|
paper trail rule for black boxes.
United States Election Assistance Commission 1225 New York Avenue N.W. Washington, DC 20005 July 20, 2005
EAC Advisory 2005-004: How to determine if a voting system is compliant with Section 301(a) – a gap analysis between 2002 Voting System Standards and the requirements of Section 301(a) The United States Election Assistance Commission (EAC) has received a number of inquiries from several states as to whether one or more particular voting systems comply with Section 301(a) of the Help America Vote Act of 2002 (HAVA). In addition, in one of its recent public meetings, EAC was asked to conduct an analysis to identify the gaps between the 2002 Voting System Standards adopted by the Federal Election Commission (FEC) and the requirements for voting systems under Section 301(a) of HAVA. EAC is not required by HAVA to preclear or approve voting systems purchased by states and local election jurisdictions. Furthermore, EAC does not believe that it was the intention of Congress or HAVA for EAC to assume this role. However, it is evident that states and local election jurisdictions as well as testing laboratories are in need of information that will help in determining whether a voting system meets the threshold requirements of Section 301(a). Thus, EAC offers the follwing analysis of Section 301(a) in light of the 2002 Voting System Standards.
Title III of HAVA, entitled “Uniform and Nondiscriminatory Election Technology and Administration Requirements,” imposes certain requirements upon states and local jurisdictions conducting federal elections. Section 301(a) sets forth the standards that voting systems must meet after January 1, 2006. Those requirements include functions and features that, among other things: (1) allow the voter to review his or her selections privately and independently prior to casting a ballot; (2) allow the voter to change his or her selections privately and independently prior to casting a ballot; (3) notify the voter when he or she has made more selections in a single race than are permitted (overvote); (4) provide for the production of a permanent paper record suitable to be used in a manual recount; (5) provide voters with disabilities, including visual disabilities, the same opportunity for access and participation (including privacy and independence) as for other voters; (6) provide accessibility in minority languages for voters with limited English proficiency as required by the Voting Rights Act of 1965; and (7) provide for an error rate in operating the voting system that is no greater than the error rate set forth in Section 3.2.1 of the 2002 Voting System Standards adopted by the Federal Election Commission (FEC).
Although the 2002 Voting System Standards set forth measurable standards that predict compliance with some of the Section 301(a) requirements, those standards do not provide sufficient and adequate guidance as to what is required to meet the accessibility requirements of Section 301(a)(3); do not prescribe testable measures for language accessibility required by Section 301(a)(4) of HAVA; and do not prescribe standards that adequately explain the requirements for overvote notification required by Section 301(a)(1) of HAVA. As such, EAC issues the following policy statement to identify the gaps between the 2002 Voting System Standards and the requirements set forth under Section 301(a) of HAVA and to explain what is needed to meet the requirements of Section 301(a) above and beyond the testing requirements established in the 2002 Voting System Standards.
Section 301(a)(1):
The requirements of Section 301(a)(1) of HAVA are met if the voting system (1) conforms and complies with Section 2.4.3.3 of the 2002 Voting System Standards and (2) notifies the voter through a visual and/or audio message prior to casting the ballot when the voter makes more selections than are legally allowed in a single race or contest (overvote):
(a) that an overvote has occurred and (b) the effect of overvoting.
Following that notification, the voting system must allow the voter to change his or her selection(s), if so desired. Voting systems that preclude and prohibit overvoting meet this requirement. Notwithstanding the above, certain paper ballot voting systems may meet the overvote requirements of Section 301(a)(1)(A)(iii) of HAVA by meeting the requirements set forth in Section 301(a)(1)(B).
Section 301(a)(2):
The requirements of Section 301(a)(2) of HAVA are met if the voting system conforms and complies with Sections 2.2.5.2.1 and 2.5.3.1 of the 2002 Voting System Standards.
Section 301(a)(3):
Section 301(a)(3) of HAVA requires that by January 1, 2006, at least one voting system in each polling place be accessible to persons with disabilities such that the voting system allows an individual with a disability the same access and opportunity to vote privately and independently as is afforded a non-disabled voter. Compliance with Section 301(a)(3) requires that the voting system is accessible to persons with disabilities as defined by the Americans with Disabilities Act, including physical, visual, and cognitive disabilities, such that the disabled individual can privately and independently receive instruction, make selections, and cast a ballot. However, accessibility involves more than the technical features of the voting system. The accessible voting system also must be used in a manner that is consistent with providing access for disabled voters (e.g., the accessible voting system must be set up for use in a space that is accessible to a disabled voter who uses a wheelchair).
