A T (Oct. 26, 2006).
305. Daniel Nasaw, Voting Machine Claim to be Studied by County, A D-G, May 20, 2006.
306. Height Could Make a Difference in this Year’s Elections (THV Broadcast, May 20, 2006) (transcript available at
http://www.todaysthv.com/news/news.aspx?storyid=28836).
307. Anna Mathews, Computer Glitch Blamed for Miscount in JP Voting, C C S T (Nov. 10,
2004).
308. LeAnn Askins, Commission OKs Results of Elections, J S (May 28, 2004).
309. George Jared, No Explanation for Ballot Machine Malfunction, S M (May 27, 2004).
310. adeus Greenson, Software Glitch Yields Inaccurate Election Results, T S, Dec. 5, 2008, available at http://
www.times-standard.com/localnews/ci_11145349.
311. Kim Zetter, Serious Error in Diebold Voting Software Caused Lost Ballots in California County, T L (Blog of
W M), Dec. 8, 2008, http://www.wired.com/threatlevel/2008/12/unique-election; Jaikumar Vijayan,
California Finds E-Voting Software Had Errors, Data Deletion Functions, C, Mar. 4, 2009.
312. Karen de Sá, More Touch-Screen Machines Malfunction, S J M N, Nov. 5, 2008, available at
http://www.votersunite.org/article.asp?id=8173.
313. Darrell R. Santschi & Cindy Martinez Rhodes, Software Glitch Led to Precinct, Vote Uncertainty in SB County, P
E, June 5, 2008, available at http://www.pe.com/localnews/inland/stories/PE_News_Local_N_
glitch05.41ce262.html.
114 | Brennan Center for Justice
314. Bobby Caina Calvan, Machine Vendor Miscalibrated Vote Scanners, County Officials Say, S B, Mar. 24,
2008, available at http://www.votersunite.org/article.asp?id=7613.
315. John Cote & Jonathan Curiel, Gaming Initiatives in Lead as Voters Crowd Polls, S.F. C., Feb. 6, 2008, available at
http://www.sfgate.com/cgi-bin/article.cgi?f=/c/a/2008/02/06/BA7NUSERR.DTL.
316. Nicole C. Brambila, Glitches in System Slowed Vote Count, D S, Feb. 7, 2008, at 1A; Correction, D S,
Mar. 8, 2008, at 1A.
317. Antone Gonsalves, San Francisco Finds Fix for Election Day Tech Snafu, IW., Nov. 5, 2004, 15:03
EST.
318. Ian Hoffman, Lawmakers Cut E-Votings Paper Trail: Manufacturers Demonstrating New Printers in Nevada Were
Embarrassed When Machine Failed to Recognize Votes, T-V H (Aug. 13, 2004); Kim Zetter, Wrong Time for
an E-Vote Glitch, W N (Aug. 12, 2004).
319. Conaughton Gig, New Electronic Scanners Miscounted Some County Votes, N C T (Apr. 7, 2004).
320. Luis Monteagudo Jr. & Helen Gao, Some Votes Miscounted in Primary, Officials Say, U-T (Apr. 8, 2004).
321. Luis Monteagudo Jr. and Helen Gao, Glitches in Voting Machines Examined, S D U T, Mar. 4,
2004, at B-1.
322. Hearing Before the Cal. Sec’y of State Voting Systems and Procedures Panel (Apr. 21, 2004) (statement of James
Dunn).
323. Ernest Luning, Adams County ‘Quarantines’ Machine at Switched Candidate’s Vote, C. I., Oct. 29, 2008,
available at http://coloradoindependent.com/13187/adams-county-quarantines-machine-that-switched-candidates-
vote.
324. John Ingold, Vote Scanners Still Face Doubts, D P, Feb. 17, 2008, available at http://www.denverpost.com/
headlines/ci_8283953.
325. Ann Imse, Big Bar Code Backfire, R M N (Nov. 14, 2006).
326. Allyn Harvey, Phantom Votes Haunt Pitkin County Election Officials, A T (Nov. 3, 2005).
327. Editorial, Voting Problems: Count Every Vote, But Not by Hand, N-T, May, 30, 2008, available at
http://www.votersunite.org/article.asp?id=7727.
328. State Officials Expecting Record Turnout for Presidential Primary, B G, Feb. 5, 2008, available at
http://www.votersunite.org/article.asp?id=7497.state_officials_expecting_record_turnout_for_presidential_primary/.
329. Dave Collins, Registrars Busy Keeping Up with Demand for Ballots, N, Feb. 5, 2008, available at
http://www.votersunite.org/article.asp?id=7489.
330. Nikita Stewart and Elissa Silverman, Primary Vote Still Doesn’t Add Up, W. P, Sept. 22, 2008 at B1; Nikita
Stewart, District’s Primary Results Certified, WASH. POST, Sept. 26, 2008 at B1.
