IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLORADO
Civil Case No. __________________
WILLIAM R. SINGER, on behalf
of the United States of America,
FILED UNDER SEAL
JURY TRIAL DEMANDED
HART INTERCIVIC, INC.,
QUI TAM COMPLAINT
RELATOR WILLIAM R. SINGER brings this qui tam action in the name of the
United States of America, by and through his undersigned attorneys Michael S. Burg,
Peter W. Burg, and David P. Hersh, and alleges as follows.
1. This is an action by qui tam Relator William Singer, on behalf of the United States,
against Defendant Hart InterCivic, Inc. (“Hart”) to recover penalties and damages arising
from false statements Hart made regarding the accuracy, testing, reliability, and security
of its voting system, in an effort to secure federal monies allocated to the states pursuant
to the Help America Vote Act of 2002 (“HAVA), 42 U.S.C. §§ 15301
et seq. Mr. Singer
worked as a computer technician for Hart from 2001 through early 2004 (at times as
Hart’s only computer technician nationwide). As such Mr. Singer was privy to intimate
details concerning the hardware and software comprising the electronic voting system that
Hart attempted to certify and market throughout the nation. Mr. Singer frequently
accompanied Hart representatives to perform demonstrations, testing, and support
maintenance of the machines in various locations, and thus heard firsthand a number of
misstatements made by Hart in its attempts to win voting system contracts, as well as
misstatements made to conceal the voting machines’ frailties and vulnerabilities. In
January 2004, Mr. Singer resigned from Hart under protest, citing many of the fraudulent
acts and misrepresentations giving rise to this action. In July 2004, Mr. Singer wrote the
Secretaries of State for the States of Texas and Ohio, to alert them to Hart’s misconduct.
He received no substantive response. Mr. Singer provided discrete bits of information to
the press in hopes of attracting attention to Hart’s misconduct. Having “accomplished
nothing” in Mr. Singer’s words, he decided to seek legal redress.
2. Relator William Singer is a citizen of the State of Texas.
3. Defendant Hart InterCivic, Inc. is a foreign corporation that has since 2000 been
registered to do business in Colorado (ID 20001002610). Its principal office is in Texas.
It maintains offices, including a research and development office, in Lafayette, Colorado.
It routinely conducts business and accepts correspondence at its Lafayette address.
4. Hart Intercivic, Inc., operates in Colorado, and therefore within this district, under
the trade names Hart InterCivic (ID 20001195122); Hart Election Services (ID
20011001217); Paragon Systems Group (ID 20001064111); Hart Information Services,
Inc. (ID 20001195121); Hart Information Services (ID 20001195127); C.F. Hoeckel
Blank Book & Lithographing Company, Inc. (ID 20001064109); and C.F. Hoeckel Co.,
Inc. (ID 20001064110).
5. Hart may be served with process of this Court through its registered agent: Neil
McClure, Hart InterCivic, Inc., 1650 Coal Creek Drive, Suite E, Lafayette, CO 80026.
JURISDICTION & VENUE
6. This action arises under the False Claims Act, 31 U.S.C. §§ 3729
7. This Court maintains subject matter jurisdiction over this action pursuant 31 U.S.C.
§ 3732(a) (False Claims Act) and 28 U.S.C. § 1331 (Federal Question).
8. Venue is proper in this Court pursuant to 31 U.S.C. § 3732(a) because: (i) Hart
resides in this district; (ii) Hart transacts business in this district and did so at all times
relevant to this complaint; and, as averred below, (iii) Hart committed acts proscribed by
28 U.S.C. § 3729—acts giving rise to this action—within this district.
9. Before filing this complaint, Mr. Singer served a copy of same upon the United
States, together with a written disclosure statement setting forth and enclosing all material
evidence and information he possesses, pursuant to the requirements of 31 U.S.C. §
10. Mr. Singer has complied with all other conditions precedent to bringing this
11. Mr. William Singer is the original source of, and has direct and independent
knowledge of, all publicly disclosed information on which any allegations herein might
be deemed based, and has voluntarily provided such information to the Government
before filing this action. Specific disclosures include: (i) letters to the Secretaries of State
for Ohio and Texas in July 2004; (ii) subsequent letters to the AGs for Ohio and Texas;
(iii) disclosures to David Allen of blackboxvoting.com; (iv) an interview with Wired
Magazine (never published); (v) disclosures to “Brett” of VelvetRevolution, who referred
Mr. Singer to Brad Friedman of BradBlog; (vi) disclosures to BradBlog; (vii) interviews
for an uncompleted and yet-to-be aired documentary; (viii) disclosures to ABC News
(never published or aired); (ix) a letter to Congressman John Conyers; (x) an interview
with, and open letter to, the Ft. Worth Star-Telegram, and (xi) republication on numerous
websites of the original BradBlog and blackboxvoting disclosures. (Composite Ex. 1,
Copies of All Known Available Disclosures).
The Help America Vote Act.
12. In 2002 HAVA became law. 42 U.S.C. §§ 15301
et seq. HAVA initially
provided federal funding to states conditioned upon the states’ certification that the
federal funds would be used to improve elections, including the improvement and
replacement of existing voting systems. 42 U.S.C. § 15301(a)-(b). The initial
appropriation of so-called “early out money” was for $650 million. 42 U.S.C. § 15304(a).
13. One-half of the “early out money” was appropriated specifically for replacement
of punch card and lever voting systems. 42 U.S.C. § 15302, 15304(a)(2). To qualify for
federal funds, the newer machines—including optical scanning and direct recording
electronic systems—were and remain subject to a number of minimum functional
requirements including the capacity for voters to verify and correct errors in their
selections before casting their ballots; privacy and ballot confidentiality; accessibility
features that enable disabled voters to vote privately and independently and that enable
voting in alternative languages; creation of a permanent paper record for audit purposes;
and maximum permissible error rates. 42 U.S.C. § 15481. In short, HAVA mandates
reliability, security, accuracy, verifiability, and accessibility of voting machines. Failure
of a state to comply with federal requirements by given deadlines would oblige the state
to repay federal monies. 42 U.S.C. § 15302(d).
14. HAVA requires each participating state to create a “state plan” setting forth,
among other things, how the state will distribute and monitor federal funds, what criteria
will apply to determine the eligibility of local governments and other entities to receive
the funds, and how the state will adopt voting system guidelines consistent with HAVA’s
minimum requirements. 42 U.S.C. § 15404(a). Each state must establish a separate
election fund for receipt and expenditure of federal monies received pursuant to HAVA.
42 U.S.C. § 15404(b).
15. In addition to “early out money,” HAVA also provides for additional, ongoing
payments to states (known as “requirements payments”). A state qualifies for
“requirements payments” so long as the chief state election official files written
confirmation that the state complies with HAVA’s minimum requirements, including the
filing of a state plan as described above. The state plan must explain how the state will
adopt guidelines consistent with HAVA’s minimum requirements found at 42 U.S.C. §
15481 and described in part at ¶ 13 above. 42 U.S.C. § 15403(a)-(b), 15404(a)(4).
