An investigation in Humboldt County, California has discovered a miscount of up to 197 ballots by its commercially purchased vote-tallying machines from Premier Election Solutions (formerly Diebold). An article from Wired has identified how the miscount was unearthed: a citizen-made open-source auditing system.

Since 2002, Humboldt County has promoted the Humboldt County Election Transparency Project, an initiative set up to ensure adequate rigor in vetting the electronic balloting process. The software—called Ballot Browser—was developed by Mitch Trachtenberg, a local area software engineer. After the ballots (some potentially with chad, just wanted to throw a reference in there somewhere) are scanned by the Premier system, they get scanned a second time. Trachtenberg’s software is then run on these scans to identify where the punches were made.

After the Transparency Project’s audit, it uncovered 197 ballots missing from Premier’s counts. After some troubleshooting with Premier, the cause was revealed to be a bug in their proprietary software, which inadvertently deleted portions of the ballots it had previously counted. As the story from Wired goes, each stack of ballots fed into the scanner is given a “deck” number starting at zero (0, 1, 2, and so on). If during the process an operator interrupts any subsequent deck, the software deletes the deck 0. The scary bit is that the logs would show no evidence of the deleted deck zero, or of its existence entirely.

The error had been revealed earlier in 2004 with a simple workaround: delete the “deck 0” before it has any meaningful data in it, and you’re set. Not a great workaround, but serviceable.

The major concern is this: what would have happened without this auditing system? Odds are that the votes would have been lost into the ether. Not a frightening prospect in wide-margin victories, but what about the tight races? Any thoughts on these Premier/Diebold balloting machines? I’m certain that the software error wasn’t a vast right-wing conspiracy, but does it underline a need to be wary of proprietary software? Or is that alarmist, and this is an isolated incident?

Proprietary software has garnered significant investment by development firms, in part, thanks to its clear-cut business model. Perhaps open source will see further attention as government realizes that the plus of near-ubiquity in proprietary software comes with its fair share of transparency-related caveats. With a few more strong showings by open source in the governance arena, policymakers may have to recognize it as a viable contender.