331. S V S, R   D  C B  E  E 1 (Sept. 22,
2008).
332. Michael E. Ruane, District Voters Line Up at Polls, W. P, Feb. 12 2008, available at http://www.washington
post.com/wp-dyn/content/article/2008/02/12/AR2008021200818.html?sid=ST2008021200978.
333. Christian M. Wade, Voting Glitches Mar Election, T T., Nov. 5 2008, available at http://www2.tbo.com/
content/2008/nov/05/na-voting-glitches-mar-election/news-politics/.
Brennan Center for Justice | 115
334. Jane Musgrave, Palm Beach County Vote Results Still Incomplete, P B P, Nov. 5 2008, available at
http://www.palmbeachpost.com/localnews/content/local_news/epaper/2008/11/05/1105votecount.html.
335. Paige Kornblue, Palm Beach County Precinct Problems, WPTV., Nov. 4 2008, available at http://truevote.us/
nucleus/index.php?itemid=749.
336. Christopher O’Donnell, Glitch Prevents Manatee County From Issuing Final Election Results, H T., Nov. 4
2008, available at http://www.heraldtribune.com/article/20081104/BREAKING/811040175?Title=Glitch_prevents_
Manatee_County_from_issuing_final_election_results.
337. Christopher O’Donnell & Todd Ruger, Vote Software is Under Fire, H T., Nov. 6 2008, at BN1, available at
http://www.heraldtribune.com/article/20081106/ARTICLE/811060374?p=3&tc=pg.
338. Mary K. Garber, Examining Florida’s High Invalid Vote Rate in the 2008 General Election Part I: How Voting System
Design Flaws Led to Lost Votes, F F E C, June 23, 2009, available at
http://www.ffec.org/documents/Invalid_Vote_Report_Revised_23June2009.pdf.
339. Lawrence Norden, Concerns About New York’s New Voting Machines, RNY, (Jan. 25, 2010)
http://reformny.blogspot.com/2010/01/concerns-about-new-yorks-new-voting.html.
340. Jane Musgrave, Creases Delay Tally of Absentee Ballots, P B P, Oct. 31 2008, available at
http://www.palmbeachpost.com/politics/content/local_news/epaper/2008/10/31/a1a_absentee_1101.html.
341. Dave Wax, Early Voting off to Rough Start, FCN., Oct. 20, 2008, http://www.firstcoastnews.com/
news/breaking/news-article.aspx?storyid=121489&catid=17.
342. Lines, Voting Problems Continue on 2nd Day, N4J., Oct. 21, 2008, http://www.news4jax.com/news/
17770664/detail.html.
343. Problems Force Flagler to Switch to Old Voting Machines, F N-J., Oct. 22, 2008, available at
http://truevote.us/nucleus/index.php?itemid=381.
344. John A. Torres, Tally Delay Blamed on Machines, F. T, Aug. 28, 2008, at A7.
345. Steve Bousquet, Janet Zink, Anne Lindberg & Craig Pittman, Voter Turnout Worst in Ten Years, S. P T,
Aug. 27, 2008, at 1B, available at http://www.tampabay.com/news/politics/elections/article785994.ece.
346. Id.
347. Seán Kinane, Vendor Blamed for Optical Scan Problems, WMNF E N, Aug. 27, 2008, available at
http://www.wmnf.org/news_stories/vendor-blamed-for-optical-scan-problems.
348. Computer Glitch Delays Vote Count in Sarasota County, WWSB ABC 7 MSC., Aug. 27, 2008,
http://www.mysuncoast.com/Global/story.asp?S=8911224.
349. omas R. Collins, Nearly 700 West Palm Votes not Tallied Election Night, P B P, June 28, 2008, available at
http://www.palmbeachpost.com/localnews/content/local_news/epaper/2008/06/27/0627wpbelex.html.
350. Bugs Removed from Pinellas Vote Tabulation System, S. P T, June 4, 2008, available at
http://www.tampabay.com/news/localgovernment/article605711.ece.
351. Derek Simmonsen, Delay, Glitch Creates Headache for St. Lucie Voters, TCP, Jan. 31, 2008, available at
http://www.tcpalm.com/news/2008/jan/31/delay-glitch-tallied-in-st-lucie-voting/.
352. David Fleshler, Debut of Voter Check-in System Marred by Complaints, S-S, Jan. 30, 2008, at 1B.
353. Hector Florin, Kathleen Chapman, & Jennifer Sorentrue, County Officials’ Good Day Turns into Rough Night, P
B P, Jan. 30, 2008, at 10A.
116 | Brennan Center for Justice
354. James Miller, Canvassing Board Orders Another Count of Early Ballots, D B N J., Jan. 29, 2008, available
at http://voteraction.org/node/342.
