Transparency, Open Source, and Paper Ballots

I previously reported that "transparency" was key word for people’s positive response to our our recent DC demonstration of our digital voter registration system (DVRS). There is also a similar transparency issue with voting systems, and voting systems also have another transparency issue around paper ballots; and then there is the issue of open source. Here’s the how the 3-way connection works.

In the case of the DVRS, the transparency benefit is public access to the who/what/when of every change to the voter registration database. This access results from the DVRS logging logging every action and event in system, and publishing the logs on the Internet (with suitable privacy filtering). The same transparency issue applies to election management systems and voting systems, where OSDV’s technology offerings will do the same kind of logging and publishing, but for different types of events. Here’s my favorite example.

An election official is submitting a deck of mail-in ballots to an optical scan device, and one of them is flagged as having no vote in one contest. The official inspects the ballot to "interpret the voter’s intent" and finds a line drawn through a bubble, rather than a properly filled-in bubble. The scanner didn’t see the line because of calibration issues, but the person does see it, and tells the counting system to count a vote for the chosen candidate, even though the scanner didn’t find one. That’s part of their job, but each such action is certainly of public interest because the election official is using their discretion to change vote totals; they should be personally accountable for each such action, and one way to achieve that (and transparency) is to publish the logs of all these actions.

In a similar way, election integrity activists have been advocating the use of paper ballots because they create a durable record of voters’ intent, that can be audited to check up on the results produced by election officials. That’s a form of transparency that requires some work, but here’s an example of reducing the work: publishing the digital images of the paper ballots as scanned, which is being done in Humboldt County CA’s Transparency Project. Not only is this an effort in transparency, the project also enabled discovery of a voting system flaw (see the Brad Blog for an amusingly sensationalistic account) in which votes were not counted.

This is exactly the sort of "people checking up on the machines" that is a key part of the "hybrid" approach to paper ballots and voting devices. The other half is the "machines checking up on the people" which includes the use of scanner devices (rather than hand counts) in the first place, but also the type of logging and publishing I mentioned above. The two sides of the coin are essential in the hybrid scheme, in order to have election practices that don’t rely solely on either computers or people to operate perfectly.

Lastly, here is the connection to open source systems. In this note I have suggested that you might want to have election systems that fosters transparency via public disclosure of election operations. If so, then you need to have some assurance that the systems are reporting correctly and completely. Having the software be open source is one part of the recipe for assurance, in contrast to closed systems that cannot be easily inspected or audited. But another ingredient in that assurance recipe is the paper ballots, which provide a physical check on the correct operations of the machines. And then the machines can help by publishing ballot images along with log data. That may sound circular, but to me it is as beautiful as the two sides of a Gothic arch curving together to support one another, to create one of many elements of stability in a cathedral.

— EJS

PS: At the risk of spoiling a beautiful ending, I should also add that the inclusion paper ballots scanners in OSDV’s tech roadmap is fundamentally a respond to demand, rather than an ideological position. The roadmap also includes "voting machines" (technically "electronic ballot marking devices") for jurisdictions that like citizens to vote with computers rather than pens; and in fact such machines are almost legally required for the enhanced access that’s specified in Federal law (HAVA).