16. In addition to the $650 million dollars in “early out money” appropriated under
HAVA, the Act authorized federal payments to the states totaling $3 billion in public
“requirements payments” during FY 2003-2005, to remain available without fiscal year
limitation until expended. 42 U.S.C. § 15407. HAVA likewise authorized the
expenditure of an additional $100 million during FY 2003-2005 to carry out the
provisions of the Act mandating access to polling places for voters with disabilities, to
remain available without fiscal year limitation until expended. 42 U.S.C. § 15424.
Additional federal monies were allocated for research grants relating to voting technology
($20 million in FY 2003), 42 U.S.C. § 15443; testing of new voting technologies ($10
million in FY 2003), 42 U.S.C. § 15453; payments for disabled protection and advocacy
systems ($60 million during FY 2003-2006 and as-needed thereafter), 42 U.S.C. § 15462;
and funds for mock elections ($200,000 in FY 2003 and as-needed thereafter), 42 U.S.C.
Voting Systems Certification
17. HAVA also established the Election Assistance Commission (“EAC”). 42 U.S.C.
§ U.S.C. § 15321. The EAC is charged with overseeing the development of voluntary
voting system guidelines and the promulgation of rules governing the testing,
certification, decertification, and recertification of voting system hardware and software.
42 U.S.C. §§ 15322, 15342, 15371. The EAC also is charged with promulgating
standards ensuring the security of voting systems. 42 U.S.C. § 15361(e)(2).
18. Under HAVA’s statutory scheme, the EAC is authorized to accredit independent,
non-federal laboratories to carry out the testing, certification, decertification, and
recertification of voting systems. These labs commonly are referred to as Independent
Test Authorities (“ITAs”). States may provide for testing, certification, decertification,
and recertification of voting systems by the ITAs. 42 U.S.C. § 15371.
19. The EAC has effectively delegated oversight of federal voting systems standards
certification to the National Association of State Election Directors (“NASED”), which
has approved certain ITAs (though none is yet certified by the EAC) to certify voting
system compliance with federal standards, which is indicated by issuance of a
certification number for each certified model.
20. NASED qualifies only specific models. In NASED’s words, “Models which are
not identical are not NASED qualified. NASED warns purchasers of voting machines
“Make sure you know, in writing, from the vendor what has and has not been tested. Be
sure to be specific — and make the vendor be specific — about the version and revision of
software. Make them put in writing that the version they are selling you is “identical” to
the version which was tested and NASED qualified.” (Publicly available at: http://www.
nased.org/ITA%20Information/NASEDApprovedSystems1.03.pdf) (emphasis in
21. Voting machine vendors submit their voting systems and system modifications for
testing under the NASED program. Testing is paid for by the vendors and is performed in
secret by the ITAs. The ITAs’ detailed reports are deemed confidential and are not
publicly disclosed (though some have been disclosed to the press and some produced in
response to public records requests after vendors have given copies to elections officials).
Only the ITAs’ and NASED’s final decision on qualification is made public.
22. Pursuant to 42 U.S.C. § 15362(e), the voting system standards (“VSS”) most
recently adopted by the Federal Election Commission (“FEC”) were deemed adopted by
the EAC upon passage of HAVA. The most recently adopted VSS were the 2002 VSS,
which were adopted by the FEC on April 29, 2002; however, the EAC (through NASED)
inexplicably continued to qualify voting systems based upon the 1990 VSS and vendors
were not required to comply with the 2002 VSS until January 1, 2006. Hence, during
most of the time period relevant to this complaint (2001-2004), the 2002 standards were
in effect as a matter of federal law, but the 1990 standards were utilized in testing and
qualifying voting systems.
23. Each state maintains its own standards for certification of machines. A state may
or may not adopt the 1990 or 2002 VSS. In no event may a state employ standards less
stringent than the minimum requirements, distinct from those in the 1990 and 2002 VSS,
mandated by 42 U.S.C. § 15481 (referenced in ¶ 13 above).
24. Vendors such as Defendant Hart engage in a highly competitive bidding process,
state-by-state, and often county-by-county, making multiple sales presentations and
deploying their respective voting systems in mock and actual elections. Every
representation made by a given vendor to state and local government officials during the
course of this process is made in an effort to obtain payment of federal (HAVA) funds.
25. Accordingly, every knowing misrepresentation by a vendor to a state or local
governmental official regarding the testing, accuracy, reliability, security, accessibility,
verifiability, or legal compliance of a vendor’s voting system, or any part of it, is a false
“claim” in violation of the False Claims Act. Any post-certification alteration of a voting
system’s software or hardware without recertification would effectively create a different
(and uncertified) system distinct from the system originally considered and purchased by
a given customer.
26. A substantial share of HAVA funds has been directed to voting system vendors
such as Hart, both for the voting systems themselves and for the vendors’ technical and
27. Since the passage of HAVA, virtually every purchase by any state or local entity
of any electronic voting system was made in whole or in part with federal funds.
28. Vendors were instrumental in the passage of HAVA and all were aware that the
testing, certification, and purchase of any electronic voting system necessarily involved
the payment of federal funds.
29. The vast majority of state and local governments employ electronic voting
machines manufactured by one of four major vendors—Diebold, ES&S, Sequoia, and/or
30. Electronic voting systems generally fall into one of two types—optical scan and
direct recording electronic (“DRE”).
31. Optical scan technology (a/k/a marksense) typically uses a “bubble sheet” that is
filled in by the voter and then is fed into a scanner, which counts the votes. HAVA
requires use of scanners capable of informing a voter that he or she has “overvoted,” i.e.,
selected more than the permitted number of candidates for a given office, and allowing
the voter to correct this mistake before the scanner counts the votes. For absentee
marksense ballots, HAVA permits use of a central scanner without this capability.
32. On a DRE, the voter typically enters votes using a touchscreen or, for some
disabled voters, an audio or other alternative interface. The votes are recorded
electronically. The voter cannot verify that the electronic ballot mirrors the ballot
displayed on the computer screen. The machine ordinarily stores the voting results data
in a removable memory component. Hart’s eSlate voting system is a DRE.
33. The integrity of electronic voting systems is a matter of paramount public
importance. If voting machines are unreliable, the entire democratic process is thrown
into doubt. And serious errors have occurred. In March of this year during a real-world
primary, a glitch in Hart’s voting system created 100,000 phantom votes (increasing the
voter turnout from 58,000 to more than 158,000). During the same election Hart units in
Galveston County, Texas, caused delayed vote counts and suffered other malfunctions,
forcing voters to resort to emergency paper ballots. In Tom Green County, Texas, during
the same primary election, Hart machines were accused of “not printing ballots for every
vote cast on the machines,” causing the suspension of a recount. Since 2003, Hart voting
systems have been publicly cited for errors in real-world elections including, among many
others, in Harris County, Texas (voters told to come back later when machines repaired);
in Richmond, Virginia (voting machine “fried”); in Orange County, California (machines
broken, “[h]undreds of voters turned away from the polls” and “more ballots cast than
registered voters” and “at least 5,500 voters had their ballots tabulated for the wrong
precincts” while “an exact account of miscast ballots [was] impossible”); in Honolulu,
Hawaii (“machine flashed a low battery indicator after [voter] selected his party [then]
flashed a ‘cast ballot’ indicator and his ballot was accepted before he was able to vote for
anyone” and “a couple of dozen Green Party ballots were recorded, even though there
were no candidates”); in Tarrant County, Texas (“poll workers fielded numerous
complaints from voters . . . .”); and in Boulder County, Colorado (“the system could not
process information for an individual race [and] the second time through, the system
found more damaged races on the same ballot and read another race as an ‘undervote’”).
