Schneier on Voting Machine Security

Ed Cone interviews the security guru Bruce Schneier about voting machines:

There are a couple of reasons that things like automatic teller machines and gas pumps are more secure. The first one is, there's money involved. If someone hacks an ATM, the bank loses money. The bank has a financial interest in making those ATMs secure. If someone hacks a voting machine, nobody loses money. In fact, half the country is happy with the result. So it's much harder to get the economic incentives aligned.

The other issue about voting machines is that ballots are secret. A lot of the security in computerized financial systems is based on audits, based on being able to unravel a transaction. If you go to an ATM and you push a bunch of buttons and you get out ten times the cash you were supposed to, that's a mistake, but that mistake will be caught in audit, and likely, you will be figured out as the person who got the money by accident, and it will be taken out of your account. Because ballots are secret, a lot of the auditing tools that we in the community have developed for financial systems don't apply.

More like this

Hope you don't get the problems with the electronic voting machines that we did in Scotland in the 2007 elections; there was a huge number of spoilt papers, the machines broke down and eventually many counties abandoned the electronic count and had to go back to the traditional hand count.

The Scottish machines were made by DRS.

Who provides the American ones?

A lot of them, including those in my home state of Georgia, are made by Diebold, a big republican campaign donor.

Designing a reliable and trustworth voting system is not difficult. A system which does not have the a tangible, fairly immutable, human readable, physical token as the official 'vote' has all sorts of difficult problems to solve... fortunately a paper ballot isn't exactly exotic technology.

Anyway, the US voting system is a mess. Talking about voting machines is a bit of a red herring actually, since even a perfect machine would be insufficient with a crappy system and a well designed system does not require trusting the voting machine (or any other single part of the process). Voter registration is a larger problem anyway.

Unfortunately for us in the US, one of our major political parties gets a benefit from disenfranchising voters and undermining the electoral system. We have known how to design and implement a much better system for a good long time, and after 2000 we had no shortage of actual experts spell out those plans (and the problems with the current system) in excruciating detail. Sadly, it is a political problem actually implementing such fixes.

BTW: The Caltech-MIT Voting Technology Project reports are a good source. Anyone who knows any statistics can pretty confidently say that more people voted for Gore in 2000 in FL to a pretty good p level based just on the different reported error rates of the measuring equipment (voting machines) in different counties. Again, sadly, one party gets a gain from having crappy measurements (we can't even use statistical corrections on the census here)... reality has a 'liberal bias' apparently.