Dear Election Assistance Commissioners,
We would all agree that HAVA guarantees by law the opportunity for voters to vote both privately and independently. One of the keys to achieving this in our present "high-tech" election system is reliable voting systems hardware.
Hardware Reliability is measured as Mean Time Between Failures (MTBF), but a more useful metric for voters would be the allowable failure rate, or percentage of voting systems permitted to fail on Election Day. Shockingly, the failure rate allowed under the Voluntary Voting Systems Guidelines (VVSG) approved by the EAC last December, is almost 10% in any 15-hour period (9.2% to be exact). To put it another way, one out of every 11 DREs or optical scanners are allowed to fail either partially or completely on Election Day, and a much higher proportion during extended Early Voting periods.
Clearly such a high failure rate, which has been known to occur in actual elections, can disenfranchise voters, impede their access to electronic ballots, and possibly even affect the outcome of elections. Certainly it also violates the spirit, if not the letter, of any law designed to ensure that voters be given the ability to vote both privately and independently, since poll worker or vendor assistance would inevitably be required in the event of a voting system failure.
So how did this happen?
The standards process that led to the current VVSG began with the Institute of Electrical and Electronics Engineers (IEEE) Project 1583 Committee, who were supposed to arrive at their recommended standards by a process of consensus. But there was no consensus reached with respect to voting systems Reliability.
As early as December, 2002, a bona fide expert Reliability Engineer from New York named Alfred DuPlessis had proposed a failure rate of only 0.03% in 16 hours, more than 300 times better than the failure rate in the 1990 and 2002 VVSG.
In public comments on the IEEE's draft, as well as the 2005 VVSG, another engineer, Dr. Stanley A. Klein of Maryland had proposed a 15,000-hour Mean Time Between Failures, equivalent to a failure rate of 0.1% in 15 hours. Although Klein's proposed standard was about three times more lax than DuPlessis', and is a higher failure rate than that of a typical personal computer, it is still 92 times better than the Reliability standard in all three versions of VVSG, including the 2005 version recently approved by the EAC for publication.
Klein's proposal would allow 1 failure per 1,000 voting systems on Election Day. To our knowledge, this compromise was never presented to the EAC by the IEEE, or even to a plenary meeting of its Technical Guidelines Development Committee (TGDC) by those who vetted the public comments on the 2005 VVSG. The result is the same inadequate Reliability standard that has existed for central count voting systems since 1990, but is now being applied to polling place machines and systems in huge numbers under HAVA: an MTBF of only 163 hours, or a failure rate of 9.2% every 15 hours.
Furthermore, this standard does not even take into account failures related to software, but rather only those related to hardware. This is analogous to building a house on a shaky foundation that only survives about 90% of the time.
There is no hope of correcting this before the 2010 elections at the earliest, unless the current 2005 VVSG is amended immediately to require vendors to build and test to a higher Reliability standard.
At a TGDC meeting on March 29, 2006, Dr. Alan Goldfine of NIST stated the following regarding this standard in a presentation on the 2007 VVSG currently being drafted: "We're not totally sure of the history of this, where the number 163 came from. The feeling - the consensus that we've got - is that it's probably too small a number." But we believe that waiting for the 2007 VVSG, which does not take affect until 2009, to finally hold the vendors accountable for producing reliable voting systems is a recipe for disaster.
We need to halt the voting systems standard development process now, before the 2005 VVSG is published in the Federal Register, and immediately adopt a higher hardware reliability standard under the guidance of NIST. Had this been done by those charged with protecting the rights of voters via the standards development process, there would be no need to approach you with such urgency. However, improving the reliability of our voting systems should be a priority of the Election Assistance Commission.
DRE Reliability: Failure by Design?
Voting Systems Batch Test Results – Reliability
The Long Road to a Reliable Voting System
Comment on This Article
You must login to leave comments...
Other Visitors Comments
You must login to see comments...