Electronic voting: Open Source is not a solution
It is election day in the US, so the time is particularly favorable to attracting public attention, again, to electronic voting machines... probably the most funny example of this is Oprah's vote getting lost on a touch screen; knowing Oprah's tremendous reach, maybe a non negligible amount of North American citizens now knows that computers don't do good voting devices. Unfortunately, Oprah's blamed this on herself, in the all too common old mood of blaming oneself for somebody else's bad software design that made companies like Microsoft immensely rich.
But Oprah is not a computer scientist, nor an IT expert, right? Computer scientists or IT experts should get these issues right, one would expect. Well, you will be surprised to know how many computer scientists and IT experts can get these issues wrong too: it's just that voting, political voting I mean, is something happening to so many people, and yet so specific, and happening so infrequently, that people is both believing they know it perfectly, and yet not having enough experience to grasp its complexity and preconditions.
A typical example of this is the post I just stepped onto by IDG's Paul Venezia claiming that using Open Source is the solution to e-voting woes.
I may understand the good will and the reasoning of Paul Venezia, and I am not pointing a finger at him specifically, but I am sorry to have to say that the argument is completely wrong: I am a long time advocate of free software, I have complained about closed source monopolies, and advocated free and open source software for more than a decade, and yet I will only accept to trust an electronic voting machine if my trust in the vote's sincerity will not depend in the least on the software running on it.
Being a computer scientist, a hacker and programmer, I have felt as naked, as powerless, as cheated upon as any other citizen when I have been forced to vote on a Nedap machine in my electoral district in France; and this is not changed in the least by the fact of knowing that somebody, or even everybody, has access to the source of some code from which it is supposed that the binary code running on the voting machine has been produced.
Think a bit about this, IT experts around the world: how can you be sure, when you perform a cash withdrawal, that the software running on a teller machine is the same you have received in source form sometimes before? You cant. And this is not a particularly new observation, it is just old common sense among the people dealing with security in computer systems (if you have some spare time, please read Ken Thompson classic article "Reflections on trusting trust").
Let me say this out loud and clear: Open Source is *not* the answer to the e-voting inherent faults.
I did spend quite some time in explaining these issues in a clear and accessible way first in 2003/2004, and more recently a couple of years ago , but living in France, all this was in french.
Seeing the stakes, I will try and translate some of all this in English soon.