355. At the Polls: Problems with Voting Machines; Reaction Mixed on Tax Amendment, H T, Jan. 29, 2008,
available at http://www.heraldtribune.com/article/20080129/BREAKING/63466273.
356. Luisa Yanez & Martin Merzer, Glitches Blemish South Florida Voting, M H, Jan. 29, 2008, available at
http://www.voteraction.org/node/344.
357. Beau Zimmer, Touch-Screen Problems Reported in Hillsborough, 10C., Jan. 29, 2008,
http://www.wtsp.com/news/local/story.aspx?storyid=72767.
358. Luis F. Perez, Palm Beach County Records 32,000 Early Votes for Tuesdays Election, S F S-S
(Nov. 2, 2006).
359. Dozens of Voters Complain about Glitch, H T (Nov. 8, 2006).
360. Doug Sword & Anna Scott, House Could End up in 13th District Dispute, S H T (Nov. 9, 2006).
361. Todd Ruger, Voting Glitch Prompts Warning, S H T. (Nov. 5, 2006).
362. See F D  E, A  R  O  U   2006
G E, General Over Undervote Table, available at http://election.dos.state.fl.us/reports/index.shtml.
363. Mary K. Garber, Voting System Performance Problems – Conclusions and Recommendations, in Assessing Election
Accuracy in Florida’s 2006 General Election: A Comparison of Voting System Performance Using Undervote Rates in State-
wide Races in Florida’s 2006 General Election (Dec. 2008) (unpublished manuscript, on file with the Brennan Center).
364. Charles Rabin & Darran Simon, Glitches Cited in Early Voting; Early Voters are Urged to Cast eir Ballots with Care
Following Scattered Reports of Problems with Heavily Used Machines, M H (Oct. 28, 2006).
365. Jeremy Milarsky & Rafael A. Olmeda, Broward County, Fla., Fixes Computer Error To Correct Statewide Gambling Vote,
S F S-S (Nov. 5, 2004).
366. D M  J L H, PA  E-V P H N
 H S  T, NRC W 18 (2004) [hereinafter Mulligan and Hall]
367. Mary Ellen Klas and Gary Fineout, Touchscreen Voting Machines Have Software Flaw, Florida Officials Learn, M
H, Jun. 12, 2004, at 1B.
368. Mary Ellen Klas, Flaw in Florida Touch-Screen Voting Machines Was Known in 2002, M H, Jun. 12, 2004, at
1B.
369. Election Glitch Missed 103,000 Votes in Florida County, CNN (Nov. 8, 2002).
370. Bev Harris, Can We Trust ese Machines?, B B V 4 available at http://www.blackboxvoting.org/bbv_
chapter-2.pdf.
371. Oscar Corral, Technicians Error, Not Machines, to Blame in Dade Election Mix-Up, M H (Apr. 4, 2002).
372. Rhonda Cook, Senior Citizens Among ose Who Have Voting Problems, A J-C, Nov. 4.,
2008, available at http://www.ajc.com:80/traffic/content/metro/atlanta/stories/2008/11/04/problems_election.html.
373. Mary Lou Pickel, COUNTDOWN 2008: Computer Woes Slow Georgians at Polls, A J-C,
Oct. 28, 2008, available at http://www.ajc.com/ajchomefinder/content/printedition/2008/10/28/advancevote.html?
cxntlid=inform_artr
Brennan Center for Justice | 117
374. Glitch Keeps Fulton Voters Waiting For Hours, CBSA., Oct. 23, 2008, available at
http://www.cbsatlanta.com/news/17785206/detail.html.
375. Alan Riquelmy, Turnout Higher than Expected, L-E, Feb. 7, 2008, available at
http://www.votersunite.org/article.asp?id=7517.
376. Stephanie Reid, State Workers Fan Out When Problems Arise at Polls, A J-C, Feb. 5, 2008,
available at http://www.votersunite.org/article.asp?id=7472.
377. Terry Dickson, Grovner Beats Deverger in a Court-ordered Rematch for McIntosh County, F. T-U Feb. 6, 2008,
available at http://jacksonville.com/tu-online/stories/020608/geo_244103741.shtml.
378. Rhonda Cook, Georgia Votes 2006: Machine Glitches: Smooth Day? No Way, A J-C (Nov. 8,
2006).
379. Primary Election Runs Into Problems; Some Problems Caused by Electronic System, KITV 4 N (Sept. 24, 2004).
380. Rebecca Boone, Dems Critical of Idaho Election Officials, C S T (Nov. 9, 2006).
381. Scott Wuerz, Shiloh Mayoral Hopeful is One of 5 Candidates Seeking Recounts in April Elections, B
N-D, May 2, 2009, available at http://www.votersunite.org/article.asp?id=8392.
