The Outsourced, Unaccountable New Hampshire Election System

Friday, January 18:

The Outsourced, Unaccountable New Hampshire Election System
and Implications for the Recounts Underway

by Bruce O'Dell, EDA Co-Coordinator for Election Data Analysis

It's disturbing but hardly surprising to hear that LHS, the vendor that runs New Hampshire elections, took custody of the Diebold optical scan memory cards after the 2008 primary election.

In New Hampshire, LHS uses the notorious GEMS central tabulator software to program individual memory cards for each optical scan device that will be counting votes in each voting location. The memory cards contain ballot definition files that are used by the optical scan software to interpret
which marks on the ballot correspond to which candidates on the ballot at that location. The memory card also contains software, but of course, we know that software is perfectly benign... if you believe the vendor, because of course no one is allowed to independently examine those cards before or after the election. Oh, yes, the memory cards also store the"results" of the election.

After the election, each optical scan machine prints out a "poll tape"which reports the vote totals by race and candidate for that machine at that location. The poll tapes are signed by local election officials and couriered to the Secretary of State by NH state troopers. Unlike most jurisdictions, New Hampshire does not use GEMS to tabulate the vote; elsewhere, in most places that use Diebold optical scan voting equipment, the memory cards are couriered to the county election office and plugged
back into the county's GEMS server for upload and "automatic" tabulation.In New Hampshire, the poll tapes are manually tabulated by an individual using Excel.

So it's not surprising LHS retains custody of the memory cards, since the New Hampshire Secrtary of State doesn't use memory cards to tabulate the vote. But several years ago Harri Hursti and Black Box Voting publicly demonstrated just one means of altering the contents of a Diebold memory card in such a way as to both alter the outcome of the election and also produce a matching, fraudulent poll tape report.

What is astonishing to me is that New Hampshire continues to use precisely the same Diebold Optical Scan model - 1.94w - on which Hursti first demonstrated his insider attack.

What compounds my astonishment is that last autumn, Hursti presented the details of that exploit in person to the New Hampshire State House Subcommittee on Election Equipment - and yet the state still decided to keep running its elections on precisely that model of Diebold equipment!

Finally, although I did my best to convince New Hampshire legislators in my own testimony there last September, they also decided not to pioneer the UBS mandatory in-precinct election-night hand-count audit of a random 10% sample of all ballots.

This is a remarkable situation, so again let me recap: first, Hursti personally explained to New Hampshire officials how their specific election equipment could be manipulated by insiders (oh, say - just hypothetically -at LHS) in such as way as to change the outcome of an election, while at the
same time producing fraudulent poll tapes that confirm the mistabulation. The only way to detect Hursti's exploit would of course be to hand-count the ballots.

Second, New Hampshire also chose to not hand-count a statistically valid sub-sample of ballots in-precinct on election night - a technique that can detect accidental or deliberate mistabulation of 1% or more, of both hand-counted and optical scan ballots, with greater than 99% confidence.
UBS could easily have confirmed - or discounted - allegations that the unexpected ten point shift from final pre-election polls to election results in the Democratic primary did not truly reflect the will of the people of New Hampshire.

Had New Hampshire simply chosen to count a 10% sample of their ballots in precinct on election night, they could almost certainly have avoided the need to recount all the Republican and more than half of the Democratic primary ballots by hand - a week later, in adversarial circumstances, and
under a cloud of suspicion about chain of custody and the legitimacy of secret vote counting.

Instead New Hampshire chose to continue to rely on secret vote counting by machine, while at the same time relying completely on the integrity of the chain of custody of ballots stored out of sight for almost a week before they were brought back into public view in a different location.

In the absence of citizen-verified chain of custody in the week after the ballots were cast, the accuracy of the 2008 "recount" in New Hampshire will- to anyone familiar with even basic auditing principles - remain forever in doubt.

What I regret is that it did not have to be this way. I bet the good people I met on the New Hampshire State Subcommittee on Election Equipment last fall might very well feel the same way.

And it could still be different in November - not just in New Hampshire, but in Minnesota where I live, and across the country. But will we demand citizen oversight of secret vote counting in time to make a difference?

-Bruce O'Dell