Verified Voting Blog: Virginia – the new Florida?

WinVote-Screen

There are many ways in which Virginia 2012 could resemble the Florida 2000 – only worse. At least in 2000 there were paper ballots to recount in Florida.  But only 7 out of 134 Virginia localities (Virginia terminology for counties and independent cities) do not use paperless  Direct Recording Electronic (DRE) voting machines. If a DRE loses or miscounts ballots, it is essentially impossible to determine the correct results.

As if to guarantee that it will be impossible ever to verify an election in Virginia, Virginia law actually prohibits manual post-election ballot audits of paper ballots, except in extremely narrow and unlikely circumstances. This prevents election verification even in the 7 localities that have no paperless DREs (Chesterfield, Gloucester, Hanover, New Kent, Wythe, Fredericksburg, and Williamburg), together with the 30 other localities that have a mix of paper ballots and paperless voting machines.  Unless the anti-verification law is repealed, Virginia will continue to be a poster child for how not to run an election, even after Virginia replaces all of its antiquated paperless DREs with paper ballot based optical scan systems, as it should. But it gets worse.

Two notoriously unreliable paperless DRE systems are still being used in Virginia, years after their inadequacies had become common knowledge. The AVS WINVote, used only in Hind County, Mississippi and in the state of Virginia, failed to qualify for federal certification in 2007, even to the lower testing of the two voting systems standards. Since then, AVS seems to have folded, with maintenance  done by Election Services Online, a Philadelphia based company with ties to the Shoup family, that founded AVS predecessor company over a century ago. A distinctive feature of the WINVote that makes it particularly vulnerable is it’s use of wifi to communicate between equipment in the polling place.

The other unreliable paperless DRE system is the Unilect Patriot, which gained notoriety in November, 2004 in both North Carolina and Pennsylvania. In Carteret County, N.C. a Patriot machine lost almost 4500 votes in early voting, while in Pennsylvania the Patriot appears to have lost a significant number of votes in the 2004 presidential race. Pennsylvania Secretary of State Pedro Cortes issued a report claiming that the Patriot was not “capable of absolute accuracy” and was not “safely and efficiently usable”. Pennsylvania decertified the Patriot; it is now used only in Virginia. Should either the AVS WINVote or the Unilect Patriot malfunction on Election Day with the Presidential or Senate race hanging in the balance, our country could be in uncharted territory.

What can be done?

Unfortunately, it is too late to replace Virginia’s aging and unreliable voting systems before the 2012 election.  The following steps may help ameliorate the situation in Virginia.

All localities that have both paperless DREs and paper-based optical scan systems should direct voters to use the optical scan systems, instead of allowing them to choose.  Since many voters are unaware of the security and reliability risks of paperless DREs, a majority have chosen to vote on DREs.  Fairfax County, one of the locations with a hybrid voting system, contains 15% of the population, so such a requirement could have a significant impact.

All polling places that use only paperless DREs should be required to provide emergency paper ballots.  Voters should be encouraged to vote on paper ballots if any DRE fails or the wait to vote is greater than a reasonable threshold, for example 30 minutes.  Furthermore, all voters who don’t trust DREs should be allowed to vote on paper ballots, with the guarantee that those ballots will be counted.  Virginia allows emergency paper ballots to be deployed if any machine fails, but it appears not to do so in practice.

The inadequate system of ballot accounting and reconciliation for DREs needs to be fixed.  Currently, if the number of cast ballots recorded by the DREs exceeds the number of voters who signed the poll books, the DREs are assumed to be correct.  This irrational practice could lead to ballot box stuffing and should be outlawed.  A forensic investigation using independent computer security experts should be mandated by law whenever there are unresolved discrepancies.  All discrepancies should be publicly reported, all of the results of the investigation should be made public, and the certification of the election should be delayed to allow time for legal challenges after the public release of the investigation report. During a March 2009 school board election in Fairfax County, election officials discovered that while a total of 707 votes had been cast, one of the two machines in the precinct reporter 348 votes, while the second reported 724 votes.  After officials printed the cast vote records from the questionable machine, they determined that only 359 votes had been cast and were able to release the election results.  Even though no meaningful recount could have been conducted on the paperless ballot representations, the discrepancy was resolved, but only by relying on and trusting software that had already malfunctioned.

The inadequate system of ballot accounting and reconciliation of optically scanned paper ballots should be reformed by requiring a manual count of the paper ballots if the number of ballots exceeds the number of voters.  While such a recount might not eliminate the discrepancy, it would catch some problems, such as the multiple counting of ballots.  Currently, after the paper ballots have been tabulated, the numbers are compared to the total of registered individuals who voted by paper ballot, as recorded by the poll books.  If the comparison indicates that ballots folded together were likely cast by the same voter, the “extra” ballots are not counted.  If ballots cast still exceed the number of names on the poll books, a blindfolded elections officer must randomly withdraw ballots equal to the number of the excess.  Although an improvement over simply leaving the excess ballots in the count as is done in DRE jurisdictions, this practice also lends itself to ballot box stuffing and should be outlawed.

All jurisdictions should be required to reconcile precinct totals with locality totals.

Reformers should push for manual post-election ballot audits where paper ballots are deployed and counted by computerized systems.  Because Virginia law prohibits manual post-election audits of paper ballots and essentially makes human recounts of ballots illegal (ballots can be recounted only by the same systems that might have initially miscounted), it could be impossible to determine the correct outcome of a close election, unless a court orders a proper manual audit or recount.  The law ignores the risk that computer-reported results could be incorrect, either because of software failure or hidden malicious software that manipulates results.  Instead of forbidding them, manual audits and recounts should be required for all major elections to verify that the computers-produced results are accurate.

Because it is impossible to conduct manual post-election audits of paperless voting systems, paperless voting systems should be eliminated as quickly as possible and replaced with paper ballots based optical scan systems.

Share

Comments are closed.