In the past two days, the New York Times has featured two interesting pieces on voting technology: (1) this skeptical take on electronic voting machines; and (2) this proposal to monitor fraud via “randomly selected ballot receipts.” Both are worth reading.
For political science research on voting technology and related subjects, see the indispensable Cal Tech-MIT Voting Project, overseen by Michael Alvarez, Jonathan Katz, Ron Rivest, and Ted Selker, with participation from others as well (the full cast is here).
Among the many interesting papers on the site is this oldie-but-goodie. Written in March 2001, it neatly presaged the current controversy over electronic voting machines. In the 1988-2000 presidential elections, both electronic machines and punchcard systems produced significantly more over- or under-votes compared to lever machines. By contrast, traditional paper ballots or ballots did well by this criterion, as compared to level machines.
Thus, the most old-fashioned systems — levers and hand-marked paper ballots — appear to function better than the electronic machines privileged in the Help America Vote Act of 2002. Given the paucity of lever machines still in use, paper ballots seem the best stopgap for 2008.