The rest of the United States could learn a lot from my home state of New Jersey as it’s gone through its series of ups and downs in the process of coming up with a workable electronic voting system. Recent evidence that our new voting machines, manufactured by
Sequoia Voting Systems, seem to have trouble keeping track of how many people actually voted is only the latest in a series of problems that, at best, are a textbook example of poor hardware, software and user interface design practices.
Powered by Bronze Age (actually 1970s) Z80 processors and lacking modern Flash technology to store critical permanent information (they use easy-to-tamper battery-backed CMOS), about the only good news about these $5000 machines is that their design flaws and security holes are so obvious and ill-conceived that it’s tough to imagine they are the result of some shadowy conspiracy to rig votes. Sequoia’s recent heavy-handed efforts to suppress a non-partisan inquiry into the well-documented anomalies in our February 5 primary election do, however, raise worries that they are trying to cover up systemic inadequacies in their equipment that would not be tolerated in even the cheapest consumer goods.
New Jersey troubles with Sequoia began soon after the machines arrived in 2004 without any capability to provide a verifiable paper record of the votes they counted. At first, Sequoia failed to take the state’s request to add paper trail capability seriously, and it took a 2006 court order to force them to commit to delivering a printer retrofit by 2008. Unfortunately, a combination of foot-dragging on Sequoia’s part and the Byzantine workings of outdated NJ government procedures have stretched out the process so that it’s highly unlikely that the printers will be in place for the 2008 Presidential election this fall. This would not be quite so worrisome, except that data collected from the machines after our fall primaries raises real questions about how accurately the machines record citizen’s votes in their electronic memories.
The irregularities are well-documented by
Dr. Ed Felten, a professor of computer science at Princeton University, and director of the school’s
Center for Information Technology Policy. In his personal blog,
Freedom to Tinker, Felten lays out a fascinating story of what happened when at least five New Jersey counties using Sequoia AVC Advantage voting machines displayed anomalous results during the February 5 primary election.
After several poll workers noticed differences between the number of people registered in a particular party who voted on a particular machine and the number of ballots for that party cast on it, Dr. Felten used public record requests to get the
vote tally summary tapes generated by 15 of the 60 machines reported to have known problems. In Cranford, for example, a machine reported that 55 Republican voters had cast 57 ballots. You can look at the printed tapes that document the anomalies, as well as a published response from Sequoia that offers one possible
theory of how they could have arisen.
An initial analysis of the Sequoia miscounts indicates that they occurred due to an error condition generated by an incorrect sequence of button pushes on the voting machine operator control panel that is used by the poll worker to activate the machine for each vote. Even Sequoia admits that the error condition is not clearly indicated and that it still has the potential to let a vote to be mis-recorded. Unfortunately Sequoia’s
analysis places the blame on operator error and fails to acknowledge that the combination of a poor user interface and non-existent error-handling in their design are what make the machines vulnerable to a casual button push in the first place. Felten’s March 20, 2008
blog entry makes a good case that, even if Sequoia’s explanation accounts for the errors, it raises grave questions about what other hidden error states lurk within the machines, and whether they can be trusted to deliver an accurate vote count without a verified paper trail.
Apparently, our State government officials felt the same way, because the counties wanted to commission Princeton’s Information Technology Policy Center to do a study of the anomalies and what caused them. Union County even offered to supply several machines to analyze as well as all their configuration files, documentation, and raw poll results. This came to a screaming halt when Sequoia got wind of the proposed study and used the old “opening up our machine and looking at our code constitutes IP infringement” argument in a letter to Felten which threatened dire legal consequences if the Center proceeded (The actual letter is posted on
Freedom to Tinker.) While not fully documented at this time, there are reliable reports of similar letters being sent to county officials asking to back off.
This might have been the end had it not been for an unexpected windfall in the form of a recent decision from another
lawsuit against New Jersey, filed back in 2004, which contended that the use of paperless voting machines was non-compliant with New Jersey law and that the machines had not been properly certified in the first place. Then, in March, the presiding judge
ruled that the prosecution should be given access to the details of the voting machines’ construction and programming, and is now expected to issue a similar subpoena for Felten’s group to allow them access as well.
With a little luck, and a bunch of persistence, the 2008 elections will be the last time New Jersey residents have to wonder whether their votes were actually counted or if their votes were counted against other imaginary votes that were generated inside the guts of a poorly-designed, poorly-tested voting machine. It will take some more work to ensure we have the same guarantee to fair elections throughout the United States, but the tide does seem to be turning towards democracy again. Despite the failure of the
Senate 2007 Ballot Integrity Act (S1487) to get out of committee, the
Congress Election Reform Bill (HR 811) is alive, well and scheduled for debate in May of this year.
As electrical engineers who are more aware of the potential frailties of electronic voting systems than most, it’s our duty as citizens to make sure that this important legislation makes its way out of committee and on towards becoming a law. Be sure to take a few moments to contact your local representatives to let them know how you feel.
Remember, the democracy you save could be your own.
Comments? Questions? Interesting new applications for recycled paperless voting machines? Write me at
lhg at en-genius dot net or post your comments on our blog.
