Our creations have taken over and, unchecked, are deciding your fate. The next presidential election way be less the people’s choice, than the voting machine computer’s whimsical recording of that choice.
This grim scenario is painted by Edward Felten, professor of computer science and public affairs at Princeton University, who has monitored our voting machines and testified before the Supreme Court as to their accuracy. Felten details the electoral process’ problems and challenges at the Princeton Library’s Tech Talk event, “E-voting Machines: Ready for Prime Time or Not?” This free talk takes place Tuesday, October 7, at 7 p.m. in the library. Visit www.princeton.lib.nj.us.
For most of us, voting seems simple. It means tapping our choices, punching the “Cast Vote” button, then walking away with serene confidence that we have made our mark. But there follows a technical, electronic side that very few individuals comprehend as well as professor Felten.
A native of Madison, Wisconsin, Felten attended the California Institute of Technology. After earning his bachelor’s in physics there in 1985 Felten stayed on in Cal Tech’s research laboratory. He later did some commercial research, before gaining his Ph.D. in computer science from the University of Washington. For the last 15 years Felten has been at Princeton, where he also is the director of the university’s Center for Information Technology Policy. His blog, www.Freedom-to-Tinker.com, carries information on everything from voting machines to Sara Palin’s E-mail.
In a race for speed and reduced voting budgets, most counties nationwide have switched from paper and machine booths to some form of electronic voting. Does it provide better accuracy? Does Felten think we will do better than the debacle of the Gore/Bush election of 2000? “We might do better,” he says. “It’s a possibility, but there are no definite indications.”
Electronic accuracy. Most of New Jersey’s voting machines now, as in 2004, are products of Sequoia Voting Systems Inc. The company claims a much greater time and cost advantage over what it terms “the slow paper ballot or antiquated machines” formerly in use. Using Sequoia’s AVC system, the voter interacts with the machine by touching a glass panel covered with paper. The pressure-sensitive panel responds to the selection by lighting the appropriate green “X.” Upon hitting “Cast Vote” the decisions are recorded into the machine’s computer and tabulated into the whole count.
Everything depends on the computer’s memory functioning flawlessly. At no time is there a hard copy or evidence of the vote, other than on the magnetized recording. If the electronic system tabulates smoothly, Barack Obama or John McCain each stand a strong chance of stepping into the Oval Office. If a power surge sets the recorder back four years, a stunned George W. Bush might just find himself swept in for a third term. While such an egregious electronic miscarriage might be no more remotely possible than, say, the selling of Merrill Lynch, lesser errors are frighteningly common.
“Computers tend to misbehave and errors, as we all know, are routine,” says Felten. During this past February primary, 16 of new Jersey’s 21 counties discovered some very puzzling discrepancies. At days end, each election machine prints out a paper tape saying the number of voters, and the number of votes for each candidate listed and written in. It is legitimately possible for the candidates’ total votes to be less than the overall total of votes cast.
Individuals occasionally get into the booth, think better of the whole thing, and walk out without voting. This leads to what is called undervotes. But in these 16 cases, the candidates totals were greater than the computer’s version of the total votes cast.
“This represents a true problem,” says Felten. “In the case of elections, with every voter’s ballot is a secret, there is no way of checking or rectifying a discrepancy.” In the end, it was determined that the flaws were the result of a Sequoia programming error. They could fix the blame, but not the process. Felten notes that New Jersey’s reliance on such machines has placed it among the 12 worst states when it comes to voting accuracy.
The human touch. On the evening prior to the last presidential primary, professor Felten took a tour. He strolled into schools, town halls, and the various voting areas through Middlesex and Mercer counties. What he found was that most machines were unguarded, sitting in unlocked rooms.
“Tampering would not be all that complex,” says Felten. “It would take, suffice it to say, an expert in computer technology to do the programing, but he would only then have to get at the back of the machine and switch the computer chips.” The necessary level of expertise might be found, Felten admits, in a very bright member of Best Buy’s Geek Squad.
Yet while computers are coming under justifiable scrutiny, Felten notes that humanity represents an almost equal source of error. A 100 percent accurate vote count remains an heretofore unachieved ideal. Clerks routinely make counting mistakes — whether more or less than the computer is up for debate.
Better ways. Rebecca Mercuri, another voting machine expert, has through her Robbinsville company, Notable Software, developed the Mercuri method. In this machine, the voter makes his choices, then views them recorded on a paper ballot beneath a thick plate of clear glass before he pushes the final “Vote Cast” button.
A similar method has the electronic vote read and scanned onto a paper record, which allows straight line traceability to the computerized tally.
Felten notes that the good news is that New Jersey has finally joined California and several other states mandating voter-verifiable paper trails for touch pad voting machines. The bad news is that it will not take effect in all counties until after next month’s presidential election is over.