Tuesday, July 17, 2001

MIT and Caltech Researchers Propose Changes in Voting Technology By FLORENCE OLSEN In a report released Monday, researchers at the California and Massachusetts Institutes of Technology propose voting-technology reforms that they say could avoid millions of "lost" votes in 2004, but would require spending 50 percent more than is currently spent on elections. The report also recommends a national research-laboratory program. Industry and academic laboratories, including ones operated by land-grant universities, would conduct tests on new voting equipment and ballot formats before they are certified by election officials for use in real elections. The new report, based on a six-month study, says the 2000 election data show that approximately 1.5 million votes for president were intended to be cast but were not counted because of faulty equipment and confusing ballots. Punch-card, lever, and older electronic-voting systems should be replaced in every precinct either by optical-scan systems or by newer electronic-voting systems that have been field tested, the report says. The report also says that up to 3 million lost votes could be avoided with better procedures for managing voter-registration databases and with generous use of "provisional" ballots for voters who encounter registration problems at the polling sites. "This report will have a big impact," says Kim Alexander, president of the California Voter Foundation, which studies how new technologies can improve democracy." The 92-page report looks at administrative data about American elections, including data which have never before been collected and analyzed. "It's the first major contribution in the voting-technology debate to come out of academia," Ms. Alexander says. The report presents results of research conducted in engineering laboratories at M.I.T. and Caltech to discover better design principles for future electronic-voting systems. It gives a set of guidelines for making electronic-voting equipment more reliable than some of the older Direct Recording Electronic, or D.R.E., systems in use today. Electronic-voting systems should be "modular," or made up of separate machines that work together and are simple enough that election officials and the public could check their accuracy, the report says. Moving to a system that partly relies on PC's at polling places is reasonable, the report says. But it adds that Internet voting by citizens from their home or office PC's is still far too risky because no one has yet figured out how to prevent hackers from committing voter fraud on a large scale. The report also recommends restricting absentee voting, which election officials say has gotten out of hand in recent years. And it advises expanding early-voting options at official sites to reduce the risk of fraudulent votes and votes cast under coercion. Using future electronic systems envisioned in the report, a voter might check in at a polling site and be given an electronic-memory card containing a blank ballot, the voting-precinct number, and the name of the election official in charge. The voter would step into an enclosed booth, insert the memory card into a slot in a PC, and the ballot would then be displayed on the screen for the voter to mark his choices. Next, the voter would carry the memory card to a separate enclosed station, where simple and inexpensive card-reader equipment would display in standard text format the voter's choices. After checking to see that all of his choices were correctly made, the voter would press a "vote" button. At that point, the memory card would be "locked" and kept for the purpose of counting and auditing the election results. If the voter noticed an error or a change that he wanted to make before pressing the "vote" button, he would retrace his steps, reinsert his card into an available PC, and make corrections or changes. Once the voter pressed the "vote" button on the separate card-reader/card-writer device, his vote would be cast. At that point, the vote would be sent electronically to another machine to handle the process of recording and counting the votes. The report notes that advances in cryptography, such as digital certificates, make it possible to verify that electronic votes have been recorded and counted correctly and have not been changed. The guidelines say that proprietary software could be used in the PC's that display the ballot, but that only "open source" code, which can be publicly audited, should be used in each of the separate devices that record and count votes. According to the guidelines, each separate part of the voting system must also keep a log of all activity, including maintenance, within each machine. Because the PC's would be needed only a few times a year for polling purposes, the report says, those same computers could be used the rest of the time by public-school districts. Background article from The Chronicle: Computer Scientists and Political Scientists Seek to Create a Fiasco-Free Election Day (4/20/2001) Easy-to-print version E-mail this article

Stem-cell researcher will leave the U.S. for Britain, where regulations are fewer Ousted president of Sinte Gleska U. wins reinstatement Dean at Central Methodist College is charged with setting 2 fires on campus Canadian lottery officials issue warning to professor who questioned their payout claims Pakistani students clash with police over appointment at U. of Balochistan M.I.T. and Caltech researchers propose changes in voting technology Online instructor cautions against having too many activities