Why You Shouldn’t Trust Electronic Voting Machines

After the 2000 Florida election debacle, a lot of people started advocating the use of computers to replace existing voting systems. But anyone who knows anything about computers can see why current implementations of such systems are horrible and untrustworthy, and a correct implementation would be very difficult to create.

Here’s an example of why current electronic voting machines aren’t to be trusted.

The Risks Digest (vol 23 issue 2) reports that in Fairfax County, Virginia, it was found that the WinVote machines being used to collect vote were quietly discarding about 1 out of every 100 votes for the Republican candidate. This was noticed only after several voters complained that the check mark besides the candidate’s name disappeared after a few seconds. (This is in addition to the various other problems Fairfax has had in recent elections with their voting machines — see previous issues of the Risks Digest for those.)

The fundamental problem with electronic voting systems is that there’s no transparency. How do I know that the vote I cast is being counted? The problem is, you don’t. If the voting software is closed-source (and almost all of it is), how do you know someone didn’t write code to deliberately alter the vote counts? This could be the cause of this problem in Fairfax — someone could have written code to discard 1 out of every 100 votes for the Republican candidate. If the screen hadn’t been updated to reflect the system ignoring the vote, who would have realized what was going on? With paper ballots you at least know that your vote has been recorded.

That’s not to say that open-source voting software is automatically better. But being open-source allows for independent auditing of the source code to make sure there isn’t code to clandestinely alter the vote tallies. Of course, in practice you’d have to audit all software on the system — from the voting software all the way down to the graphical toolkit, the libraries, and the kernel itself, not to mention the hardware. That’s a lot of work, and we don’t yet have any good way to prove the correctness of non-trivial programs.

However, the current “state of the art” in voting software is deplorable. Ever hear of Diebold? A little Googling should show you what I mean.

Comments are closed.