"Voting Technology: Innovations for Today and Tomorrow," a two-day conference organized by the Caltech/MIT Voting Technology Project, opened with a discussion of lessons that could be learned from 2000 and applied to voting procedures this November.
The conference was held in the Stata Lecture Hall Oct. 1 and 2; it attracted faculty and students from MIT and other universities, election officials from around the country, vendors and local citizens.
Ted Selker, assistant professor of media arts and sciences at MIT, moderated the Friday morning session. The three panelists, all members of the Caltech-MIT Voting Technology Project (VTP) since its inception in 2000, were MIT professors of political science Stephen Ansolabehere and Charles Stewart, and R. Michael Alvarez, professor of political science at Caltech.
Selker described the VTP team as a relationship between technologists and social scientists that had already succeeded in getting media attention for problems in voting.
Later sessions in the conference addressed ballot design, voter authentication, security, reliability and Internet voting. Vendors offered demonstrations of voting systems. Anticipating these, Selker asked, "Do we want the government to step in and be the people who build and monitor voting machines?"
Alvarez discussed the metrics of success for the Voting Technology project itself, including higher-quality post-election reporting, improvements in poll-site voting, fewer reports of voter registration problems and an overall decline in residual votes.
"We have a pressing need for high-quality data that's consistent and publicly available," Alvarez said. Noting that voting machines don't undergo any certification process and, as voting technology evolves, there are "fewer humans in the process," Alvarez advised voters and election officials to anticipate the November election by taking certain steps. He referred to seven actions voters could take to empower themselves, and he offered suggestions to election officials. These included requiring reporting of election data and developing complaint procedures.
Georgia on his mind
Stewart's presentation involved "drilling down into the details of Georgia, the one state that has retired since 2000 every type of voting technology used nationally, providing our best and only case of what can happen," as well as raising fascinating questions about poll worker training, he said.
Four years ago, Georgia was the second worst state for lost votes--worse than Florida--and it could have been a "sorrier story, not just for residual vote rate but also for variability," said Stewart. "In one county, 20 percent of the votes weren't counted."
Counties with the highest rates of lost or residual (uncounted, unmarked, spoiled or over-voted) were smaller, less affluent and more frequently populated by African-Americans than other counties. Those counties were also the ones that made the greatest gains after the Diebold company installed direct recording electronic voting machines and trained poll workers to use them, Stewart said.
"Diebold and Georgia had a joint interest in making the election work well, and we need to find better ways of partnering state election commissions and vendors," Stewart said. But Georgia's success raised a conundrum, too, about the relative effects of training and technology.
"It's hard to evaluate improvements in technology. The voting process and the ballot were logically thought through in Georgia. Also, Diebold put nearly 6,000 poll workers into the field and did lots of training. But was it the technology or the process that really drove the gains for voters?" Stewart asked.
The Caltech/MIT Voting Project is a collaboration between the two universities that aims to prevent a recurrence of the problems experienced during the 2000 U.S. presidential election. Specific tasks of the project include evaluating the current state of reliability and uniformity of U.S. voting systems, establishing uniform attributes and quantitative guidelines for performance and reliability of voting systems, and proposing specific uniform guidelines and requirements for reliable voting systems.
A version of this article appeared in MIT Tech Talk on October 6, 2004 (download PDF).
