Take My Election, Please
“I miss voter fraud” – John Stewart, The Daily Show
While there have been numerous reports of voting irregularities on election day, there were no massive, widespread failures of electronic voting machines, and no evidence emerged indicating manipulation of the electronic vote totals. And so concern over e-voting has disappeared from the news. But this wasn’t a vindication of e-voting – it was more like dodging a bullet. It’s been amply demonstrated that e-voting machines are prone to failures or are susceptible to data manipulation (see Newsweek’s Black Box Voting Blues, the Washington Post’s Electronic Voting Raises New Issues, and my previous post on this).
Picture in your mind a close, hard-fought election. Imagine a successful, nefarious plot to rig the totals from electronic voting machines, and then imagine the plot is exposed after the election. It’s not possible to do a recount, because the voting records themselves were manipulated. Imagine if this happened in Florida in 2000. The tangle of state and federal rules (or lack of rules) in this unprecedented situation would provide no obvious guidelines on what to do.
So what would happen? Would there be calm and rational discussion of the options? Don’t bet on it – there’d be people in the streets, and an incredibly nasty, drawn-out legal battle, the result of which would not bestow broadly accepted legitimacy on the winner. You need look no further than the current crisis in Ukraine. The Christian Science Monitor described it as “a key test for a fledgling democracy.” It would be a test for even the oldest democracy, where there are no clearly defined procedures concerning what to do if the validity of the votes themselves are in doubt.
What’s been lost in the heated discussions about electronic voting is that it’s potentially superior to any mechanical system, so long as it leaves a paper trail. Even some of the electronic machines currently out there – when handled competently – demonstrate this potential. They can almost completely eliminate the “undervote” and “overvote” problems that lead to “spoiled ballots” which are never counted. There were 1.9 million of these in 2000, and the vast majority of them were from heavily Democratic minority districts. See the Miami Herald article Touch screens reduced spoiled ballots. In fact, it was the tens of thousands of spoiled punch card ballots in Ohio that led some to argue that Kerry would have won Ohio if it were possible to reliably count them (see Greg Palast’s Kerry Won Ohio).
However, poorly designed and implemented electronic systems are no better than poorly designed and implemented mechanical systems – see Analysis reveals flaws in voting by touch-screen. I’ve been designing and developing web applications since 1996, including systems that deal with confidential financial and personal data, so I know a thing or two about electronic security and interface design. What’s maddening to me is how badly written the software is in most of the electronic voting machines (see the Newsweek article above). I used to work at E*Trade, and compared to online stock trading (or even ATMs), meeting the requirements for a secure, reliable, recountable, and easy-to-use voting system just isn’t that hard. So why are the e-voting machines so bad?
- All decisions regarding the purchase and evaluation of these machines happens at the local level. Frankly, most of the folks on local election boards don’t have the expertise to set performance and security requirements for e-voting machines, nor do they know how to properly test the machines once they receive them. I’m not questioning the dedication of these folks, but I doubt the vast majority of them have a strong background in human-computer interface design or electronic security. Assessing voting machines isn’t like comparison shopping for a TV: not understanding what goes on inside a TV doesn’t seriously limit your ability to choose one well, but the same cannot be said for a voting machine.
- We’re currently relying on the free market to create these machines, and that’s a bad idea. A free market entails multiple suppliers competing for buyers, and over time the bad products are weeded out, the good products win market share, and overall product quality increases. For how many elections will we endure this market clearing process? How many of our elections will be handled by badly designed systems as market forces eventually lead us to the good ones? With the inner workings of the machines considered trade secrets, how can they be effectively assessed? And who exactly is the “customer” who decides whether a system is good or bad? At the very least, national standards should be established indicating a minimum set of requirements that all e-voting systems must meet. Believe it or not, right now there are no such standards. Maybe we could start with a regulatory scheme similar to what Nevada has for its gambling machines (see this New York Times editorial).
- It doesn’t take a lot of analysis to realize that a quality, nationwide e-voting system would favor the Democrats, as spoiled ballots would be virtually eliminated. I’d be willing to bet that if Gore had won in 2000, or Kerry in 2004, they would have pushed harder for the development of such a system than Bush has. Bush came into office in 2000 under a cloud, and had to at least make a semblance of an effort on this front. Two years later, the Help American Vote Act was passed. Even though a multitude of problems have been discovered in the e-voting machines it spawned, have you heard anything in the past two years from Bush or the Republican Congress about fixing them?
