Skip to content ↓

Far from perfect

With election days away, Stewart charts progress of U.S. voting security

In the run-up to the Nov. 4 presidential election, the News Office has asked MIT experts to weigh in on the presidential candidates, their policy ideas and aspects of the campaign. In this final installment in the series, Charles Stewart III, the Kenan Sahin (1963) Distinguished Professor of Political Science and head of MIT's Department of Political Science, examines issues of voting security.

Q: In light of concerns over voting in 2000 and 2004, how secure are voting systems for 2008?

A: The more experience we have with voting systems, the more we realize there's a real distinction between security and reliability. There have been some well-publicized cases where teams of experts have exposed security vulnerabilities with electronic systems, and there have been other cases where researchers have "hacked" into systems. But there still isn't any hard evidence of major security problems emerging in actual elections. However, there continue to be an unnerving number of cases where systems are shown to be unreliable. For instance, Premier Election Solutions (formerly Diebold) recently reported that a bug in the software that accepts and counts election results from individual voting machines has a flaw that can result in some ballots being dropped from the system before they are counted. It's a stretch to call this a security problem, but it's not a stretch to call it a reliability problem.

The reliability problems aren't just with electronic voting machines. The Premier Election Solutions example applies to optically scanned ballots as well as to electronic machines.

I still am confident that votes are counted better now than they were in 2000, but we still have a long way to go before anything close to perfection is achieved.

Q: Have any significant gains been made in terms of security of voting machine or paper ballots?

A: I think that most computer scientists would say that no real gains have been made with the electronic systems because they are inherently insecure and unreliable. Even the addition of "paper backups" to electronic machines hasn't been a panacea. (Again, the Premier example I mentioned before is a good example -- it is possible to compare the paper reports generated by the voting machines with the electronic versions that remain after the download. However, almost no one has been doing this comparison -- and these are the professional election administrators.)

The physical security of paper ballots (and of electronic machines) is probably better because states are now more aware of the need to establish a "chain of custody" of voting machines and ballots. For instance, in a lot of states, officials used to allow poll workers to take the machines home the night before the election. These so-called "sleepovers" for the voting machines made it easier for the poll workers to get the precincts opened for voting on Election Day, but raised questions about opportunities for tampering. Sleepovers are going the way of the dodo.

So, again, I think things are better now than in 2000. However, we are much more aware of how informally elections are run in the United States, which continues to provide Election Day horror stories.

Q: What innovations do you see coming in terms of voting systems?

A: I continue to be amazed that the major voting machine companies don't adopt an open source model of voting machine software. Computer experts will tell you that open source election software won't guarantee security, either, but it would allay fears of the public, and probably allow the most egregious software errors to be caught. One of the barriers to the development of a robust electronic voting machine market is the lingering distrust that many elements of the public have in the quality of the software.

Q: Should the nation adopt a system of voting identification cards to protect the election process?

A: It's pretty clear that voting identification has become highly politicized, with Republicans believing that more stringent voting identification will end high levels of fraudulent voting, and Democrats believing that these ID laws will intimidate minority voters and disenfranchise the elderly. Republicans love these laws and Democrats hate them. It's pretty clear that both sides have engaged in a great deal of hyperbole in making their cases. Accusations of large-scale fraudulent voting almost always dissolve. And, while it is certainly true that certain types of people have a harder time getting ID cards (such as the poor and the elderly), I have a strong suspicion that the lack of identification is nowhere close to the main reasons these folks don't vote.

That said, I've always believed that there is a compromise that would achieve what both the left and the right want achieved. Have every state automatically register every 18 year-old to vote and send them a vote ID card. The left would get universal, automatic registration; the right would get a voter ID card. But, that's my own personal pipe dream.

Q: If you suspect a problem while voting, what should you do?

A: It depends on what the problem is, but in general, you should talk to the poll worker who is in charge of the voting precinct right away. Don't wait until you've checked out, because there's no way to undo a vote once you've checked out of the precinct. If there are registration issues, voters may also insist that they be given a provisional ballot, allowing them to resolve the registration problem the next day at the local election department (or town clerk) office. I, myself, have observed one precinct in Cambridge where people were just turned away from voting when there was a minor registration question, rather than even being told they could cast a provisional ballot (to be clear: these were people who were registered, or thought they should be on the registration list, but there was some question about the registration). When poll workers say there is no way to clear up registration problems on Election Day, they aren't following the rules.

A version of this article appeared in MIT Tech Talk on October 29, 2008 (download PDF).

Related Links

Related Topics

More MIT News