382. Scott Wuerz, With Discovery of Uncounted Ballots, Vernier Now Appears to Have Won Re-Election in Shiloh, B
N-D, April 8, 2009, available at http://www.votersunite.org/article.asp?id=8371.
383. Nancy Gier, Democrats Reporting Irregularities in 14th District Voting Machines, C. D H, Mar. 8, 2008,
available at http://www.dailyherald.com/story/?id=150092.
384. Russell Lissau, Consultant to Investigate Lake County Malfunction, C. D H, Feb. 7, 2008, available at
http://www.dailyherald.com/story/?id=130066&src=3.
385. Stephen Blackman & Megan Crepeau, Election ‘08: City Precinct Misses Half of Ballots in Count, D
N, Feb. 6, 2008, available at http://www.dailynorthwestern.com/2.13893/election-08-city-precinct-
misses-half-of-ballots-in-count-1.1922000.
386. John McCormick,Voting Equipment Glitches Lingering, C T (Nov. 2, 2006).
387. Susan Kuczka, Returns Are In: Software Goofed – Lake County Tally Misled 15 Hopefuls, C T (Apr. 4,
2003).
388. Bill Engle, County’s Count Last in State; Glitch Delays Wayne Election Results Until After Midnight, P-I,
Nov. 6, 2008, available at http://www.votersunite.org/article.asp?id=8181.
389. Editorial, Ballot-Counting Problem Should Have Been Avoided, H B., Nov. 4, 2008, available at
http://www.theheraldbulletin.com/opinion/local_story_309220039.html?keyword=topstory.
390. Disabled Voters Disappointed with Touch-Screen Problems, WISH-TV (Nov. 10, 2006).
391. Rick Yencer, Voting Machines to be Repaired, S P, Sept. 26, 2006, at 3A.
392. Jason omas, Poll Machine Flaw Hidden, State Says, I S, Sept. 25, 2006, at 1.
393. Computer Error on Ballot Count Changes Results for County Commissioner in Indiana, A P (Nov. 12,
2004).
394. Jens Manuel Krogstad, Rogers Requests Total Recount; Danielson Clings to Lead by 14 Votes, WCF C, Nov. 13,
2008, available at http://www.wcfcourier.com/news/local/article_0ef93257-c12f-5706-b764-04ac8697e04b.html.
118 | Brennan Center for Justice
395. Kurt Allemeier, Computer Glitch Slows Scott County Vote Reporting, Q-CT, Nov. 6, 2008, available at
http://www.qctimes.com/news/local/article_23535b59-8472-5286-9633-b75cd40a1c96.html?mode=story.
396. Tim Rohwer, Faulty Voting Machines Delay Results; Counting Under Way, T D N O (June 7,
2006).
397. Lynn Campbell, Broken Vote-Counting Machines, Moveon Draw Complaints In Iowa, D M R (Nov. 3,
2004).
398. Kevin Flaherty, Error Comes from Voting Machine Programming, M S, Nov. 6, 2008, available at
http://www.morningsun.net/news/x635432391/Error-comes-from-voting-machine-programming
399. Gary Demuth, Errors Found in Voting Machines, S J., Apr. 8, 2009, available at http://www.saljournal.com/news/
story/Faulty-Election-booths-4-7.
400. Aug. 6 Ballot Problems Alleged; Clay, Barton County Candidates Seek Review of Races, L J-W 
N (Aug. 22, 2002).
401. Bill Robinson, Local Turnout Heavy as Computer Glitch Delays Count, R R., Nov 4, 2008, available at
http://www.richmondregister.com/localnews/local_story_310073742.html?keyword=secondarystory.
402. Patrick Crowley, E-Slate Ballot Glitch Ignites Blame Game, C E., Nov. 5, 2008, at 1A.
403. Melissa Gagliardi, Council Results Finally Released, L C-J., May 22, 2008, available at
http://www.votersunite.org/article.asp?id=7714.
404. Warren Stewart, All Four Major E-Voting Machines Flip Votes in Early Voting, VTUSA (Nov. 5, 2006), available
at http://www.votetrustusa.org/index.php?option=com_content&task=view&id=1967&Itemid=26.
405. Steve Wilson, Electronic Voting Glitch Reported in Delcambre Race, D I, Oct. 9, 2008, available at
http://www.iberianet.com/articles/2008/10/09/news/doc48ee56fed1702965734605.txt.
406. Colin Hickey, Results Slowed by Machine Trouble, J  M S (Nov. 9, 2006).
407. E.B. Ferguson III, Election Workers Prepare for Long Day as Polls Open, T C, Feb. 12, 2008, available at
http://www.voteraction.org/node/401.
408. Steve Urbon, Ballot Glitch A Mystery, S-T, Sep. 10, 2008, available at http://www.southcoasttoday.com/
apps/pbcs.dll/article?AID=/20080910/NEWS/809100332.