(Composite Ex. 2, Selected Media Stories re Hart Election Failures).
Hart’s eSlate System
34. Hart’s electronic (DRE) voting system is known as eSlate. The eSlate system is
comprised of several distinct components.
35. The Ballot Origination Software System (“BOSS”) is used by election officials to
define and create individualized electronic ballots. Data is entered once into BOSS and
then flows through all components of the eSlate system.
36. The Mobile Ballot Box (“MBB”) is a reusable, portable flash memory card. It is
used to store and transfer election information. When inserted into the Judge’s Booth
Controller (see ¶ 37), the MBB supplies election information and ballot styles, and it
stores cast votes. Once voting has concluded the MBB is removed and its contents are
tallied by the Tally software (see ¶ 39).
37. The Judge’s Booth Controller (“JBC”) is the “brain” that manages the system,
enabling poll workers to know which voting booths are in use at any time. The JBC
issues access codes for the voters’ use. It can control up to 12 daisy-chained eSlate units.
38. eSlate refers to the system generally and to the device that voters use to cast
ballots, unless using paper ballots, in which case Ballot Now is used—see ¶ 42. (Ex. 3,
Taylor County, Texas, eSlate Voting Instructions). The eSlate may be equipped with a
disabled access unit (“DAU”) for use by disabled voters. The eSlate units are physically
connected to the JBC, which stores cast-vote records.
39. Tally is a software application that reads, stores, and tabulates the cast-vote
records from the MBB (the portable flash card that transferred the cast-vote records from
the JBC). Tally tabulates all early voting, absentee, and election day results, and produces
40. Rally is a software application that is capable of reading, storing, and transferring
cast-vote data from polling places or collection centers with respect to early returns.
41. The System for Election Records and Verification Operations (“SERVO”)
software is an election records archiving and asset management system. SERVO is
designed to recover data from equipment in the case of a lost or damaged MBB. SERVO
also is designed for various recount purposes.
42. Ballot Now is a digital-scan paper ballot system that manages the printing,
scanning, and resolution of mailed-in paper ballots. It also records the electronic castvote
records to an MBB to be read and tabulated with Tally.
43. In a typical election, election officials create ballots using BOSS. The ballot
definitions are fed by the JBC to the eSlates attached to it (including at least one DAU
needed to comply with the HAVA disability access requirement). A “zero tape” is run on
the JBC to ensure no votes reside in the system. When voters arrive they are given a
JBC-generated PIN number that they enter on the eSlate. They then vote and the votes
are stored on the eSlate, MBB, and JBC (though not in identical form, as Hart claims).
The cast-vote records stored on the MBB (itself located inside the JBC) are removed and
loaded into Tally by physical transfer of the MBB card. A day-end report also is printed
using the JBC. (See generally Ex. 4, Election Flowchart & System Interface Graphic from
2003 Compuware Review of eSlate System & Hart’s eSlate Architecture Graphic).
44. On or about November 1, 2001, NASED qualified (using the 1990 VSS) BOSS
versions 2.0, 2.2, and 184.108.40.206; Tally versions 2.1, 2.3, and 2.6.10; and Ballot Now version
1.3.11. (Ex. 5, NASED List of Qualified Systems, 1/3/03).
45. In or about March 2002, NASED qualified (using the 1990 VSS) the eSlate 3000
DRE Precinct, including BOSS version 2.08.8 and Tally version 2.8.21. (Ex. 5).
46. In or about May 2002, NASED qualified (using the 1990 VSS) the eSlate 3000
DRE Precinct Voting Machine version 1.15.07, including JBC version 1.15.08. (Ex. 5).
47. By 2003, some version of eSlate was certified in at least 20 states, including
Colorado. (Ex. 6, Internal Hart Spreadsheet Indicating Certification Status).
Relator William Singer
48. Mr. Singer worked as a technical services specialist, i.e., a computer technician,
for Defendant Hart from November 26, 2001, until January 5, 2004.
49. Attached as Exhibit 7 is an electronic copy of Mr. Singer’s annual Performance
Review dated November 25, 2002. The review concluded that Mr. Singer was “highly
effective” in several fields, including “Subject Matter Expertise” and “Job Knowledge” in
his role as a technical services specialist. Mr. Singer received an “Overall Rating” of
“highly effective,” and he was lauded for his “tireless dedication and willing spirit.” The
review deemed “Job Knowledge” Mr. Singer’s “strongest attribute” given his “high level
of expertise” and “in-depth knowledge and technical skills.”
50. At the time Mr. Singer was hired, he was one of three computer technicians on
Hart’s staff. After approximately one year the computer technician staff was reduced to
Mr. Singer and one other technician. For a period of several months during 2003, Mr.
Singer was the only computer technician at Hart.
51. As a technical specialist, Mr. Singer’s responsibilities included providing tech
support for all Hart voting systems hardware and software; setting up and transporting
purchased voting machines; setting up voting machines for sales demonstrations,
including setup and monitoring of machines during actual elections; testing the computer
equipment; and preparing the system components for certification tests performed by
outside entities. The details of his work included inventory control; demonstrations;
database storage; eSlate configuration; training of other employees on the voting systems;
serving account managers—including assistance with election procedures and hardware
issues; gathering system specifications and build quotes for customers; buying and
shipping equipment; tracking changes in the eSlate system and communicating them to
customers; and general research of new hardware, software, and utilities. Mr. Singer was
at various times responsible for tracking, testing, and communicating to account
managers, salespeople, trainers, and the like, any changes to the eSlate system.
52. Mr. Singer accompanied Hart representatives on sales presentations throughout
the United States and has personal knowledge of the sales methods and messages
routinely employed by Hart. Sales personnel regularly sought Mr. Singer’s advice on
how best to translate technical issues into common parlance for sales presentations.
53. Mr. Singer helped to maintain the salespersons’ “personal” demonstration models.
54. Mr. Singer frequently traveled to provide technical support during actual
elections. He traveled on one occasion to provide technical support during a state
55. The voting system often was used in actual elections as a part of product
marketing, so that Mr. Singer’s participation simultaneously supported real-world
elections and system sales.
56. During the course of his employment Mr. Singer repeatedly voiced concerns to
his supervisor, Travis Harrell, regarding acts and omissions by Hart that Mr. Singer
deemed deceptive. Mr. Singer was reassured on such occasions that he did not have the
“big picture” or “all of the facts,” and that if he had, he would not be concerned.
57. These responses quelled Mr. Singer’s concerns until approximately the time he
became the only technical specialist at Hart. He witnessed acts, omissions, and
misrepresentations that could not be explained away as anything other than false and
fraudulent conduct, and which confirmed his prior suspicions of same.
