WORKSHOP ON DEVELOPING A RESEARCH AGENDA FOR ELECTRONIC
VOTING TECHNOLOGIES
September 17-18, 2004
Convened by the American Association for the Advancement of Science
Main | Program | Participants | Synopses
Henry E. Brady
RESEARCH QUESTION RELATED TO ELECTRONIC VOTING TECHNOLOGIES
Although most of the discussion about voting technologies now centers on electronic voting systems and their security vulnerabilities, it is important to remember that national attention to voting systems began with the issue of accuracy in the 2000 election in Florida where there were no electronic systems used to capture people's votes. Most of the vote recording systems were either punch cards (Votomatic style or DataVote) or optical scan. Moreover, it seems sensible when asking about electronic systems to ask "compared to what?" Hence, any set of the "most pressing research questions related to electronic voting technologies" should consider all voting systems-at least all the ones that are widely used and likely to persist into the future such as optical scan, electronic, and punchcard (although only DataVote may have a future).
In addition, the term "electronic voting systems" is somewhat ambiguous. In some discussions, the term refers to the actual voting device on which people record their vote which I will call the voting (or vote recording) system. There are five vote recording systems in use in America (paper, lever, punchcards, optical scan, and electronic), only one of which is electronic. In other discussions, the term is also used to refer to the systems that count the votes which are almost always electronic except in the case of paper ballots and lever systems, although even votes counted using these voting systems are often entered into computers for the final tally. I will call these vote counting systems.
I find it useful when thinking about vote recording systems and vote counting systems to consider six criteria:
Accuracy in recording and counting votes - How well does the system capture
voter intent? Does it clearly indicate when a mark has been made? Does it
provide feedback to the voter that a mark has been made? Does it summarize choices?
Does it prevent overvotes? How well does the system count votes? Does it read
marks appropriately? Does it include cross-checks for accuracy? Is there an
audit trail?
Accessibility-How accessible is the system for people with disabilities and other barriers to voting? Can it be used by the blind, the disabled, language impaired? Can it be used by those speaking foreign languages?
Anonymity and Privacy-Does the system ensure that the vote is anonymous? Does the system protect people's privacy by divulging the least possible information while ensuring accuracy and accessibility?
Reliability-How reliable is the voting system when used in the rough-and-tumble conditions of an American voting precinct with poll workers who, at best, have only had a day or two of training? Is it hard to set-up, to use, to explain to voters, to run properly? Does it depend upon special technical knowledge or technical capabilities of the site? How reliable is the vote counting system? Is it vulnerable to programming errors? What kind of expertise does it require and what kinds of errors can be made?
Security-How vulnerable are the vote-recording and vote-counting systems to fraud or mistakes? Is there an audit trail? How likely is it that the vulnerabilities will be exploited? What is the overall risk (vulnerability times likelihood of its exploitation)?
Legitimacy-Is the overall system perceived as legitimate by citizens? Do they believe that the results are accurate?
Based upon this background, I think that some of the most pressing research
questions are:
1. How can we measure the accuracy and accessibility of voting systems? How
can we test systems to make sure that they accurately capture voters' intentions?
What kind of accessibility standards should we have? How can we create a monitoring
system that will make sure that performance standards are achieved? How can
we do this while maintaining anonymity and privacy?
2. How can we go beyond the standard certification process to ensure that voting
and vote counting systems are reliable? What kinds of reporting system do we
need to record incidents where systems have not performed reliably both in precincts
and in central counting locations?
3. How can we estimate risk and not just vulnerability? How can we estimate
the likelihood that vulnerabilities will be exploited? Does open source software
or a "voter verified ballot" increase or decrease vulnerability? Do
they increase or decrease the perceived legitimacy of the voting systems?
4. What factors affect the perceived legitimacy of voting systems? How can we
create an institutional structure that reassures voters and convinces them that
their votes have been properly recorded and counted?
5. To what extent does poll worker and election official training affect the
quality of election administration? Indeed, how much are the performance problems
with the current system due to inadequate training and faulty administration?
Do electronic vote recording or counting systems pose special problems for election
administration?
[an error occurred while processing this directive]