Thursday, July 29, 2004

Putting E-Voting to Rest

From the Desk of David Pogue


To the immense credit of this column's readership, not a single person has written to complain that I've spent too much time and space on the topic of computerized voting machines. Still, with today's installment, I hope to wrap up my coverage of this topic, at least for now. It's becoming clear that this issue is just as contentious and polarizing as any other this election season, and few people's minds will be changed by continued discussion. Meanwhile, the more it becomes clear that there's no way to solve the problems in time for this fall's Presidential election, I'm getting a little depressed by the whole thing.

Hundreds of you wrote to share your thoughts. For example, I heard from several readers overseas, who claimed to be disgusted by the entire voting-machine flap. "In a country where fewer than 50 percent of the citizens even bother to cast a vote, you've got a much bigger problem with your democracy than haggling over which kind of voting machine to use," wrote one.

I also heard from readers who derided the belief that, in the face of concerns that voting-machine software might be flawed or rigged, a voter-verified paper trail is the solution. I've written that such a paper record is critical if a recount is needed; otherwise, the only "recount" you can perform is to check the memory card yet again, which is pointless if you questioned its total the first time.

But the paper trail, several of you wrote, is no panacea. Yes, it provides a secure method of performing a recount -- but that's valuable only if it occurs to someone to PERFORM the recount. Trouble is, how will we know whether a recount is necessary? If one candidate wins the election by a 20 percent margin, will anybody realistically demand a recount? Does that mean we'll have to hand-count the results of every
voting machine in America?

The paper trail is certainly better than nothing -- if one paper-trail recount shows evidence of software tampering, then at least a wider investigation can begin -- but it's worthless unless somebody does, in fact, conduct a recount.

I heard from many, many people who felt that the rush to computerized voting machines is an ill-planned overreaction to the Florida hanging-chad episode. They wrote that non-computerized systems, particularly optical-scan and lever machines, have worked well for years--and they're a lot more secure than the new computer programs.

But many others wrote it to say that in the hands of a ruthlessly determined conspiracy, ANY system is, in theory, hackable. "Regarding the lever machines," wrote a voting official from California, "it's not in the final counting, but in the prep. My grandfather was a politician in Brooklyn in the 30's and 40's. The way machines got 'set up' then was to put a toothpick into an opponent's counter on the back of
the machine during set up, breaking off the excess and letting the counters carry the piece into the workings. This will slow down the roll of the numbers. You don't do this in the units line [of the counter], which could be too obvious, but in the tens or hundreds line, where it would be less noticeable. Ingenious, huh?"

Even paper ballots aren't tamper-proof. "Each side used to keep men called 'short pencil' guys at each polling place," that reader went on. "They would keep graphite under their fingernails so that they could run them across opponents' ballots to make them ineligible for counting (since no marks outside the boxes were allowed)."

I heard from a state voting official in Virginia who agreed that, in the end, no system is totally tamper-proof -- and offered some concrete ways for you to get involved. "In the final analysis, you have to trust the process, and to do this you have to know the details of the process. So volunteer to be an election official, get to know about YOUR system, and make sure that it and your election process works."

Lots of you wrote to suggest superior voting-machine technologies. Dan Wallach, for example, is a co-author of the Johns Hopkins study that ripped apart the security practices of Diebold, the largest maker of e-voting machines.

"I think you've missed one of the best technologies we've got for voting," he wrote. "It's called precinct-based optical scanning. The voter fills in a bubble with a pen on a printed paper ballot, then inserts it into the scanner above the ballot box. This sort of system satisfies what computer scientists have been grumbling about (it's voter-verifiable -- voters see the ballot when they mark it up) and it has a low error rate. And, oh by the way, op-scan is much, much cheaper."

There are even devices that combine the best of both worlds, he went on: touch-screen machines that offer the simplicity and accessibility of Diebold-type systems multi-lingual, adjustable type size, software that prevents voting for more than one candidate, and so on) -- but that print out an optical-scan ballot rather than allowing the software to tally the votes.

These are all terrific suggestions. These systems would do a lot to quell voters' fears and restore confidence in our election process.

The problem, of course, is that many states have already spent millions of dollars on self-contained touchscreen machines with no paper trail and no "op-scan" ballots. The states are not about to throw out all that equipment.

When I made this point to one of my correspondents, he wrote back: "A few million dollars? So what? We're spending $5 billion a MONTH trying to build a democracy in Iraq. Why not spend a tiny fraction of that to ensure a working democracy at home?"


Forum: David Pogue's Columns

Share your thoughts on the feedback board.