E-voting in Belgium
A post by Gunnar reminded me that I wanted to blog about this: e-voting in Belgium is going to change slightly.
Previously, e-voting worked with a computer with a lightpen on which you would enter your vote, a magnetic card that would contain your vote in encoded form, and a magnetic card reader that would count the votes. After more than a decade—when I was first eligible to vote back in 1996, it had already been implemented in the city of Antwerp, where my vote would be cast, so I've never voted on paper—it had still not been rolled out everywhere, and some of the computers used for the voting were now starting to show their age.
A few days ago on the news, they reported that the e-voting system was to change so that rather than magnetic cards, it would use pieces of paper. This way, voters would be able to check that the information on the piece of paper conforms with their vote.
Investigating what this means, I found this website by the federal government, where they have some pictures of a test run of the new system which they ran in October last year. Rather than using magnetic cards, the new system uses paper slips with the voted information in two forms: once as a (huge) QR code, and once as a human-readable text.
In theory, this is a more transparent system than using a magnetic card, and that's a good thing; additionally, it allows for a traditional recount (by people, not computers) should the need ever arise. However, I do still see some problems:
- The website says you need to first fold the slip, and then unfold it when you get at the scanner so you can scan your barcode. This risks violating the secrecy of the vote, as people will be able to glance what other people have voted as they unfold the slip. I'm not sure how this can be fixed.
- If the vote is printed on the paper slip in two forms, you need to be able to guarantee in some way that the two forms are the exact same information. People with smartphones can scan the QR code and compare it to the human-readable code, but not everyone has one of those.
Having said that, I do think the new system is an improvement over the old (in fact, it's exactly the kind of system I had thought about myself earlier when thinking about ways to improve e-voting); and if the two problems are solved in some way (it's not impossible to do that, just requires some clever thinking), I believe it has potential as a way to implement electronic voting in a good way.
I have stumbled across lots of pseudo-articles written by (and for) the SmartMatic company regarding their e-voting implementation "solutions", and they often quote Belgium as one of the early success stories and flagships for their product. This is indeed good news!
Still, it begs a couple of questions, such as... What kind of information is encoded in the QR code? The nature of your vote can be represented as a smallint (even as an 8 bit character), and a QR code rarely makes sense for less than a ~10 character string. What other information are they printing?
So, yeah, I'd been wondering about that too. The website that I linked to in my blogpost shows one ticket with enough resolution that a QR decoder should be able to make sense of it, but I can't seem to find a QR decoder which takes a JPEG rather than a camera lens as input.
It's definitely more than an 8 or 16 bit integer, however. That QR code is huge.