[an error occurred while processing this directive]

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

Michael Shamos

RESEARCH AGENDA

A fundamental problem with electronic voting is that the public, for the first time in 25 years, has been persuaded by computer scientists that DRE machines cannot be relied upon. A critical research agenda, therefore, is to develop sound means by which the machines may be made trustworthy that are readily understandable by the public.

Technologies to be investigated include:

Auditing hardware and software to make an independent record of votes cast.
Self-diagnosing systems built from observer chips to monitor system operation.
Methods for detecting systems changes or demonstrating that no have been made.
Voter-verifiable audit trails not dependent on paper records.
Separation of candidate names, parties and office titles from the vote-casting and recording mechanism so no useful tampering is possible.
Secure software distribution mechanisms so that uncertified software cannot be installed in a voting machine.
The role of open source software methodology in reducing errors and increasing public trust.
Cryptographic methods allowing universal post-election verifiability.
Thoughtful investigation of Internet voting proposals, including Project SERVE, to pave the way for what is an inevitable method of voting.
A reliable system of reporting and investigating irregularities in DRE voting, including thorough forensic examination.

The standards by which DRE machines are evaluated are inadequate and out-of-date. What is needed is not a static set of standards or tests, but a process by which such standards can be kept current.

The DRE debate has raised some fundamental computer science questions, about which experts disagree, especially the question whether it is possible to tamper with software in a way that is fundamentally undetectable. Essentially no research has been done on this question or on methods for detecting changes to software.

A much more pressing problem than DRE security is the inability of millions of voters to vote on election day because they are outside their home jurisdiction and have not been able to comply with absentee procedures. The Federal Voting Assistance Program (FVAP) estimates the number of voters so disenfranchised at over 5 million for each Presidential election. Far more effort needs to be devoted to developing a sound remote voting mechanism usable by all overseas Americans.

There currently exists no field of "voting science," which would encompass voting technology, policy studies, human-computer interaction, election law, accessibility, computer security, voter identification methods and all other aspects of the complex process of voting.

[an error occurred while processing this directive]