58. In January 2004, Mr. Singer resigned from the company, “[d]ue primarily to
unresolved ethical differences” with the company. (Ex. 8, Singer Resignation Letter). He
cited specific acts of misconduct and “fraud” and closed with his hope that “this letter has
some positive impact in creating change . . . so that I am the last person that has to leave
for the sake of their conscience.” (Ex 8).
59. After departing from Hart, Mr. Singer sent an email to his closest associates at
Hart explaining that Hart had been spreading misinformation about him, advising them
that if they remained at Hart and supported the company, he could no longer protect their
names in any disclosures he might make about the company. (Ex. 9, Singer Email to Hart
60. In July 2004, Mr. Singer wrote to the Secretaries of State for the States of Texas
and Ohio advising them of Hart’s misdeeds in hopes that this would prompt an official
investigation into the company’s practices. Mr. Singer subsequently disclosed those
letters to the press, including to BlackBoxVoting.com and, later, to BradBlog.com—a
web log known for its coverage of election-related issues. (Composite Ex. 1, Copies of
All Known Available Disclosures).
61. Mr. Singer engaged counsel only after exhausting these other avenues.
Background Facts & Hart’s False Claims
62. During Mr. Singer’s tenure with Hart (November 2001-January 2004), he
acquired firsthand knowledge of the false claims described below. Each described act or
omission was knowingly committed by Hart in an attempt to secure federal funds through
purchase of its voting systems or was knowingly committed by Hart in an attempt to
retain federal funds already paid it and to avoid disgorgement of those funds.
Failure to Test Software Properly. Most fundamentally, Mr. Singer has firsthand
knowledge that Hart routinely failed to adequately test its software. Specifically, it failed
to completely alpha test its software and did not beta test it at all. Alpha and beta testing
are standard practices within the industry and represent the first and second groups, or
waves, of comprehensive tests for all software produced professionally. Alpha and beta
testing are designed to catch major and minor flaws, respectively. Frequently beta testing
itself involves several stages, some of which may be external to the company producing
the software. Hart had no beta test group, had no one tasked with performing beta tests,
and conducted no external beta testing. (Ex. 10, Singer Email re “Top Ten” Product
Improvement Suggestions, 11/26/02) (“6. Create Beta Test (internal) group”). On
separate occasions, Hart’s alpha tester Seiji Okamoto and its BOSS programmer Bill
Barstad physically showed Mr. Singer stacks of “incomplete” testing documents relating
to products that had been sent for ITA testing. Hart’s failure to beta test its voting system
renders false or fraudulent every representation Hart made to potential purchasers
regarding the system’s reliability, security, accuracy, or compliance with legal
requirements, and every claim for payment relating to purchases made in reliance upon
Faked Certification Tests. Mr. Singer has firsthand knowledge that Hart created a
“dummy” machine to obtain system certification in at least one state—Ohio. At the time
of the testing (mid-2003), Mr. Singer was the only person at Hart who performed system
setups for customers and for virtually all demonstrations. As such he maintained current
corporate documentation (as well as personal documentation) of Hart’s system
configurations, and was the only person at Hart who knew where current configuration
documentation was stored. Configurations may be thought of as technical descriptions
needed to build, test, accept, operate, install, maintain, and support a system. A
“configuration” describes the physical combination of hardware components that make up
a system and the software settings that allow various hardware components of a computer
system to communicate with one another as desired. Analogizing to a chemical
compound, even the slightest change in configuration results in a distinct system—H
(water) is quite different from H
2O2 (hydrogen peroxide). It was impossible for Hart to
set up the Ohio certification system according to its standard customer configurations
without consulting Mr. Singer. Nevertheless, Hart used an alpha tester (Seiji Okamoto) to
set up the “demo” system to be tested in Ohio. Mr. Singer knew that the standard
configurations could not have passed certification tests. Mr. Singer confronted his
supervisor, Travis Harrell, explaining that Hart was making misrepresentations during the
state certification test by holding out the dummy system as its standard system without
altering the standard system to conform to the tested configurations after-the-fact. He
received no explanation why the programming team was used for the demo setup and why
non-standard configurations were employed. Hart knowingly misrepresented the dummy
system as its standard system in a successful effort to obtain system certification, enabling
Hart to pursue and procure federal funds from the sale of its voting system.
Dummy Reports to Election Officials. Mr. Singer has firsthand knowledge that
Hart created at least one dummy report during the Ohio certification process.
Specifically, the Ohio Secretary of State’s office asked whether the eSlate system could
generate a specific form of report (an “election abstract”). It could not. Because this
would affect and possibly preclude certification, a Hart representative (Sheri Charleston)
physically created a mock-up or dummy document that appeared as if it were an eSlategenerated
document demonstrating the system’s ability to produce an election abstract.
Mr. Singer knew that eSlate could not produce such a report. Hart knowingly
misrepresented the dummy report as a standard report within its system in a successful
effort to obtain system certification, enabling Hart to pursue and procure federal funds
from the sale of its voting system.
Dummy Reports to InfoSentry. InfoSentry Services, Inc., is an independent
testing service that was hired by the State of Ohio in 2003 to perform a security
assessment of all computer voting systems and vendors in Ohio. Based upon statements
in that report, Mr. Singer reasonably infers that Hart manufactured dummy reports
regarding a disaster plan and security audits, and that Hart submitted those reports for
review by InfoSentry to give the false appearance of Hart’s (and, by implication, eSlate’s)
true capabilities. Valid reports of this sort, as described in the InfoSentry report, (Ex. 11),
could not have been generated without consulting the technical team, and there was no
technical consultation. Hart knowingly presented the dummy reports in a successful
effort to obtain system certification, enabling Hart to pursue and procure federal funds
from the sale of its voting system.
Semi-Random Vote Storage. Based upon Mr. Singer’s attendance at sales
representations and based upon Hart’s promotional literature he encountered, Mr. Singer
has firsthand knowledge that Hart presented its voting system as one that stored votes in a
random manner, suggesting complete randomization. Based upon Mr. Singer’s personal
knowledge of the system’s vote storage, he knows this to be untrue, as did Hart
principals. Specifically, the eSlate system stores votes in a sequentially random format
that is not truly random. (Ex. 12, Singer Email re CVR Reports, 11/1/02). Because the
vote order changes with the system’s redundancy (recording votes simultaneously in
various locations), the system could be exploited with the machine timestamp and identity
to determine with some accuracy how a specific person voted. The system could have
been designed so that votes were completely randomized. Mr. Singer alerted his
supervisor, Travis Harrell, to this vulnerability. Hart knowingly misrepresented the
randomization of vote storage to make the system appear secure and compliant with
HAVA and VSS standards in a successful effort to obtain system certification, enabling
Hart to pursue and procure federal funds from the sale of its voting system.
Audit Trails. While generating a sample of each report the eSlate system could
produce, Mr. Singer discovered that the system’s BOSS software audit trail had invalid
entries, thus rendering every BOSS audit report false. An audit trail is a security feature
that tracks everyone who has logged in to the system and records every modification
made to it. BOSS’s audit trail had invalid entries. Mr. Singer alerted the company to this
security flaw. It was corrected in some jurisdictions where Hart had customers, but not in
others. The fix was never certified and Hart concealed the fix from all customers.
