July 28, 2004
Where'd I put those backups?

I'm sure you've seen the story of Florida's latest voting woes by now.


A computer crash erased detailed records from Miami-Dade County's first widespread use of touchscreen voting machines, raising again the specter of elections troubles in Florida, where the new technology was supposed to put an end to such problems.

The crashes occurred in May and November of 2003, erasing information from the September 2002 gubernatorial primaries and other elections, elections officials said Tuesday.

The malfunction was made public after the Miami-Dade Election Reform Coalition, a citizen's group, requested all data from the 2002 gubernatorial primary between Democratic candidates Janet Reno and Bill McBride.

In December, officials began backing up the data daily, to help avoid similar data wipeouts in the future, said Seth Kaplan, spokesman for the county's elections supervisor, Constance Kaplan.

The loss of data underscores problems with the touchscreen voting machines, the citizen's group said. "This is a disaster waiting to happen," said Lida Rodriguez-Taseff, chairwoman of the Miami-Dade Election Reform Coalition. "Of course it's worrisome."

The group is concerned about the machines' effectiveness, following revelations about other problems with the system. Last month, state officials said the touchscreen systems used by 11 counties had a bug that would make a manual recount impossible. Earlier this month, a newspaper study indicated touchscreen machines did not perform as well as those that scanned paper ballots.


I've said before, and I'll say again, that I believe the touch-screen voting machines should merely be used as an interface, with a printed ballot that's counted by optical-scan machines used as the actual vote of record. I could be persuaded to support using the machine counts as official, with paper ballots as backup only and for recounts, but either way I feel that having that kind of redundancy built into the system is a Good Thing. What happens if the two don't match? Well, then you've got a problem on your hands, as the whole point is that the two are supposed to match, but I'd say the paper ballot, which each voter gets after pushing the buttons and then drops into a ballot box, is the official record any time there's a discrepancy. I really don't see what's so controversial about this.

Greg has been on a bit of a tear lately about the perceptions of security of electronic voting machines and allegations that they lack in various areas, and while I think he makes some good points, the point that I keep coming back to is that we've all basically been told to trust Diebold, Hart Intercivic, and all the other voting machine makers. What alarms me about this Florida article isn't my tinfoil-hat "oh my god they're going to steal the election!" instincts, it's the thought that it took a catastrophic failure - and its discovery by a citizens' group - for anyone to suggest that regular backups of this data might be a good idea. How can you put any faith in that kind of process? Has anyone asked Beverly Kaufman what kind of backup procedures Harris County has? At least with paper ballots, you can point to a locked storage room somewhere.

We don't really know how robust and stable these machines are, and the reason we don't know is because the manufacturers won't tell us. That would mean parting with their trade secrets. Given a choice between their intellectual property and the integrity of my voting system, I'll choose the latter every day. I'm not an open-source evangelist, but I can't see why that isn't the right model to emulate here.

I certainly have no desire to go back to punch cards. I actually like the eSlate interface, and I feel confident that with such an interface all kinds of problems such as overvotes and undervotes can be eliminated. All I'm really asking for here is transparency. Can we all agree on that?

Posted by Charles Kuffner on July 28, 2004 to Technology, science, and math | TrackBack
Comments

I think one fair debate is the matter of who should be making the voting machines and/or who should be doing the software work. I'm literally on the fence as to whether an open-source code would be advantageous or not, for instance.

Of course, when I read stuff like this, I say to myself that the problem isn't that there's a paper trail lacking, but that simple electronic security things were either overlooked or performed properly (seriously, how tough is it to back up this sort of data?).

A big loud argument over the best online security steps to take, the proper testing for such devices, and so on ... that's all well and good for my money. Adding paper, to me, just introduces one more variable to potentially go wrong in the process.

Posted by: Greg Wythe on July 28, 2004 3:07 PM