April 23, 2004

Some running notes from the final plenary

NOTE: These notes don't cover any of the questions and answers after the panel members' remarks.

David Wagner, UC Berkeley CS Dept
Wagner discussed how to deal with what he considers the fact that today's paperless DREs are unable protect the integrity of our elections. In his opinion, we need to ban paperless DREs or move to something better.

He gave some computer science background to the options with a metaphor: pots versus laptops. Pots don't crash. Not so with software. A pot is simple and has no moving parts. Computers have millions of transistors and run on millions of lines of code; there is no way to prevent inadvertant bugs or spot every maliciously introduced trojan horse.

Wagner brings up the Therac 25, which was a medical irradiation device that was controlled solely by software, and ended up resulting in 6 deaths when it exposed patients to 100 times the proper level of radiation. "If you're relying solely on complex software systems for safety critical functions, you're playing with fire."

Douglas A. Kellner, Commissioner, New York City Board of Elections
"If you can't count the votes in public, it's not a suitable voting technology. Period." But this removal from public view, according to Kellner, is the effect of paperless electronic voting machines.

The experience in NY: the city was the first to call for electronic machines, back in 1988. There were "dozens and dozens" of safeguards built into the process. He became an election official in 1992, and found the safeguards hard to crack but that those were mostly eroded in the rush for national paperless elections. NY stopped its contract in the 1990s. NY, if it goes electronic again, will have a paper trail.

Kim Alexander, California Voter Foundation
CVF is still urging the decertification of all paperless systems. The voting systems panel unanimously voted to decertify the Diebold TSx, which represents 75% of Diebold's stock in CA. If Secretary of State Shelley accepts the recommendation, the TSx won't be used in November.

Reasons that Alexander gave for the board's decision:

  1. Problems with the machines in March elections led "directly to disenfranchisement." There were problems at hundreds of voting places in Alameda, San Diego, and Orange Counties. There were also problems with the wrong electronic ballot being distributed, as well as problems with Sequoia and Diebold optical scan ballots. The latter miscounts were discovered because the counties could count the paper ballots and recalibrate the scanning machines.

    1. "The life of a paperless ballot."
    2. The voter casts the ballot on a machine.
    3. At the end of the day, all machines are tiedtogether, their results are downloaded.
    4. A central machine counts and reports its count.
    It's unclear whether the problems observed so far are mistakes or intentional.

  2. Government regulation of electronic voting so far has been poor.

  3. A lack of confidence in elections undermines the legitimacy of elections, which undermines the legitimacy of a government.
  4. Mike Shamos, Professor, Carnegie Mellon University
    Whatever the problems with paperless machines, paper is not the answer, according to Shamos. Every election that has been manipulated has been manipulated with paper. There has never been a proven case of election manipulation with electronic system. This is different from failed elections due to bad equipment, evidence of which is plentiful.

    He cites a 20 year history of success with DREs in Pennsylvania; the problems elsewhere are engineering problems. The question is what probabilit y of error we are willing to accept in the risk of miscounts and undetected tampering. The way forward is through testing; the interested parties must decide how to test the systems to be satisfied that they operate satisfactorily (not perfectly).

    Shamos concluded by stating that paper audit trails aren't the only solutions to DRE concerns; independent equipment manufacture and auditing are probably.

    Scott Konopasek, Registrar of Voters, San Bernardino County, California
    If voting machines are adequately administered, Konopasek is confident that DREs can be secure. Human beings, however, are the most vulnerable part of the integrity of elections.

    He is in favor of better verification tools. But here is his position on paper: VPAT is necessary, but a duplicate paper ballot will not solve any problems.

    Dan Tokaji, Assistant Professor of Law, Ohio State University, Moritz College of Law
    Focus on: equality. The right to vote is "sacred" and "fundamental to democracy."
    More acronyms: "contemporaneous paper record," or CPR. This is a synonym for VVAT. Relying upon these measures without proper testing is a "recipe for disaster."

    Dimensions of the debate:

    1. Technology, especially the possibility of malicious code
    2. Election administration. Procedures exist or could be implemented to improve integrity of
    3. Voting rights, which have 4 aspects. (wrt all, electronic voting systems are superior).
      1. In CA primaries, 170,000 votes were lost to hanging chads. Paper-based systems lose 1000s or even hundreds of 1000s of votes. Optical scan machines do considerably better.
      2. Electronic voting machines do substantially better in reducing racial disparities.
      3. Disability. This community is standing up for its right to vote and vote secretly.
      4. Multilingual access: again, electronic machines are superior.

    A lack of familiarity with election admin has led to four mistakes:

    1. The CPR is very hard to implement.
    2. The value of CPR has been overstated and won't resolve the problems identified with electronic voting.
    3. An underestimation of the efficacy of current measures.
    4. An overestimation of the integrity of current paper-based systems.

    Posted by aaronb at April 23, 2004 04:42 PM
Post a comment

Remember personal info?