Conformance with Section 301(a)(3) is a complex matter, which must take into account the disability of the voter, the advancement of technology and its availability, and the efforts of the elections officials to make the voting process accessible to disabled voters in a private and independent manner. The following are some factors that must be considered in determining accessibility in conformance with Section 301(a)(3) of HAVA:
(1) Section 2.2.7 of the 2002 Voting System Standards; (2) Section 2.4.3.1 (a) of the 2002 Voting System Standards; (3) Section 3.4.9 (a-e) of the 2002 Voting System Standards; (4) The voting system must afford a disabled voter the ability to perform the same functions (e.g., receiving and reading the ballot, making selections, reviewing selections, changing selections, and casting the final ballot) as are afforded to a non-disabled voter. These functions may be provided to the disabled voter through features of the voting system that are different than those used by non-disabled voters. The disabled voter need not and in many cases cannot have an identical voting experience as a non-disabled voter (e.g., a voter with a visual disability is afforded the same access to reading the ballot as a sighted voter when the ballot is read to the visually disabled voter using an audio component of the voting system). (5) Accessibility of the voting system to the voter includes accessibility to all equipment needed to cast and count ballots. Many jurisdictions use a paper ballot voting system that requires the voter to submit his or her own ballot after casting for purposes of ballot counting. Where such voting systems are in use, such jurisdictions must to the extent reasonably and technologically possible afford a disabled voter the same ability to submit his or her own ballot, in a private and independent manner, as is afforded a non-disabled voter. In this example, visually disabled voters must be allowed to submit the ballot independently, as the disability is one that is capable of being accommodated, and technology and practice provide a means that can be used to allow the visually disabled voter to submit a ballot with the same degree of privacy and independence afforded to a sighted voter (e.g., a privacy sleeve). (6) There may be certain disabled voters whose disabilities prevent them from voting independently (i.e., without assistance from a person of their choosing or a poll worker). While HAVA requires voting systems to allow independence and privacy, it does not preclude a disabled voter from requesting and obtaining the assistance of another person as provided in Section 208 of the Voting Rights Act of 1965. (7) Section 301(a)(3)(B) contemplates that an accessible voting system can include a direct recording electronic (DRE) voting system or other voting system equipped for individuals with disabilities. This advisory should not be read to preclude the innovation and use of accessible voting systems other than DREs for purposes of meeting this requirement. A Bad Bill, With Some Good Things in It
June 14, 2005
Daniel P. Tokaji Associate Director, Election Law @ Moritz
Voting activists descended upon Washington last week to urge enactment of the “Voter Confidence and Increased Accessibility Act of 2005” (H.R. 550). This bill is one of several that aims to enhance the security of electronic voting machines, by requiring that they produce a contemporaneous paper record of the electronic ballot, or “voter-verified paper record.” While H.B. 550 has some worthy features, Congress should not enact the bill in its present form because its core requirement would do more harm than good.
What the Bill Would Do
H.R. 550 is a modified version of a bill that was considered but not enacted by the last Congress (H.R. 2239). The chief sponsor of both bills is Representative Rush Holt (D-NJ). This year’s version has 135 co-sponsors in the House, almost all of them Democrats.
H.R. 550 would amend certain provisions of the Help America Vote Act (“HAVA”), the election reform law passed in 2002. The centerpiece of the current bill is its requirement that all voting systems “produce or require the use of an individual voter-verified paper record of the voter’s vote.” For electronic touchscreen machines, this would require an attached printer that would generate “a paper print-out of the voter’s vote.”
While HAVA already requires that voting systems generate a paper record, H.R. 550 would go further. In particular, it would require that electronic voting machines generate a paper record contemporaneously – that is, at the time of voting – so that the voter has the opportunity to “verify” his or her choices. This requirement would take effect on January 1, 2006, the effective date of HAVA’s other voting system requirements.
This sounds good on paper (apologies for the pun), but the results would be chaotic and possibly disastrous in practice. The most conspicuous problem is timing. It is unrealistic to require all jurisdictions using electronic voting technology to obtain printer add-ons by the 2006 election. That’s particularly true, given that there aren’t any federal standards for this device in place yet.
As many election officials discovered in the wake of the 2000 election, rushing the transition to new voting technology can cause more problems than it solves. A 2006 deadline doesn’t give states and counties nearly enough time to investigate their options, go through the procurement process, implement the new equipment, train their poll workers, and educate their voters.
Worse still, there is no provision exempting counties that have already made a successful transition to electronic voting technology. The State of Georgia, for example, moved to electronic voting throughout the state and saw the number of uncounted votes decline dramatically in 2004, particularly in heavily-minority precincts. Under H.R. 550, Georgia election officials would have just a few short months in which to replace their existing system – or retrofit it with an as-yet unproven printing device. That’s inviting trouble. It would be unfortunate if Congress enacted legislation that had the effect of reversing the substantial improvements in technology since the 2000 election.