409. Robert Carroll, Recount Affirms: Abington Approves Another Override, B G, May 18, 2008, available at
http://www.boston.com/news/local/articles/2008/05/18/recount_affirms_abington_approves_another_override/.
410. Charles Winokoor, Local Vote for Hillary, Mitt, T D G, Feb. 7, 2008, available at
http://www.tauntongazette.com/news/x1224705221.
411. Ben Cunningham, Malfunction Delays Hasting Results, G R P (May 4, 2006).
412. Steve Gunn & Lynn Moore, Election Turns Around When Inspectors ‘See e Light, M C (Sept. 4,
2004).
413. Problems Delay Final Vote Count, H-A (May 7, 2005).
414. Ed Kemmick, Software Flub Leads to County Recount, B G (Nov. 8, 2006).
415. Nate Jennings, Problem Machines Spur Call for Recount, L J S (Nov. 14, 2004).
416. Countinghouse Blues: Too Many Votes, WOWT 6 O (Nov. 5, 2004).
Brennan Center for Justice | 119
. Joe Dejka & Chris Olson, A Late Night in Sarpy; Glitches Delay Results, O W-H (Nov. 6, 2002).
418. Frank Zoretich, Election Results Certified After Software Blamed, A T, Nov. 19, 2002 (on file with
the Brennan Center) [hereinafter Zoretich].
419. Gordon Fraser, Faulty Keyboard Leads to Erroneous Vote Result in Salem, E T., Mar. 13, 2008, available at
http://www.eagletribune.com/punewsnh/local_story_073055256.html?keyword=topstory.
420. Bob Jordan, Problems Persist with Monmouth Election Computer System, A P P, June 3, 2008, available at
http://www.votersunite.org/article.asp?id=8435.
421. Paul Cox, Voting-Machine Malfunctions Crop up at Some Polling Places, S-L, Nov. 4, 2008, available at
http://www.nj.com/news/index.ssf/2008/11/votingmachine_malfunctions_cro.html.
422. Ken orbourne, Hudson County Vote Count Hits Major Snag, J J., Nov. 4, 2008, available at http://www.nj.com/
elections/index.ssf/2008/11/hudson_county_vote_count_hits.html.
423. Bob Jordan, Delays Reported in Recording Election Results, A P P, June 5, 2008, available at
http://www.votersunite.org/article.asp?id=7740.
424. Diane C. Walsh, Machines Get Vote of No Confidence After Errors in Primary, S-L, Feb. 20, 2008,
http://blog.nj.com/newsletters/2008/02/am_edition_19.html.
425. Andrew W. Appel, et. al., Princeton University, I  I   S AVC A
9.00H DRE V M 117 (October 17, 2008).
426. Erica Zarra, Two Electronic Ballot Machines Replaced in Montclair During Vote, M T, Feb. 7, 2008,
available at http://www.votersunite.org/article.asp?id=7512.
427. Brian T. Murray, An Error at Hudson Polls Delays Official Tally of Primary Ballots, S-L, Feb. 7, 2008, at 19.
428. Dave Evans, Voting Problems Reported in NJ, WABC-TV N J (Nov. 7, 2006).
429. Sara K. Satullo, Voting Machines Stopped Working, E-T, Feb. 7, 2008, available at http://voteraction.org/
node/376.
430. Sharna Johnson, County Vote Canvass Approved, C N J., June 13, 2008, available at
http://www.cnjonline.com/news/county-28628-canvass-election.html.
431. Barry Massey, New Mexico to Check Troubled Voting Equipment, A P, Mar. 21, 2008, available at
http://votersunite.org/article.asp?id=7610.
432. Warren Stewart & Ellen eisen, Summary Report on New Mexico State Election Data (www.HelpAmericaRecount.org
Jan. 4, 2005), available at http://www.democracyfornewmexico.com/democracy_for_new_mexico/files/
NewMexico2004ElectionDataReport-v2.pdf.
433. Zoretich, supra note 418; 12,000 Early Votes Skipped in Count, A T (Nov. 16, 2002).
434. New Voting Machines Falter, Could Cost Big Bucks, WPTZ., Nov. 10, 2008, http://www.wptz.com/news/17949935/
detail.html.
435. Liz Lawyer, One Sequoia Machine Faulty; No Other Major Hiccups in Tompkins County, I J., Nov. 4, 2008,
available at http://www.votersunite.org/article.asp?id=8135.
436. Sue Book, Revised Numbers Add to Turnout in Craven Election, S J., Nov. 13, 2008, available at
http://www.enctoday.com/news/votes-42894-nbsj-election-county.html.
120 | Brennan Center for Justice
437. Wesley Young, Gap in Judicial Contest Narrows, W-S J., Nov. 8, 2008, available at
http://www2.journalnow.com/content/2008/nov/08/gap-in-judicial-contest-narrows/.