Thereafter Hart routinely knowingly concealed this security flaw and thus misrepresented
the security of its voting system in a successful effort to obtain system certification,
enabling Hart to pursue and procure federal funds from the sale of its voting system.
Corrupt M2B3 Devices. Hart marketed a memory card reading device called an
M2B3, claiming that it was faster and less prone to corrupt MBB data than were the card
readers Hart previously sold. However, internal testing revealed that the M2B3 was not
faster, nor did it entirely resolve the data corruption issues of concern. The results of this
testing were concealed from past and prospective purchasers. Hart knowingly concealed
this system flaw from purchasers until it was encountered, as occurred in Tarrant and
Harris Counties, Texas. Hart continuously misrepresented M2B3 capabilities when
selling version 2.1 to subsequent purchasers. These misrepresentations were made with
the intent to avoid any sort of repayment of federal funds already spent on the M2B3
devices and with the successful intent to sell voting systems, which were purchased with
System 3 Redundancy. Toward the end of Mr. Singer’s tenure with Hart, the
company began marketing System 3—a new version of eSlate. Hart was anxious to have
System 3 perform in an actual election so that the company could tout that fact in
subsequent sales presentations. Hart’s standard sales presentations and promotional
materials contained misstatements regarding the system. Specifically, Hart knowingly
misrepresented System 3 as capable of triple redundancy (meaning that cast votes are
stored in at least three separate places). Earlier eSlate versions were touted as tripleredundant
in part because lost votes could be recovered through a program called JBC
Recovery (a program not available for customer use). (Ex. 13, Hart Instruction Sheet
“Using JBC Recovery”). However, the System 3 programming team did not develop a
counterpart utility suite to JBC Recovery that might be used to recover lost votes in the
event of a system failure. Hence, System 3 lacked the promised redundancy. Hart also
concealed the fact that the system had not been adequately tested. Hart knowingly
misrepresented the reliability and functionality of eSlate System 3 in a successful effort to
sell voting systems that were purchased with federal funds. These voting systems were
used during real-world elections and that fact later was touted by Hart in sales
presentations that succeeded in selling even more voting systems, which also were
purchased with federal funds.
Disabled Voting. One of HAVA’s primary goals was to ensure accurate, private
and independent voting for disabled citizens. Hart designed special eSlates (disabled
voting units, or “DAUs”) that could be disconnected from the daisy-chain of eSlates in a
polling place and brought to a disabled voter. DAUs were particularly prone to lose votes
due to system design flaws that were well known within Hart (and later admitted to
customers once the problems became impossible to conceal). Based upon Mr. Singer’s
firsthand knowledge, Hart sales presentations concealed these design flaws and the
DAUs’ tendency to lose votes. Moreover, Hart knowingly misrepresented its ability to
consistently rebuild vote records from corrupted eSlates, including DAUs. The devices
had to be shipped to the computer programming team in Lafayette, Colorado, where the
votes were secretly rebuilt. Hart’s programming team refused (and refuses) to reveal its
method of rebuilding votes, its success or failure rate, or the percentage of votes actually
recovered. At the end of this secretive process, the programming team merely provides
an MBB with cast-vote records and nothing more. Hart knowingly misrepresented the
accuracy and reliability of its DAUs and Hart knowingly concealed the system’s flaws. It
did so with the successful intent to sell these systems, which were purchased with federal
Unconfigured Systems. Mr. Singer has firsthand knowledge that in Harris
County, Texas, Hart furnished voting systems that were unconfigured (and which
therefore varied from the systems presented for purchase during sales demonstrations),
thus permitting/requiring election officials to configure the systems in ways that
inevitably varied from the standards that governed the voting systems demonstrated for
purchase. Furnishing unconfigured machines gave rise to glaring security vulnerabilities
(in violation of HAVA and the VSS). Hart knowingly furnished these unconfigured
systems for the purpose of sustaining its account with Harris County, thereby ensuring the
future purchase of its systems and support with federal funds, and thereby retaining
federal funds already paid on the premise that Harris County would receive a compliant
voting system rather than the non-compliant system it actually received.
Concealed SERVO Problems. Mr. Singer has firsthand knowledge that the eSlate
warehousing and recount application, SERVO, had major ongoing problems with
corrupting data. Hart knew of this flaw in its system but concealed it from prospective
and past purchasers of its system. Hart frequently provided untested versions of SERVO
that corrupted vote information and destroyed the “triple redundancy” that Hart touted in
its sales presentations. In September 2002, Mr. Singer’s supervisor, Travis Harrell,
circulated a draft set of release notes for System 2.1. The release notes included “Known
Defects (defects not fixed in this release.)” First among those for SERVO was “Cast
Vote Record data corruption” as well as several defects with SERVO’s internal audit
log.” (Ex. 14, Draft System 2.1 Release Notes at page 17 of 18 [sic]). Hart knowingly
misrepresented the system’s redundancy and general reliability, and knowingly concealed
SERVO corruption issues, in a successful effort to sell voting systems that were
purchased with federal funds, and in a successful effort to avoid being required to repay
federal funds already paid to it for faulty systems.
Clandestine Updates to Tally. Mr. Singer has firsthand knowledge that in
Charlottesville, Virginia, the Tally software suffered from a bug, and was updated to a
newer version without the customer’s knowledge. Mr. Singer, under protest, assisted
with the update by supplying the software and assisting Hart employee Sheri Charleston
with the update by phone. Hart knowingly misrepresented the system’s capability to
voting officials, and knowingly concealed the Tally corruption and update, in a successful
effort to sell voting systems that were purchased with federal funds, and in a successful
effort to avoid being required to repay federal funds already paid to it for faulty systems.
SERVO Recount Inadequacies. Hart’s SERVO utility was designed in part to
assist in vote recounts. Mr. Singer has firsthand knowledge that SERVO suffered from
problems that compromised any potential recount, most notably its inability to count
blank votes. (Ex. 12, Singer Email re CVR Reports, 11/1/02) (“NOTE THAT VOTES 2
AND 4 WERE BLANK BALLOTS, these don’t appear on the servo reports.”). In some
elections, ascertaining the total number of ballots cast is critical, e.g., where passage of a
ballot measure turns on whether more than 50% of the ballots, as opposed to votes, cast
on the measure contained a “yes” vote. Hart knowingly concealed these issues during all
sales presentations and on every occasion its system was purchased with federal funds.
“Used” eSlates. In Tarrant County, Texas, and elsewhere, Hart temporarily
provided extra eSlates and JBC units when a customer jurisdiction’s needs were higher
than expected. The additional units had higher failure rates due to defective capacitors.
Hart provided the machines despite its knowledge of this hardware defect. It concealed
the defect. When questioned after-the-fact concerning the higher failure rates, Hart
described the machines as “used” and thus more prone to failure. Hart knowingly
misrepresented the machines’ vulnerabilities and concealed its knowledge of the defective
capacitors in a successful effort to sell voting systems that were purchased with federal
funds, and in a successful effort to avoid being required to repay federal funds already
paid to it for faulty systems.
Specific Knowing Misrepresentations in Colorado Proposal. On June 16, 2003,
Hart transmitted a proposal to the Purchasing Division of Boulder County, Colorado. (Ex.
