Wednesday, June 18, 2008

NY Advocates to State Board of Elections: Make Election Audits RISK-BASED

A coalition of New York's election integrity advocates have asked the State Board of Elections not to approve the current draft of the Part 6210.18 auditing regulations, which are loosely based on the State Election Law requirement to manually count the ballots from 3% of electronic vote counting systems in each county (expected to be computerized paper ballot optical scanners).

Among other deficiencies, the regulations contain a loophole that would increase the hand counts well beyond the 3% called for by the election law for some counties and contests, while apparently neglecting others.

A letter sent to the State Board on June 16, after several months of intermittent negotiations and discussions, signed by Bo Lipari (New Yorkers for Verified Voting), Lawrence Norden (Brennan Center for Justice), Aimee Allaud (League of Women Voters), Neal Rosenstein (NYPIRG) and yours truly, among others, proposes risk-based statistical audits and the formation of a state election auditing task force within 60 days. The proposed group would conclude its work at least 60 days prior to the first 2009 election. The audits proposed in the letter are favored by many national election integrity advocates as a more efficient and effective method than percentage-based audits to confirm, independently of software and with a high level of confidence, the winners of elections called by electronic vote counting systems.

Certification of the systems slated to replace New York's mechanical lever machines has been fraught with difficulties such as continued non-compliance with federal and state voting system standards by vendors, and the withdrawal of federal certification of New York's first voting system testing lab, Ciber, Inc. in 2006.

These and other reported problems and failures, including documented cases of incorrect electoral outcomes reported by e-vote counting software, have led an increasing number of advocates, experts, and officials to conclude that the risk-based post-election audit, possibly resulting in a full hand recount, is the only way to know with any certainty that the outcomes of elections reflect the will of the voters.


Anonymous said...

The meeting of the NY Commissioners of Elections was held last Thursday, June 19th. So what happened? Did they agree to hold off on the formulation of auditing methods, or did they go ahead notwithstanding the advacates' letter and pass the proposed %-based regulation? If it was passed, can we get it changed by law by the NYS legislature?

Howard Stanislevic said...

Hi Laura,

They delayed the vote on the regulations, did not agree to a task force, but agreed to continued discussions. We hope to have those soon. I will report on any developments here.

I think there is some agreement that the Election Law is a floor but not a ceiling. It could be changed but it's not bad as long as we can do better by regulation.

Anonymous said...

Thanks, Howard. I know you & the other groups will be doing your best to sell the Commissioners on the advantage of risk-based audits. I would suggest you emphasize to the math-phobic among the Commissioners that, although the risk-based assessment sounds arcane and complicated, it is not that much harder to implement than the 3% audit, and that you can provide the language for the regulation and the programming (open to scrutiny by their own experts, of course) for determining how many votes to audit in which localities. What civil servants are interested in above all are ease of execution and a solution that is predictable and won't turn out to be a political can of worms. What might stump them and cause resistance is figuring out who they can use as an expert to verify to their satisfaction that you are not selling them a bill of goods. Non-mathematicians are understandably leery of statistics.

Howard Stanislevic said...


If election results were "predictable", we wouldn't need to have elections, would we?

The issue here is whether the voters of NY are going to trust only software to determine electoral outcomes; whether "certification" is a substitute for transparency (it's not), etc.

So far, to their credit, I have not seen any reluctance on the part of the State BoE to require hand counting significant numbers of votes. In fact, some of the stuff in the current version of the regulations would require some counties to hand count 100%! The problem is, hand counting is not being applied equitably and there will not be enough hand counting in some counties to confirm the outcomes of our most important elections.

If anyone thinks that high school algebra is somehow not sufficiently transparent, I would ask them to explain to us exactly how voting system software works. That said, the audit sample sizes could be listed in a simple lookup table in the regulations, and that possibility is being explored.

The urgent problem is, under current election law it may NOT be possible to obtain a manual recount in NY of more than 3% of the ballots cast at polling places, unless a court so orders. To convince a judge, you need to show evidence of miscounted votes, and the only reliable way to obtain such evidence is to hand count enough votes to find such miscounts. So we have a legal Catch 22!

If NY is not going to join the list of NO-RECOUNT States, we had better require risk-based statistical audits. Otherwise, we are in no better shape than FL, VA or the paperless states such as GA and PA, who can't even conduct recounts, and there would be no point in NY replacing our lever machines.

That's what's at stake.

Anonymous said...

Sorry if I sounded like I was questioning the need for risk-based audits. Nothing could be further from my sentiments. I was trying to see the situation from the perspective of the Commissioners to determine what is needed to overcome their hesitance. The "predictability" I referred to was the likelihood of the audit sample resulting in reported outcomes that will stand up to candidate challenges and more extensive recounts. Regarding that, I do not doubt your work, but was only exploring what would be needed to reassure the Commissioners.

From what you said about NYS law, we may be faced with a campaign to amend it when the Legislature resumes session in January. Hope your more optimistic initial assessment of "floor not ceiling" turns out to be the operative interpretation, but if it doesn't, the NY netroots is better organized now than it was when the campaign to fight DRE's began.

When I said "programming open to scrutiny", I was envisioning providing the SBOE with a short "wizard" to use in executing the steps of the equation. Perhaps my brief scan of Lindeman's paper on audits (and my poor math skills) gave me a false impression of how complicated the calculations need to be. It appears so if all that's needed is a table.

Anyhow I'm totally with you in believing meaningful audits are crucial. Know the we emailers, phone callers, letter writers, and amateur lobbyists have got your back.

Howard Stanislevic said...


Gee thanks! I don't think anyone's actually ever had my back before, so I appreciate that! :) There was a lot of support for the NJ Audit bill which became law last Jan. though. You can follow the history of that on Election Integrity: Fact & Friction.

You seem to have a lot of interest in this very important issue. If you'd like to discuss it further, feel free to email me.

Meanwhile, if you follow some of the links in the recently renamed "Educational Articles, Papers, etc." section of the blog, you can read all about the methods, and see some open source software demos that use the formulas which have already been peer reviewed and published. Of particular interest is that Roy Saltman, working for NIST, had most of this stuff figured out over 30 years ago!

I think the real issue is a lack of dialog with the decision-makers at the State Board of Elections because they are so busy with certification, etc. Hopefully, this has changed as a result of the letter sent by the Good Government groups. Thanks for your support!