The Punk Engineer

What's REALLY wrong with electronic voting machines

A thread on Slashdot about Wisconsin's new law requiring paper trails and open-source software for touch-screen voting machines has rekindled some thoughts I've had about voting machines.

The main thought, of course, is that the paper trail we normally associate with this sort of thing--namely, a printed receipt that says, "Yes, you DID vote for John Kerry"--is a complete MacGuffin and is essentially irrelevant.

And here's why.

First, and most obvious to me (but apparently not to most commentators), is that if someone can get in and program the voting machine to prefer one candidate can easily program that machine to print out a "voting receipt" that tells each voter that, "Yes, you did vote for your candidate," when in fact the vote was recorded for the other guy.

It doesn't matter whether the machine is a touch-screen type (like an ATM) or the fill-the-circle-and-scan-it (like taking the SAT) type we use here in Tucson, or in fact any other computerized thing you can come up with:

It can be programmed to print out the vote you cast, while at the same time recording your vote incorrectly.

The second very important point that's not noted by most commentators regards the recount trigger. Recounts are invoked in only the closest elections, say where the difference in vote tallies is less than a percent of the electorate.

(Everybody knows this scenario: the election manager pushes the touch-screen voting machine's "recount" button and it gives the exact same answer it gave in the first place. That's not very helpful. At least in the case of the scanned ballots, the pollworkers can go back to the original ballots and count them by hand.)

Now, what happens if the machine tells you the election result is not close? Well, there's no recount. Why is this important?

Because if the machine can be programmed to count, say, one of every five votes cast for Candidate JK as instead cast for Candidate GWB, the bogus votes may be enough (especially in a close election) to skew the results beyond what would trigger a recount.

Think about that. Really. Think about it. In this scenario, paper trails don't mean a goddamn thing.

It doesn't matter whether the machine uses a touch screen or an optically-scanned paper ballot. The vote is skewed enough toward a particular candidate to avoid a recount (that's the key), but not so much as to draw suspicion (say, if exit polls prefer the loser by a slim margin).

Think about it.

Some argue that a solution to this problem is, as mandated by the Wisconsin law, the use of open-source software in the voting machines. For the most part, I agree, but it doesn't go far enough and there are some problems with it.

In support of the open-source argument: I think that companies like Diebold who claim "trade secret" as a reason for not opening their voting-machine sources are basically full of shit. Why? Because any high-school student with some decent programming smarts can write the code for a voting machine in an afternoon. It's not rocket science. Really. If you require (and you do) security, have a grad student with a strong interest in cryptography write the code, and let the community vet it.

The problem with the open-source solution is not really a problem with open-source per se, rather it's the question of, "who certifies the software, and who ensures that the certified code is loaded into each machine?" It's a logistics problem.

OK, so what do I recommend? Good question. It's important to remember that neither the Constitution nor any local/state/federal laws require instantaneous election results. Democracy can wait a couple of weeks before results are announced. Hey, if hand-counting ballots is good enough for Iraq and Afghanistan (the Bush Administration's two "experiments in spreading democracy"), it oughta be good enough for the U.S. right?

So why not count ballots by hand? Make it a civic duty, much like serving on a jury. Representatives from all participating parties would of course be present for the counting and would be allowed to challenge votes.

Hand counting ballots will be time consuming and potentially expensive (although how much do those certified voting machines cost?), but our democracy is worth it.

Posted by Andy at January 4, 2006 06:15 PM