15, Pertinent Portions of Hart Proposal, 6/16/03). The proposal responded to Boulder
County’s Request for Proposal # 4437-03 relating to voting systems. The proposal
contained numerous misrepresentations and false statements made for the sole and
successful purpose of obtaining the Boulder County voting system contract. Hart won the
Boulder contract and was paid for its systems and support. The payments were made in
whole or in part with federal funds, as Hart knew it would be. Each of Hart’s
representations was made in direct response to Boulder County’s item-by-item request for
information and/or statement of voting system requirements. Hart’s proposal contained
the false representations listed below, which induced Boulder County to expend federal
HAVA money for Hart’s systems and support. Mr. Singer has personal knowledge that
each of the representations listed below was false, as described, and that each
representation was known by Hart to be false at the time Hart made the representation.
77a. At page 3-16 Hart states that the cast-vote record made on a DAU “will be
recorded in the Judge’s Booth Controller and Mobile Ballot Box just like all the other
ballots cast in the polling place.” This is intentionally misleading. As noted in ¶ 71
above, Hart knew that DAUs were subject to exceedingly higher error rates than eSlates.
77b. At page 3-17 Hart claims that the MBB remains the “master record of voting.”
This statement conflicts with Hart’s claims of triple redundancy in the preceding
sentence. In fact vote storage in the eSlate/DAU, JBC, and MBB was not identical.
Instead of providing a backup of stored information, Hart claimed to create multiple
originals, but in the case of discrepancy it would be impossible to determine which is the
“original” or “true” record of votes.
77c. At page 4-1 Hart states that “[t]he eSlate system has been certified by the
National Association of State Election Directors . . . including all ballot formatting, vote
recording, tabluation, and reporting components.” However, the SERVO utility is a
tabulation and reporting component and had not been certified by NASED. Exhibit 5
demonstrates that as of January 2003, SERVO was not certified by NASED and Mr.
Singer personally believes that it was not certified between that time and the June 2003
submission of this document, nor does any NASED document suggest so. The system
certification documentation provided in Hart’s proposal, (Ex. 15, at end), does not
demonstrate NASED approval of Servo. Hart’s own SERVO Operations Manual, (Ex.
16, Pertinent Parts of Servo Ops. Manual), confirms that SERVO is a tabulating and
reporting component of the eSlate system. (“Servo is an election records and recount
management system for the eSlate . . . .” and “SERVO provides: reports on CVRs [castvote
records] . . . . recount data . . . .”). Likewise, Hart’s proposal to Boulder itself
identifies SERVO as a component of the eSlate system. (Ex. 15, Boulder Proposal at 3-3,
3-4). Thus, SERVO was part of the “eSlate System” but was not NASED qualified as
77d. At page 4-2 Hart states that “[t]he eSlate system does not permit overvotes.” In
fact, Ballot Now is part of the “eSlate system,” (Ex. 15, Boulder Proposal at 3-3, 3-4), and
it permits overvotes. (Ex. 17, Internal Confidential “Ballot Now 1.5.09 Survival Guide” at
7) (noting how to resolve overvotes on straight party contests). Additionally, Exhibit 14
(Draft System 2.1 Release Notes) included within “Known Defects (defects not fixed in
this release)” that “[w]hile scanning ballots for the System 2.1 testing, BNIP repored an
overvote in a contest that was clearly marked correctly (autovoted).” (Ex. 14, Draft
System 2.1 Release Notes at page 12 of 13 [sic]).
77e. At page 4-3 Hart states that disabled voters shall have the same “opportunity for
access and participation . . . as for other voters.” However, as Hart knew, the DAUs had
an exceedingly higher lost-vote rate than standard eSlates and resulted in lost votes for
disabled voters as described in ¶ 71 above.
77f. At page 4-3 Hart states that eSlate provides for second languages “including
character-based languages.” In fact eSlate version 2.1 was not capable of providing
character-based languages and previously had suffered numerous problems even with its
dual-language capacity. (See Ex. 14, Draft System 2.1 Release Notes) (noting at least five
“known [unfixed] defects” in the BOSS 2.9 Spanish language feature).
77g. At page 4-3 Hart states that “[t]he eSlate System is fully tested and certified by
NASED.” As demonstrated in ¶ 77c, SERVO was not certified by NASED. Moreover,
NASED certification did not include special patches, updates, and emergency fixes that
Hart knew from past experience would be required once its machines were used in the
real world. These patches, upgrades, and fixes often, but not always, received “no look”
temporary certification from state election officials, but such certification could not
render the systems HAVA- or VSS-compliant.
77h. At page 4-6 Hart states that its system is subject to “regular scheduled release of
hardware, firmware, and software upgrades . . . .” In fact Hart had no “regular
scheduled” releases. Its releases were ad hoc.
77i. At page 4-14 Hart states that “Ballot Now performs with 100% accuracy . . . .”
This statement is untrue and was known to Hart to be untrue when made. Hart repeats
this false statement at page 4-18. In December 2002, Mr. Singer alerted Hart to a realworld
vote inaccuracy that occurred with Ballot Now in Arapahoe County, Colorado. (Ex.
18, Singer Email re Arapahoe Election, 12/13/02) (“One of the provisional ballots which
had previously autoresolved as an undervote was ‘autoresolved’ by Ballot Now as a vote
for Feeley. (that contest did not come up for resolution) We are researching this issue but
it’s likely a mark on the edge of the detection range that BNIP scanned ‘properly’ the
second time.”). Prior to that, in July 2002, Mr. Singer detailed to Hart the problems
experienced in Arapahoe County with respect to Ballot Now version 1.4—the predecessor
version to that being proposed in Boulder. (Ex. 19, Singer Email re Arapahoe Election,
7/10/02) (“I don’t believe we can conduct an election in Arapahoe . . . . Ballot now is
unreliable . . . the database is unstable . . . .”). Mr. Singer’s supervisor Travis Harrell
forwarded these concerns to other Hart employees. (Ex. 20, Harrell Email re Arapahoe
Election, 7/10/02) (“These Ballot Now issues are preventing the County from proceeding
with Logic and Accuracy testing . . . .”). Later, in 2005, Hart admitted to Yakima County,
Washington, that Ballot Now produced inaccurate results (undervotes) because of a line
running through the scanned box images “likely caused by a small foreign object (dirt or
paper debris) . . . .” (Ex. 21, Harrell Email re Yakima Election, 6/15/05). Also, Exhibit
17, Hart’s Confidential Internal “Ballot Now 1.5.09 Survival Guide,” authored by Hart’s
Ballot Now programmer, addresses numerous Ballot Now defects. The “survival guide”
also includes an entire appendix described as a “Ballot Now post-mortem of the Nov
2002 Harris County election,” with sections entitled “Process Issues/Problems” and
“System Issues/Problems.” Finally, Exhibit 14 (the Draft System 2.1 Release Notes)
identifies a number of inaccuracies with Ballot Now, including that it permits overvotes,
and is capable of creating the wrong number of cast-vote records. Hart thus knew that in
its June 2003 report to Boulder the statement that Ballot Now was “100% accurate” was
false. Boulder County would be using Ballot Now 1.5, but Hart disclosed none of this
information in its June 2003 proposal.
