I came across this story from the seminal today. It is the first segment of a multi-part piece interviewing the election reform activist Dan McCrea, and he had some interesting things to say.

He points out the conflicts of interest in Florida in 2000 and Ohio in 2004, talks about how HAVA made things worse, and points out some other interesting things. However, this caught my eye:

While Florida, and to a greater extent Ohio, remain electoral mysteries, election issues in Sarasota, Florida in 2006 seemed to offer election activists the best chance they had yet had of using the legal process to obtain greater transparency in elections.

While I understand that the premise is that seeing the code might reveal something interesting I am not sure how it could ever achieve any level of election transparency. On the other hand I do think that they should have just shown the code. There might have been some unrelated problems in the code, and it would have been a minor problem to fix flaws found in the software compared to the PR disaster of not revealing the code. From what I have seen there is clear evidence that the problem was a ballot design problem, and revealing the code would have put the flawed software idea to rest.

The Herald Tribune did do an analysis for which I am unaware of a good refutation. Michael Shamos also gave a talk at UMBC about the analysis that he and a team performed on the system. While he admitted that there were some flaws none of them could have caused that particular error, and he also indicated that it was a ballot design problem. At WOTE 2007 I had the chance to meet with Ted Selker and he basically said that it was clearly a case of bad ballot design.

What I have not seen, however, are the results of a real-world test of this idea. Ted indicated to me that he was, at least, planning on it, but I’ve not seen or heard about anything since. To me it seems like a highly plausible hypothesis and it would be interesting to see the results of such a test.

If it turns out to be true it would underscore the importance of usability in a voting system — Just because it is on a computer doesn’t mean it is automatically easier to use. There should be some minimal requirements for testing each ballot design before it can be used in an election.

I look forward to reading the next segment of the interview.

Often when we attempt to explain Punchscan to activists and others we hear something like the following:

Darn it. If my 80-year-old grandma can’t understand it, then we shouldn’t use it. Elections should be transparent.

I have been meaning to address this issue in a blog post for some time, but today I see that Ben Adida has already done a pretty good job. While people may have a point that Punchscan and E2E in general is somewhat complex, this is not the same as transparency. While not everyone will take the time to understand it, the fact remains that a normal citizen could do it if they so desired, and that is the key distinction between E2E and many other systems. There are no experts, closed sources, closed designs, or NDAs. Ben’s definition of transparency is as follows:

A system is transparent if, given a reasonable amount of time and effort, a person with a college education can understand it. Then, those without the education, time, or willingness to understand it can consult with someone they trust who does understand it.

I sort-of agree with this statement, although I would use high-school instead of college.  In reality, I think that transparency is fundamentally different, and it has to do with the level of observance people can exert on the election. Anyway, we have been saying something similar to this for a long time now. Here’s what we say on our FAQ about it:

The actual system could, we expect, with such a mock election as introduction, be taught in advanced high school science or college classes. This is many times simpler than convincing anyone of how the software in current voting systems works—even if people were allowed to see it!

I see no reason why anyone with a minimal education would not be able to understand Punchscan with the help of a 1-2 week (or day) course.

The only thing I find even a little objectionable about Ben’s post is the following:

…to anyone familiar with quality control processes, the ballot chain-of-custody is a reliability nightmare: how does one check that no one has tampered with a ballot box full of de-identified ballots that no one can look at during the 24 crucial hours where low-wage, minimally trained election workers are entirely responsible for them?

This doesn’t exactly do justice to the solution most activists are trying to promote. They want semi-translucent boxes in full view, and counting done on the same day at the same place in a specific way in front of anyone who is able to observe. There are still many problems with this but it is a bit different than centralized counting or the DREs we have today.

Kudos to Ben on an excellent post.