Open source vote counting


Here’s a great project that I just learned about in Wired magazine–an open source vote counting system written in Python and released under a GNU GPL v3 license. The system was used as part of the Humboldt County (CA) Election Transparency Project, where it discovered 197 missing ballots and a bug (i.e., another bug) in CA’s proprietary voting system marketed by Premier (formerly Diebold) Election Systems.

The software reads optical scan ballot images and counts their votes. The images are produced by scanning the ballots using a commercial off-the-shelf (COTS) scanner–i.e., a machine not connected in any way with the vote counting machines used in the election. Because the software is FOSS, anyone can download it and run it, thus producing one or more independent audits. These audits would be done by everyday citizens in addition to the mandatory random audits that are conducted by some states, including CT. All we have to do now is convince voting officials to scan the ballots and distribute them to the public.

The software was written by Mitch Trachtenberg. Carolyn Crnich, the enlightened registrar of voters in Humboldt, said yes to transparency and started the Election Transparency Project. She bought a Fujitsu Scanner and recruited volunteers to scan all of the county’s votes, which were made available to the public. Trachtenberg ran them through his software. The 197 missing ballots were not discovered through the normal vote counting, nor through the county’s 1% mandatory audit. The voting machine vendor, Premier Election Systems (formerly Diebold), later admitted that the missing ballots were due to a programming error in its GEMS software. They had known about the bug since 2004. They claimed to have provided a workaround to Humboldt, but a former employee of Humboldt forgot to tell the ROV about it. Hence the workaround was not employed. I’m guessing that Humboldt was not the only county that failed to work around this bug, which affected mailed in ballots. So there were probably dropped ballots in other counties in CA and wherever else GEMS is used to count ballots.

All of which underscores Trachtenberg’s point that proprietary voting systems and “secret counting” methods aren’t in the best interest of democracy.

What about here in CT? Through the dedicated efforts of citizen advocacy groups such as TrueVoteCT and CTVotersCount and others, CT has one of the most progressive mandatory auditing systems on the books? As pointed out by Luther Weeks of CTVotersCount, the faulty Diebold GEMS system is used by a private company in Massachusetts to prepare CT’s ballots but is not used to tabulate the ballots, which are counted by hand or by calculator or by spreadsheet from the tallies provided by precinct tabulators. Are there errors in these manual counts? Possibly. Would they detected by CT’s mandatory audit requirement? Not likely.

It would be much more likely that this kind of error would be detected by this kind of transparency project. So let’s get it implemented in CT!

  1. No comments yet.
(will not be published)