77j. At page 4-14 Hart states that “Ballot Now uses commercial off-the-shelf
scanners and printers, and operates at the full rated speed of each.” In fact ballot
processing requires higher resolution than the scanners can achieve while maintaining
77k. At page 4-14 Hart states that Ballot Now’s bar code system “eliminates the need
to sort ballots before scanning, which eases the workload for the Elections Staff and
speeds ballot processing.” Mr. Singer personally witnessed the hand-sorting of ballots
during an election in Arapahoe County, Colorado, more than six months before this
misrepresentation was made by Hart. Mr. Singer reported to Hart the need for handsorting
by way of an “after action” email report dated December 13, 2002. (Ex. 18, Singer
Email re Arapahoe Election, 12/13/02). Likewise, Exhibit 17 (the Ballot Now “Survival
Guide”) identifies as a process issue/problem the counties’ failure to pre-sort ballots,
including checking of serial numbers.
77l. At page 4-16 Hart states that the dimensional stability of paper ballots and
printing accuracy are non-issues because “these artifacts have no impact on accuracy.”
Mr. Singer has personally rescanned the same ballot through Ballot Now at different
times and yielded different results. Hart knew at the time it made its statement that such
artifacts could impact the accuracy of vote recording.
77m. At page 4-19 Hart states that Ballot Now allows scanners to work “continuously
at rated speed.” In fact internal testing demonstrated that scanners could not work
accurately at maximum speeds, particularly when handling large ballots.
77n. At page 4-24 Hart states that another vendor’s optical scan system may be
integrated with eSlate by use of “[t]he eSlate System’s Fusion utility . . . .” But Fusion
was not NASED qualified nor was it certified by the State of Colorado at this time. Thus
it could not have been used. (
Cf. Ex. 22, California Sec’y of State Evaluation, 2/24/06)
(noting that because Fusion has not been NASED/ITA-tested it could not be used absent
state certification; Ex. 15, Colorado Certification Documents, at end).
77o. At page 4-26 Hart states that it is “experienced in the integration of the eSlate
System with other Election Management Information Systems” and cites Harris County,
Texas, issuance of access codes as an example. This representation refers to a system
interface (an interface is the point of interaction between a computer and any outside
entity, such as another computer). Hart’s statement conceals the fact that the interface
could pose serious security risks to the system because it involved connecting outside
systems (with configurations entirely unknown to Hart) directly into the rear of a JBC (the
“brain”), rendering this a classic “back door” interface. Mr. Singer does not have
firsthand information as to what, if any, security measures were in place in this regard,
but he does have firsthand knowledge that the interface did not create any special audit
log information that might alert anyone should an outside program take over the JBC. In
any event, Hart concealed in its Boulder proposal the critical fact that the interface would
involve outside systems and attendant security risks. It thus designed and touted a feature
that it knew could override the audit logs and manipulate the voting system, but concealed
the security vulnerabilities from Boulder County.
77p. At pages 4-35 to 4-36 Hart describes how it ostensibly creates non-character
based and character based languages. The description Hart provides was not possible
with version 2.1, which was the version offered for use (unless the proposal specifically
identified a function as possible only in version 3.0, which this section does not).
77q. At pages 4-36 to 4-37 Hart states that no additional software would be required
to make the base system support the languages described in the report. In fact additional
software was required to make the base system support these additional languages
(though Hart did not charge for that software), even in the uncertified eSlate version 3.0,
which is the version Hart references in this section.
77r. At page 4-40 Hart states that it can provide a requested automated logic and
accuracy test routine. In fact Hart had no software capable of performing the requested
functions in an automated fashion and none was developed during Mr. Singer’s tenure
with the company.
77s. At page 4-45 Hart states the cast-vote records are created as a “triplicate
original” stored on the eSlate, the JBC, and the MBB. Such claims of triple redundancy
are misleading for reasons noted in ¶¶ 70 and 77b. Additionally, the JBC stores votes
internally and on a flash card, so that if the JBC is damaged or corrupted, the card it is
writing to is necessarily suspect. In any event, and particularly in that event, if the castvote
records differ on any of the three storage components, it is impossible to determine
which has the “true” vote record. Thus, triple redundancy is valuable only if all three
sources are available and are in agreement. Moreover, in situations of machine failure,
SERVO could not be used to extract cast-vote records because the standard interfaces
would be disabled (it is here that JBC Recovery would be used as described in ¶ 70
above). Finally, Hart concealed the fact that in many instances of machine failure, the
units would be shipped to Hart’s programming team and secretly counted with no
verifiability, as described in ¶ 71. Hart’s response at 4-45 is intentionally misleading in
omitting these facts, which would be crucial to a prospective purchaser.
77t. At page 4-46 Hart is asked to describe its post-election night process to capture
information, including “ballot images, hardware logs, device activity logs, and operator
activity logs.” Hart states that “[a]ll information is captured and transferred to the
tabulation system on Election Night.” This is false. Ballot images from Ballot Now are
not transferred, nor are the BOSS or unit audit logs. Likewise, as noted in Exhibit 14
(Draft System 2.1 Release Notes at page 9 of 10 [sic]), the tabulator Tally’s “known
defects” that were not fixed included the fact that “[A] user has the ability to assign/reject
write-ins. This action is not being captured by the Audit Log.”
77u. At page 4-46 Hart touts the post-election use of SERVO to backup and manage
crucial vote information. However, Hart concealed numerous performance issues with
SERVO that called into question its accuracy and reliability. Attached as Exhibit 23 is an
email from Mr. Singer to his supervisor, Travis Harrell, indicating problems with SERVO
and memorializing the fact that Mr. Singer had reported to Hart such problems in “every
version” of SERVO (necessarily including the version at issue in the Boulder proposal).
(Ex. 23, Singer Email to Travis Harrell, 12/17/03). Attached as Exhibit 24 is an email
string beginning with Hart Account Manager Rich Geppert on June 5, 2003, (11 days
before the Boulder submission) regarding “another Servo problem” experienced in Texas
that affected the ability to recount votes. Programming manager Michael Lawless states
that “We’re aware of this issue” and that a fix is planned for SERVO version 2.0
(whereas 11 days later Hart proposed version 1.02, i.e., the version known to be faulty, to
Boulder County). Programmer Lawless goes on to conclude “I think we would all agree -
it should have been fixed.” Hart again touts SERVO at page 4-186 without revealing
these known defects and without revealing Hart’s intention not to correct the problem in
the version it proposed for use in Boulder (version 1.02).
77v. At page 4-159 Hart states that in eSlate version 2.1 ballot font sizes could be
increased if desired. In fact increasing the font size would destroy the ballot structure.
This feature was not available until version 3.0, which was not certified at the time of this
submission (see page 4-1).
77w. At page 4-65 Hart states that “[o]nce the voter selects a language, all further text
and audio information is delivered in the selected language.” In fact error and status
messages were English-only.