438. Steve Lawson, Voting Glitch in Rockingham County, GR., Nov. 5, 2008, available at
http://www.votersunite.org/article.asp?id=8158.
439. Chris Dailey, Revised Results Give Democrats Sweep in County Board Race, T D B., Nov. 6, 2008, available
at http://tryondailybulletin.com/index.php?option=com_content&task=view&id=1370&Itemid=192.
440. Heather Gale, ousands of Votes Missed in Tuesday Tallies, J D N, May 9, 2008, available at
http://www.jdnews.com/news/votes-56535-board-onslow.html.
441. Shawn Flynn, Mecklenburg, Wake Find Vote Flaws, N14., May 5, 2008, http://news14.com/charlotte-news-
104-content/top_stories/595595/mecklenburg--wake-find-vote-flaws/.
442. Cecil Bothwell, Buncombe Machines Disenfranchise Voters, M X (Nov. 10, 2004).
443. Gary D. Robertson, Months After Election, North Carolina’s Agriculture Commissioner Agrees He Lost, A P
(Feb. 5, 2005).
444. Binyamin Appelbaum, Gaston Investigates Election Tally Errors, T C O, November 16, 2004, at 4B.
445. See Binyamin Appelbaum, Names, Ballots Fail to Tally in 29 Gaston Precincts, T C O, November
18, 2004, at 1B.
446. Associated Press, Recount Changes One Franklin Co. Race, I S, Nov. 12, 2004, available at
http://www.votersunite.org/article.asp?id=3804.
447. Jessica Rocha, Mechanic to Smooth Vote, N  O (Oct. 14, 2004).
448. Wade Rawlins & Rob Christensen, Winners” May Be Losers, N  O (Nov. 12, 2002).
449. Kim Zetter, E-Vote Machines Drop More Ballots, W, Feb. 9, 2004, available at http://www.wired.com/politics/
security/news/2004/02/62206.
450. Ellen eisen, Ballot Definition Files: No Review is Provided for a Key Component of Voting System Software,
VU (June 2004, updated Oct. 2005 & June 2006), available at http://www.votersunite.org/info/
BallotProgramming.pdf.
451. Lynn Hulsey, Audit Shows Issues with Area Voting Machines, D D N, Dec. 17, 2008, at A6, available at
http://www.daytondailynews.com/n/content/oh/story/news/local/2008/12/17/ddn121708 votingmachine.html.
452. Ben Marrison & Jill Riepenhoff, Voting Machine Complaints Trickle In, C D, Nov. 4, 2008, available at
http://blog.dispatch.com:80/vote08/2008/11/a_woman_who_voted_at.shtml.
453. Paul Souhrada, Obama, McCain Missing From Some Ballots, C D, Nov. 4, 2008, available at
http://blog.dispatch.com:80/vote08/2008/11/obama_mccain_missing_from_some.shtml.
454. Stark County Voters Bemoan Machine Troubles, C R, Nov. 4, 2008, available at http://www.truevote.us/
nucleus/index.php?itemid=801.
455. Letter from Betty McGary, Executive Director, and Lynn Kinkaid, Deputy Director, Butler County Board of Elections,
to Dave Byrd, President, Premier Election Solutions (Apr. 4, 2008) (on file with the Brennan Center).
456. Lynn Hulsey, Voting Machine Maker Says Error May Cause Votes to Be Uncounted, D D N, Aug. 23, 2008,
at A6.
Brennan Center for Justice | 121
457. Letter from Dave Byrd, President, Premier Election Solutions, to Jennifer Brunner, Ohio Sec’y of State (Aug. 19, 2008)
(on file with the Brennan Center).
458. Barbara Carmen, County’s Voting Machines Examined, C D, Mar. 16, 2008, available at
http://www.dispatch.com/live/content/local_news/stories/2008/03/16/BOEPROBE.ART_ART_03-16-08_
B1_9F9LIV3.html.
459. Mark Niquette, Officials See Few Glitches Across Ohio, C D, Mar. 4, 2008, available at
http://blog.dispatch.com/primary/2008/03/officials_see_few_glitches_acr.shtml.
460. Errors Plague Voting Process in Ohio, Pa, V (Nov. 3, 2004); Brian Gottstein, Back Up! We Need Back Up!,
R T (Nov. 11, 2004).
461. LaRaye Brown, Some Ballots Counted Twice: Discovery Raises Further Doubt About Close Treasurer Race,
N-M (Nov. 16, 2004).
462. John McCarthy, Machine Error Gives Bush 3,893 Extra Votes in Ohio, USA T, Nov. 6, 2005, available at
http://www.usatoday.com/tech/news/techpolicy/evoting/2004-11-06-ohio-evote-trouble_x.htm.
