MIT IST News
The 2000 presidential election is remembered for Florida’s hanging chads and highly controversial recount. The 2004 presidential election was similarly contentious, with concerns about improper voting procedures in several states, including the swing state of Ohio. Aspects of the entire voting process were called into question, from voter registration, to the unequal distribution of voting machines, to the accuracy of the count.
Will the voting process be fairer and the final tally more accurate in 2008? And how will voting technology come into play?
Professor Charles Stewart III, Head of the MIT Department of Political Science and a member of the CalTech/MIT Voting Technology Project, gave is&t an overview of what factors might affect election results in 2008. The Voting Technology Project, established in the wake of 2000’s controversial recount, evaluates the reliability of U.S. voting systems and proposes principles for the design of new voting technologies.
According to Stewart, there has been progress since 2000, but because each state manages its own voting process, there will never be uniform national standards.
In 2002, Congress enacted the Help America Vote Act (HAVA), which funded the replacement of mechanical lever and punch-card systems (goodbye, hanging chads). Some precincts bought direct-recording electronic (DRE) voting machines, which display ballots to voters on touchscreens. A voter’s electronic input is tabulated by software, and often backed up by a paper audit trail that the voter can verify. Optical scan systems also continue to be popular: individuals mark their votes on paper ballots, which are read into electronic scanners that tally the results. The paper ballots can be used for manual recounts, if needed.
In addition to this move to modern voting equipment, Stewart notes that the National Institute of Standards and Technology (NIST), in partnership with the U.S Election Assistance Commission (EAC), has developed voluntary voting system guidelines. While states don’t want to cede control of elections to the Federal government, they do recognize the value of striving for standards. As a consequence, most states operate voting machines certified by the EAC.
So all is well? Not quite.
One Country, Fifty Standards
The design of paper and electronic ballots isn’t uniform. Each state (or in some cases, each municipality in a state) creates its own ballots. Usability testing isn’t required, so on Election Day some voters may still get confused by unclear ballot layouts.
What's more, with 50 different state standards for voting, election software needs to be customized for each state. This lack of uniformity can lead to errors in the software, which is usually written by one of the two large vendors of voting systems – Premier (formerly Diebold) and eSys. These systems are proprietary, which doesn’t encourage either innovation or streamlined code.
Stewart notes that Premier recently discovered a flaw in its software for the 2008 elections. The vendor has come up with a fix, but it requires election officials to follow a multi-step correction procedure. This is far from ideal, since most staff at the polls do not have extensive computer skills.
The Voting Technology Project recommends standardizing all the components of voting systems – CPUs, touchscreens, and scanners. This would allow smaller vendors to compete and could lead to improvements in voting technology. But this recommendation hasn’t won favor with election officials, who prefer to deal with one vendor.
Counting the Votes
Stewart acknowledges that hacking into electronic voting machines remains a risk, and that physical security measures are still the main means of protecting these systems.
He also warns that paper audit trails can instill a false sense of security. Experiments that have tried to reconcile paper trails with e-votes do not match up. Most of the discrepancies are likely due not to hacking but to mechanical errors. Paper audits are produced by printers, which can break, get jammed, and so on. Meanwhile, voters who can view their paper audit in the voting booth often find it cryptic to read and don’t verify its accuracy.
If proprietary voting systems have their share of problems, what about Internet voting? Why isn’t that an option in 2008?
One primary concern is the security of the channel. Some people draw an analogy between the U.S. banking system, which handles millions of transactions a day, and online voting. But this analogy doesn’t hold. The banking system has built-in redundancies and, even more important, it is not anonymous. Anonymity and voting go hand in hand. You should be able to vote, but the system should not track who you voted for. For now, anonymity online means that votes cannot be verified.
Internet voting also raises concerns about maintaining the secrecy of votes. If you vote at home (or in a nursing home), what guarantees that you were the person who voted? Others could co-opt your vote or coerce you to vote a certain way.
When the Voting Technology Project got under way, says Stewart, "We thought we could solve voting problems by designing the best-ever voting machine or creating the right standards. With time, we’ve come to appreciate that the primary problems are around process and rules. Technology is not the panacea. The key to success is to design voting systems that interact with human beings."
Current MIT participants in the project, in addition to Stewart, include faculty member Ron Rivest and affiliates Steve Ansolabehere and Ted Selker. Rivest is an expert in the fields of cryptography and computer and network security. Ansolabehere examines the role of rules, like voter ID, in the election process. Selker’s research focuses on aspects of voting, including accessibility, auditing, and methodology.