Election technology: There's only one problem.

If the issues surrounding the introduction of new election technology seem confusing, the problem isn't in your head. We're trying to solve too many problems at once, and, in particular, we're trying to solve the problems of ballot creation and precinct reporting at the same time.

The latter problem is much more difficult to solve; electronic precinct reporting implies innumerable security and privacy sub-problems. This makes it tremendously "interesting" to an engineer's mind, and engineers are very prone to losing focus in the presence of an "interesting" problem. Ask any non-engineer and you'll discover that improving precinct reporting isn't actually urgent. It only seems urgent because the engineering mindset got attracted to the "interesting" properties of the solution, and now we're bogged down with a lot of hand-wringing about modems and encryption and verifiability and the reliability of flash storage devices and all the rest of that crap. Adding insult to injury, precisely no one was clamoring for a solution to the precinct reporting problem in the first place.

What's actually urgent is solving the ballot creation problem. There's good reason to believe the 2008 election may well be just as close as the 2000 election (remember Florida) and the 2004 election (remember Ohio). Our very topmost priority ought to be eliminating the possibility that anybody ever again will wonder about the meaning of hanging chad. And when you limit yourself to solving the single most urgent problem, surprise surprise, it turns out to be much more tractable. Here's how simple it could be:

You walk into your local polling place and find a dozen computers. They're not fast, but they all have touch-screens and slots for ballots. You approach the usual tables staffed by Nice Old Ladies, they check you in the way they always have, and they hand you a paper ballot the way they always have. You go to one the computers and feed in the ballot. You waltz though a series of screen images and touch the parts you like. There aren't any "interesting" computer problems to solve, because when you're done voting, you tell the computer and it prints your ballot in an OCR font and spits it back out of the slot in the computer and into your very own hot little hand. This is key. You can read the ballot, it's on paper, and you deposit it yourself in the ballot box just like you always have. The county recorder picks up the box and transports it to the recorder's office when the poll closes just like they always have. They count your ballot with a big computer just like they have for years now, except now the big computer can read your ballot exactly the same way you can because it's printed in an OCR font. They store your ballot just like they always have in case there's need for a recount. End of story.

Ask yourself why we don't hear of simple plans like this one to address the ballot creation problem. Is it because governments don't know any better than to let engineers get carried away with "interesting" problems we don't actually have? Is it because the firms which build these systems see a way to make more money by solving problems nobody cares about? Is it because politicians award contracts to their friends instead of smart people? Being more certain about the results of close elections is just not a very difficult problem to solve. We're making it more difficult for no good reason.