463. Rob Scott, Computer Glitch Stalls Northumberland County Vote Count, D I, Nov. 4, 2008, available at
http://www.dailyitem.com/0100_news/local_story_309211730.html.
464. George Mattar, Machines Malfunction in 5 Towns, B C C T, May 20, 2009, available at
http://www.phillyburbs.com/news/local/courier_times/courier_times_news_details/article/28/2009/may/20/machines-
malfunction-in-5-towns.html.
465. Early Polling Issue Reported in Green Tree, KDKA-TV, Apr. 22, 2008, http://www.votersunite.org/article.asp?id=7644.
466. Gabrielle Banks, Scattered Problems Reported with Voting Machines, P P-G (Nov. 7, 2006).
467. Michael P. Buffer, Errors Discount Election More, T L (May 23, 2006).
468. Board Should Hold a Revote for the City, Editorial, R E (July 20, 2005).
469. John Hilton, DJ Race Still Up in the Air, T S (Nov. 11, 2005).
470. Marika Kelderman, Charleston County Suffers Voting Problems on Election Day, L5N., Nov. 5, 2008,
http://www.live5news.com/Global/story.asp?s=9300365&clienttype=printable.
471. Renee Dudley, Machine Ballots Omit Candidates’ Names, I P, Oct. 22, 2008, available at
http://www.islandpacket.com/news/local/story/645376.html.
472. Mike Cherney & Aliana Ramos, Results are In: McCain Wins, S N, Jan. 22, 2008, available at
http://www.votersunite.org/article.asp?id=7423; Mike Cherney, Vote Results Further Delayed, S N, Jan. 21, 2008,
available at http://www.votersunite.org/article.asp?id=7421.
473. Robert Behre, Charleston County Voting Under Way; Ballot Misprint, Machine Quirk Get Attention, P  C
(Nov. 3, 2006).
474. Kristy Eppley Rupon, Counting Error Overstates Votes, T S (Apr. 29, 2005).
475. Vote Counting Machine Error Results in Election Vote Changes, KOTA R 1380, June 4, 2009,
http://www.kotaradio.com/news.asp?eid=5002&ID=3296.
476. Clay Bailey, Dozens in Bartlett Unable to Vote in City’s Election, M C. A, Nov. 4, 2008, available at
http://www.commercialappeal.com/news/2008/nov/04/dozens-bartlett-unable-vote-citys-election/.
477. Anne Paine, Tennesseans Report Voting Problems, T, Oct. 25, 2008, available at http://www.votersunite.org/
article.asp?id=8052.
122 | Brennan Center for Justice
478. Rebecca Ferrar, Voting Machine Issue Confusing to Some, K N S, Oct. 22, 2008, available at
http://www.knoxnews.com/news/2008/oct/22/voting-machine-issue-confusing-some/.
479. Jennifer Clendenion, Machine Problems Frustrate Early Voters, D C C., Oct. 22 2008, available at
http://www.decaturcountyonline.com/article.asp?art=1244.
480. State Rep Advises Voters to Check Ballot Twice, KBMT N, Oct. 31, 2008, http://www.kbmt12.com/news/local/
33637934.html.
481. Kim Zetter, Texas Voters Say Straight-Party Voting Flipped Votes; Activists Urge Voters to Avoid Straight-Party Voting, W,
Oct. 29 2008, http://www.wired.com/threatlevel/2008/10/straight-party/.
482. Martina Stewart, Texas Countys Machines Causing Concern for Some Voters, CNN (Oct. 28, 2008), http://politicalticker.
blogs.cnn.com/2008/10/28/tx-countys-machines-causing-concern-for-some-voters/.
483. Libby Cluett, Concerns Raised Over Electronic Voting, M W I, Oct. 24, 2008, available at
http://www.mineralwellsindex.com/local/local_story_298161535.html.
484. Marti Alexander, City Discovers Error, A C J., May 19, 2008, available at http://www.votersunite.org/
article.asp?id=7709.
485. Lauren Winchester, Problem Machines Cause Voting Delay, DT, Mar. 5, 2008, available at
http://www.votersunite.org/article.asp?id=7585.
486. Steven Alford, Late-Night Vote-Tally Snafus ‘Probably Set Some Records, L D N, Mar. 5, 2008, available at
http://www.votersunite.org/article.asp?id=7587.
487. El Paso’s Electronic Voting Machines Have Problems, KFOX-TV N (Nov. 3, 2006); Vic Kolenc, Voting Machine
Problems Checked, E P T (Nov. 4, 2006).
488. Glitch Forces Manual Tabulation for U.S. House Race, D M N (Nov. 8, 2006).
489. Tom Bower, Bexar Computer Glitch Delays Counting of Votes, S A E N (Mar. 10, 2004).
490. Jefferson County Voters Continue to Raise Concerns About Voting Machines, KFDM-TV (Oct. 28, 2006).
491. Anna M. Tinsley & Anthony Spangler, Vote Spike Blamed on Program Snafu, F W S-T (Mar. 9,
2006); Anna M. Tinsley, Judicial Candidate Files Challenge, F W S-T (Apr. 6, 2006).