The Experience in the States
Even beyond the timing issue, it’s doubtful that H.R. 550 provides a workable or effective solution to the legitimate concerns regarding electronic voting security. So far, only one state – Nevada – has actually implemented an electronic voting machine with a contemporaneous paper record on any significant scale. Anecdotal evidence indicates that few voters actually bothered to check the contemporaneous paper records. A serious and unbiased analysis of Nevada’s experiment is needed, before even considering a national mandate.
While experiments like Nevada’s are to be encouraged, it’s a mistake to lock in place a particular would-be solution – especially one that’s expensive and not yet proven. This will only discourage states from trying other mechanisms that may turn out to be more effective. In the end, it may actually make our voting system less secure than it otherwise would be.
Congress should also look to the experience of states that have passed legislation to require a contemporaneous paper record. The experience of California and Ohio, in particular, demonstrates the hazards of mandating a contemporaneous paper record.
In California, former Secretary of State Kevin Shelley issued a directive requiring that electronic voting systems generate a contemporaneous paper record effective 2006, and the California legislature wrote this requirement into state law last year. The dilemma that counties face is that it’s not at all clear whether these systems will actually work effectively – or, alternatively, whether paper jams, privacy issues, lack of accessibility, or the difficulty in reading the paper print-outs will hamper their effectiveness.
The result of this dilemma has been for the largest county in the United States, Los Angeles County, to retain its “Inkavote” system – an optical scan system that functions like the Votomatic punch card machine, but with dots that voters fill in instead of chad. This device was intended to be an interim solution, and is an improvement over the Votomatic that Los Angeles County used before 2004. But while “hanging chads” aren’t a problem with the Inkavote, it’s equally difficult for voters to check their work. The Inkavote is therefore inferior to the paperless electronic systems currently available, which not only reduce the number of uncounted votes but also provide accessibility advantages for disabled and non-English proficient voters. The result of California’s paper-trail bill has thus been to hinder the country’s largest electoral jurisdiction from moving to better technology.
The experience of Ohio, which also passed a law requiring a contemporaneous paper record last year, is equally illustrative. In Ohio, the result of this law has been to give a virtual monopoly of the electronic voting market to a single company – the oft-reviled Diebold. Until recently, there were no electronic systems certified in the state that satisfied the state’s requirement that these machines generate a “voter verified paper audit trail.” The only system presently certified is manufactured by Diebold and it is presently unclear whether any other systems will be certified in time to be implemented in 2006. Thus, in Ohio, the ironic result of the campaign to require a contemporaneous paper record was to give a virtual lock on the market to the very company that has been the poster child for the supposed ills of electronic voting.
The Good, the Bad and the Ugly
California’s and Ohio’s experience with “voter verified” paper trail laws should give pause to anyone who believes that H.R. 550 is the magic bullet to the problem of electronic voting security. More fundamentally, the bill rests upon the questionable premise that paper is less susceptible to manipulation than electronic records. While there have been a number of documented instances of fraud in past elections, most have taken place with paper ballots; none have taken place with electronic voting.
Moreover, it’s unlikely that the recount process mandated by the current version of H.R. 550 would catch fraud, even if it did occur with an electronic voting machine. The bill would require that the Election Assistance Commission (“EAC”) conduct a manual recount of two percent of all precincts in the country in each general election. At least one precinct per county would be recounted. This would impose an enormous burden on the EAC, requiring unprecedented coordination among federal, state and local election officials, but would have only modest benefits. It would do nothing to catch fraud at the local level, where it is more likely to occur given that relatively small changes in vote totals can swing an election. Even for federal races, such a recount would by itself provide little reason for increased confidence in the accuracy of the result, according to this analysis by VoteHere’s Andrew Neff. Neff's analysis finds that a two percent precinct recount of a U.S. House race would catch a five percent change in the results (either accidentally or maliciously) with only 40 percent confidence. The bottom line is that the recount process mandated by H.R. 550 would impose a heavy financial burden on the federal government, while doing little to promote electoral integrity.
It is certainly important that voters be allowed to “check their work.” The paperless electronic voting machines already available allow voters to do this, in a way that’s likely to be more effective than a contemporaneous paper record that voters may or may not actually check. The best thing that may be said of allowing voters to check a paper record is that it functions as a sort of placebo, helping to elevate public confidence that elections are being conducted on the square. Even here, however, the benefits of paper are overstated. Approximately 90 percent of ballots in Washington’s recent gubernatorial election, for example, were cast by paper ballot. Yet this did little to “convince the loser that he or she lost,” one of the supposed advantages of paper.
None of this is to say that fraud and error are impossible with electronic voting. But better testing and procedures are much more likely to promote electoral integrity than a printer add-on. The $150 million that H.R. 550 would authorize for such hardware – not to mention the unknown amounts that would be required on an ongoing basis for the 2 percent national recount – would be much better spent on improving the testing and certification of voting machines. The EAC is presently engaged in this process, but progress has been hampered by Congress’ delay in funding its work. <snip>
and so on..... sigh
:-(
|