Testimony: Rescanning On Same Type Scanner Would Gut Post-Elecition Audit

Friday was the public hearing on S.B. 364 proposed by the Registrars of Voters Association of Connecticut (ROVAC).  It would replace the manual hand count post-election audit by a rescanning by the same model of ES&S/Premier/Diebold AccuVote-OS optical scanner with a duplicate memory card.

Here is the prepared introduction to my testimony:

Chairs and members of the Committee, my name is Luther Weeks. I am Executive Director of CTVotersCount and Executive Director of the Connecticut Citizens Election Audit Coalition.  I have personally observed twenty-five (25) post-election audits. Today, I am speaking for CTVotersCount. (I am also a certified moderator and a retired computer scientist)

I am opposed to the current Senate Bill 364 which would gut the current post-election audit law by eliminating the manual count and substituting an almost useless recount by the same type of machines and duplicate memory cards.

Instead, CTVotersCount recommends our alternative bill which would clarify the ballot chain of custody, make the audit procedures enforceable, while improving the efficiency and integrity of the audit – without increasing costs.

The current bill is based on three assumptions, all of which are erroneous because:

1. A count by an identical scanner and memory card is NOT equivalent to a manual count. In the words of a League of Women Voters Report:

An audit count that simply repeated the original counting procedure, whether electronically or by hand, would add little value to the election-validation process.

2. The audits DO NOT cost too much.  Based on municipality requests for reimbursement to the State and citizen observations, the costs of the 2008 audit are estimated at $82,000 and the cost of the 2009 audit at less than $41,000.  (Under our proposal the 2008 audit would have likely have cost 20% to 30% less than it did.)

3. People in Connecticut CAN count votes accurately. Officials in other states have regularly counted paper ballots quite accurately with less than 1/8th difference rate as Connecticut officials, at a lower cost.  The keys are better methods, documentation, training, and organization.

My written testimony includes details supporting estimates for the costs of the audits, the ability of people to count votes accurately, and the advantages of our proposed alternative bill.

I have also delivered testimony to the committee from several computer scientists from Connecticut and around the country attesting to the fallacy of auditing by identical computers and the costs of manual counting in other states.

In summary: Saving money is an important goal, yet the value delivered for expenditures and the value lost in the name of savings should be recognized and considered.  Should we stop inspecting highways, bridges, school buses, and election systems because it costs money? Or should we continue because they protect the value of our investments in infrastructure, save lives, and protect Democracy?

Thank you.

My complete testimony.

Fortunately the Committee did not have to rely exclusively on my testimony and experience observing post-election audits:

  • Two Coalition Post-Election Audit Observers gave a real-world voters’ view of the current process and its limitations:  Tessa Marquis and Mary Rydingsword.
  • Christine Horrigan and the Connecticut League of Women Voters testified in opposition to the current bill.
  • Out-of-state experts who provided testimony that audits don’t cost that much and that using similar machines and memory cards is of little value: David L. Dill, Jeremy Epstein, Candice Hoke, Peter G. Neumann, and Barbara Simons, along with two Computer Scientists from Connecticut, the President and a member of TrueVoteCT who lent their names to the testimony: Michael Fischer and Ralph Morelli <testimony>

There is a sound technical basis for verifying electronic vote tallies by manually counting a sample of precincts or vote subtotals. As computer scientists and election experts, we know very well that there is no reliable way to ensure that a security-critical computer system, such as a vote scanner, is free of malicious software that can change votes — or is even bug-free, for that matter.

It has been shown time and time again that there is a clever way to defeat every defense that has been invented. Furthermore, basic errors and gross security holes have been exposed in every existing voting device examined by computer security professionals to date. Errors are routinely detected in elections – and many smaller errors are probably missed. In 2008, hand-counted tabulation audits have discovered errors that led to incorrect vote totals. Computers can greatly increase the convenience and accuracy of elections – but only if we double-check the results independently of the hardware or software by hand counting a randomly selected sample of the ballots.

  • A special thanks to Barbara Simons for her own independent testimony: <testimony>

Senate Bill 364 appears to be based on the flawed premise that simply rerunning paper ballots through vote counting scanners used for the initial count can replace the manual audits currently required by law. Unfortunately, this assumption is wrong. Because the proposed rescan is not an audit, it will not achieve the goals for which the manual audit was intended. I urge the legislature to retain the manual hand counts in the existing audit legislation.

  • Also Secretary of the State, Susan Bysiewicz, testified her opposition to the bill.

Our hope, based on the overwhelming opposition to be bill, based strongly on well established science, is that the bill will either be completely rewritten based on our recommendations or not moved forward by the Committee.

We also note that the testimony on this bill and other election bills considered, resulted in interest by the Committee and important discussions of the details of how incorrectly marked ballots are counted, the importance of careful recanvasses, and the ballot chain-of-custody.

We expect that in a few days all of the written and verbal  testimony will be available <here> and <here>

Another instructive issue:

In other testimony we agree with Registrars who testified for improvement/removal of the quirks in Connecticut law which require multiple moderators in each polling place when more than one party has a simultaneous primary.  And perhaps with even less justification, in elections, officials can be recruited from any town to serve in another, in Presidential primaries too, yet in local, state, and other Federal primaries all officials must come from the town of the polling place.

The basic problem is that our election statutes are a hodgepodge.  Some sections apply to primaries, others Federal elections, others state elections, others local elections, others local primaries, others referendums etc.

As partially covered in our proposed revisions (see our proposed bill in our testimony) there is a need for clarifying, consolidating, and rationalizing the ballot chain-of-custody.  The law has different rules for moving and preserving absentee ballots, polling place ballots cast in all paper elections, and polling place ballots cast in optical scan elections.  That last item covering the majority of ballots and elections, polling place ballots cast in optical scan elections, is a hodgepodge of unclear language partially reflecting lever machines and ballot boxes, while reverencing a hard to follow chain of law leading to sections referencing absentee ballots.  It seems to be more of a chain-of-confusion and ambiguity than a chain-of custody we can rely on or expect officials to understand!

A 74 page bill is being considered to fix some of the continuing reverences in the law that remain focused on lever machines and are inappropriate for optical scanners.  Between mixing lever machines and scanners, even DRE’s (which have not been used in Connecticut), incomplete recanvass laws, and chain-of-custody laws, there remain plenty of issues to address in the future.


One response to “Testimony: Rescanning On Same Type Scanner Would Gut Post-Elecition Audit”

  1. rjs

    While I agree that a recount is preferable to a recanvass, I fear that decades of using the mechanical lever machines has had the result of rendering Connecticut officials incapable of performing a real recount.

    Having said that, the audit process is serving to re-educate officials to this necessary function, as the audit is simply a targeted recount.

    As you may know, our November audit (in West Haven) was a complete bust, and we were constrained to retry 10 days later. In the interim, I devised a training program which took about 30 minutes (including a quick “test”). After this training, the audit proceeded smoothly and quickly.

    I can be reached through the West Haven Registrars if you’d like more information about this.

Leave a Reply

You must be logged in to post a comment.