77x. At page 4-82 Hart is advised that its system must be capable of use in open
elements and at page 4-83 it is asked to describe requirements necessary for optimum
operation. Hart responds at page 4-83 that “the eSlate system can operate optimally in all
conditions . . . .” However, Hart conceals the fact that the system will not work in direct
sunlight because the LCD will blank out. DAUs are of particular concern in this regard
because they are advertised as capable of being removed and brought outside to assist
77y. At page 4-87 Hart states that Tally “operates in closed, Windows NT/2000
configuration to provide a secure physical and logical environment.” In fact eSlate 2.1
had no Windows 2000 customer configuration that would count live votes.
77z. At page 4-88 Hart states that “[a]ll data and applications are Windows NT or
2000 compatible.” In fact data for MBB images was not readable in any standard format
such as Windows (though it could be read through special utilities).
77aa. At page 4-88 Boulder requires that “[a]ll software used to move results shall
generate logs reporting all . . . . operator activity . . . .” Hart states that “[a]n audit trail is
maintained of all data movement [and] a real time audit log is continuously printed during
the tabulation process.” Hart neglects to mention that these audit logs track only activity
within standard Hart applications. Hart made no provision for detecting or comparing
external changes so that, for example, a “hidden” program to manipulate votes would not
have been detected or audited. Additionally, as noted in Exhibit 14 (Draft System 2.1
Release Notes at page 9 of 10 [sic]), the tabulator Tally’s “known defects” that were not
fixed included the fact that “[A] user has the ability to assign/reject write-ins. This action
is not being captured by the Audit Log.” Thus, “all data movement” was not necessarily
captured by an audit log.
77bb. At page 4-91, in a response regarding security protocols, Hart states that “[t]he
remote transfer (Rally) stations connected to the ballot tabulation systems (Tally) are
mutually connected via a physically secure private network, which is not connected to
outside networks.” In fact Rally and Tally connected via standard phone lines. Hart also
says the system is “completely secure from outside intrusion.” Theoretically, one need
only have hooked up a network cable in order to intrude the system. Hart could have
avoided this vulnerability by not using network cards—as Mr. Singer fruitlessly suggested
77cc. At page 4-103 Hart states that it has “developed a vote simulation utility, used
to simulate voting on the eSlate System, and will provide this capability for Boulder
County.” At the time this statement was made the vote simulation utility was recognized
within Hart to have been so unreliable that it could not be released to purchasers. Mr.
Singer has firsthand knowledge that these utilities were never intended for public release.
Even if Hart had provided such a utility to Boulder County, it could not possibly have
provided technical support because such support would have to come through Mr. Singer,
who was denied formal access to it.
77dd. At pages 4-156 and 4-158 Hart confirms that it has a “24-hour technical
support (hotline)” and that it “propose[s] a 24-hour hotline . . . staffed with qualified
technical personnel who are able to assist with any problems that the Boulder County
Clerk and Recorder might encounter . . . .” In fact, the “24-hour hotline” was a single
phone on the back of a desk in the tech department. The computer techs routinely ignored
it (and advised their managers that they refused to answer it given their workload).
Ordinarily if a call was taken it was answered by a trainer or account manager.
77ee. At page 4-159, Hart represents that it provides “Remote dial-up/Internet
software diagnostics.” In fact Hart did not provide any such service nor did it have the
capability to do so.
Systematic Practices. Hart made each misrepresentation identified above, by act
or omission, in a systematic attempt, and pursuant to corporate policy, to knowingly cause
the payment of federal funds for its voting systems. Hart concealed known software and
hardware flaws when making sales presentations. Hart made knowing misrepresentations
concerning the testing, accuracy, reliability, security, accessibility, verifiability, and legal
compliance of its systems. And Hart concealed new problems as they arose (thus failing
to remedy them with respect to systems already purchased) in an ongoing effort to
increase its competitiveness in order to secure further federal funds and in order to avoid
being required to repay funds already paid to it for its voting systems and technical
support. Hart billed its customers for voting systems (and support) that would not have
been purchased but for Hart’s fraudulent acts and omissions. All such systems (and
support) were purchased in whole or in part with federal funds. At no time did Hart ever
disclose to any potential or actual purchaser that its voting systems failed to comply with
HAVA or VSS standards though in many respects, as Hart knew, the systems did not
comply with HAVA or VSS standards.
Ongoing Misrepresentations. Mr. Singer has firsthand knowledge that Hart made
scripted/routinized claims in its sales presentations and thus has made such known
misrepresentations to other purchasers. Likewise, one may reasonably infer that Hart
proposals to other purchasers were substantially similar to the Boulder proposal. Thus,
discovery of other Hart responses to requests for proposal unquestionably would reveal
numerous similar (if not identical) and additional misstatements to other governmental
entities in Hart’s successful attempts to secure federal HAVA funds.
COUNT I: Violations of the False Claims Act
80. Each of the foregoing allegations is realleged and incorporated hereby.
81. As described in this Qui Tam Complaint, Defendant Hart, by and through its
officers, agents, and employees: (i) knowingly presented, or caused to be presented, to the
United States Government, a false or fraudulent claim for payment or approval; (ii)
knowingly made, used, or caused to be made or used, a false record or statement to get a
false or fraudulent claim paid or approved by the Government; and (iii) knowingly made,
used, or caused to be made or used, a false record or statement to conceal, avoid, or
decrease an obligation to pay or transmit money or property to the Government.
82. Defendant Hart authorized and ratified all the violations of the False Claims Act
committed by its various officers, agents, and employees.
83. The United States Government and the public fisc have been damaged as a result
of Defendant Hart’s violations of the False Claims Act.
84. Mr. Singer requests a jury trial on all issues so triable.
85. WHEREFORE, Relator William Singer, on behalf of himself and the United
States Government, prays:
(i) that this Court enter a judgment against Defendant in an amount equal to three
times the amount of damages the United States has sustained as a result of
Defendant’s violations of the False Claims Act;
(ii) that this Court enter a judgment against Defendant for a civil penalty of
$10,000 for each of Defendant’s violations of the False Claims Act;
(iii) that Relator William Singer recover all costs of this action, with interest,
including the cost to the United States Government for its expenses related to
(iv) that Relator William Singer be awarded all reasonable attorneys’ fees in
bringing this action;
(v) that in the event the United States Government proceeds with this action,
Relator William Singer be awarded an amount for bringing this action of at
least 15% but not more than 25% of the proceeds of the action;
(vi) that in the event the United States Government does not proceed with this
action, Relator William Singer be awarded an amount for bringing this action
of at least 25% but not more than 30% of the proceeds of the action;
(vii) that Relator William Singer be awarded prejudgment interest;
(viii) that a trial by jury be held on all issues so triable; and
(ix) that Relator William Singer and the United States of America receive all relief
to which either or both may be entitled at law or in equity.
s/Michael S. Burg
Michael S. Burg
Burg, Simpson, Eldredge, Hersh & Jardine, P.C.
40 Inverness Drive East
Englewood, CO 80112
Attorney for Relator
s/Peter W. Burg
Peter W. Burg
Burg, Simpson, Eldredge, Hersh & Jardine, P.C.
40 Inverness Drive East
Englewood, CO 80112
Attorney for Relator
s/David P. Hersh
David P. Hersh
Burg, Simpson, Eldredge, Hersh & Jardine, P.C.
40 Inverness Drive East
Englewood, CO 80112
Attorney for Relator