492. Brian Williams, Software Blamed in Precinct 8 Democratic Chair Race Mixup, L A-J (Mar. 11,
2004).
493. Ballot Glitches Reverse Two Election Results, T H C (Nov. 8, 2002).
494. David Demille, Washington County Vote Underway, S, Oct., 22, 2008, available at http://www.votersunite.org/
article.asp?id=8025.
495. Sandhya Somashekhar, Ballot Machine Malfunctions, Fairfax Race Left in Limbo, W. P, Mar. 11, 2009, at B1,
available at http://www.washingtonpost.com/wp-dyn/content/article/2009/03/10/AR2009031002068.html.
496. Julia O’Donoghue, Big Questions about Voting Machines, R C, Mar. 18, 2009, available at
http://www.connectionnewspapers.com/article.asp?article=326929&paper=71&cat=104.
497. Marsha Walton, Memory Cards Cause Problems in Fairfax County, VA, CNN, Nov. 4, 2008, http://politicalticker.blogs.
cnn.com/2008/11/04/vafairfax-county-machine-problems/.
Brennan Center for Justice | 123
498. Lauren Glendenning, Voting Glitch in Fairfax, C N (Nov. 8,2 006).
499. David Cho, Fairfax Judge Orders Logs of Voting Machines Inspected, W P (Nov. 6, 2003).
500. David Wickert, Slow Going For Local Counting, N T., Nov. 6, 2008, at A1.
501. Scattered Reports of Voters Being Blocked and Machine Malfunctions, K5. (Nov. 2, 2004).
502. Paul Fallon, Software Glitch Delays Vote Count, T W.V., Nov. 5, 2008, available at http://www.timeswv.com/
todaystopnews/local_story_310033002.html.
503. Glitch Stops Vote Counting in Marion, WHSV., Nov 4, 2008, http://www.whsv.com/westvirginiaap/
headlines/33863909.html.
504. Paul J. Nyden, ES&S Official Left Company in May, C G, Oct. 26, 2008, at 1A.
505. Paul J. Nyden, More Voting Problems Reported, C G, Oct. 23, 2008, at 1A.
506. Joselyn King, Voting is a Touchy Business, W N R, Oct. 21, 2008, available at
http://www.news-register.net/page/content.detail/id/515728.html.
507. Paul J. Nyden, Some Early Voters Miffed Jackson County Touch-Screens Switched Votes, 3 Residents Say, C
G, Oct. 18, 2008, at 1A.
508. Paul J. Nyden, Early-Voting Problems in Putnam, C G, Oct. 18 2008, at 1A.
509. Christian Alexandersen, Printers Slow Voting Machines, H H-D, May 21, 2008, available at
http://www.herald-dispatch.com/homepage/x642013931.
510. Voting Machine Problem Causes Delay, G, Feb. 18, 2009, available at http://gazettextra.com/news/2009/feb/18/
voting-machine-problem-causes-delay/.
511. Jake Rigdon, About 600 Medford Ballots Cast in November Ignored, M N-H (Mar. 12, 2004).
512. Matthew Van Dusen, Clerk Changes Election Vote Totals, C S-T (Aug. 21, 2004).
NEW AND FORTHCOMING BRENNAN CENTER PUBLICATIONS
Renewing Democracy After Citizens United
Susan Liss, Michael Waldman
Small Donor Matching Funds: e NYC Election Experience
Susan Liss, Angela Migally
New Politics of Judicial Elections, 2000-2009: Decade of Change
James Sample, Adam Skaggs, Jonathan Blitzer
e Filibuster: Its Use and Abuse
Mimi Marziani
Voter Registration in a Digital Age
Christopher Ponoroff, edited by Wendy Weiser
User Fees for Criminal Justice?
Rebekah Diller
Buying Justice: e Impact of Citizens United on Judicial Elections
Adam Skaggs
A Citizens Guide to Redistricting, Revised and Updated: 2010
Justin Levitt
Domestic Surveillance: New Rules, New Risks
Emily Berman
Domestic Counter-Terrorism and Radicalization
Faiza Patel
Corporate Campaign Spending: Giving Shareholders a Voice
Ciara Torres-Spelliscy
Foreclosures: A Crisis in Legal Representation
Melanca Clark with Maggie Barron
Racial Disparities in Federal Prosecutions
Brennan Center for Justice &
the National Institute on Law and Equity
For more information, please visit www.brennancenter.org
brennan
center
for justice
At New York University School of Law
161 Avenue of the Americas
12th Floor
New York, NY 10013
www.brennancenter.org