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
Steve Ansolabehere
ELECTRONIC VOTING:
IMPORTANT RESEARCH QUESTIONS
Use of electronic vote recorders has tripled over the last four years. In 2000, one in ten voters used touch screen or push button machines that directly recorded their vote in electronic format and provided an electronic tally for election administrators. In 2004, nearly one in three votes will be cast on such machines. Concerns about the security, reliability, and accessibility of such machines, however, have led many election officials to steer clear of these voting machines and have raised concerns about integrity of the electoral process in areas using electronic voting machines.
Electronic voting should not be viewed in the abstract or in isolation. It is part of a system of election administration, and it is from that perspective that I think the most fruitful questions about electronic voting may be raised.
1. How many people cannot vote or do not have their votes counted because of election administration practices? What are parts of the election process, e.g, voting machines and registration, pose the main obstacles to voting?
2. Did HAVA help? Have statewide registration, provisional ballots, and new voting equipment led to marked improvements?
3. What degree of accuracy do we require of the election system? In a large electorate, the probability that a recount reverses the initial outcome is quite low, unless the election is very nearly tied. This is a simple consequence of the law of large numbers, assuming an unbiased counting process. What is the degree of accuracy of the initial count in elections? Is there evidence of bias in the initial counts, say, toward winners or the party of the local election officer? Has electronic tabulation improved or worsened the accuracy of tabulations? I think these questions can be addressed for many systems through recounts. It is unclear how to address such questions for direct recording electronics.
4. Some of the more prominent problems documented in Florida concerned ballot designs, such as the Butterfly Ballot, that led some people to vote for someone they did not intend to. How frequent are "wrong votes"? What are the main causes? Are such mistakes preventable?
5. The existing system of testing and certification relies on laboratory testing of the durability of equipment. Many of the problems reported in the press involve difficulties with operation. Is laboratory testing appropriate? What is a suitable protocol for field testing, and how can it be implemented legally?
6. How much fraud occurs? How can researchers and election officers detect and measure fraud? What parts of the election process, such as absentee ballots or registration, are most susceptible to fraud? How does the risk of fraud through electronic voting compare with that through other parts of the process?
7. There are two levels of assurance that election administrators may give to voters - confidence that the vote was counted and confidence that the vote was counted as cast. What cryptographic solutions have been proposed for each of these problems? Do any of these proposals appear to be real solutions? How difficult is their implementation?
8. How much is trust a really factor for voters? Do people make use of existing safeguards? Does trust or confidence correlate with actual performance? A long literature in political science suggests that trust and efficacy have only modest effects on voter participation. Has the controversy over electronic voting and voting machines generally had any effect on voters' trust of the system and willingness to vote on specific systems?
9. How does electronic voting change the business of voting? Counties are used to keeping their voting machines for many years. Computers require frequent use and become obsolete quickly. What is the appropriate model for public-private partnerships here? At current levels of funding, can counties afford the switch to electronic voting over the long-run?
10. Absentee balloting is on the rise. What is the trajectory for absentee balloting in 20 years? How will absentee balloting affect the current discussion about electronic voting? For example, how do we guarantee secrecy in any system that uses absentee ballots extensively?
11. Internet voting has now been used in two primary elections to process absentee ballots, apparently without difficulty. I believe that Internet Voting will continue to grow as a system for absentee balloting. What do we need to know about the operation of Internet Voting systems in order to endorse or oppose Internet Voting? Can these systems be secured cryptographically? How does their security compare with other systems, such as banking and paper based absentee ballots?
12. How does election administration performance vary across jurisdictions?
Charles Stewart and I find some evidence that rural counties have unusually
high residual vote rates, especially when using non-paper ballot systems. Do
different jurisdictions need different methods of casting and counting ballots?