This is the last article of Tristan Nitot who challenged me with this idea of open source voting machine. Here is the excerpt of his article where it is question:
Bruce Schneier explains that the transparency provided by the fact that the source code is public is essential in this environment. There are other examples, such as the famous voting machines, the black boxes of democracy ( Ireland has just scrapped its own, for lack of reliability), or the cookies announced in the framework of the Hadopi law , which can lead to the shutdown of Internet access. As Bruce Schneier explains, from the moment the issue is important (elections, court decisions), we must be sure that the software that helps decision-making is transparent, and that we can compile it. – even to be sure that the source code we are analyzing is the one we are executing. This implies in essence that the software in question is free.
Is an open source voting machine the guarantor of the transparency of the vote? Let’s study the question a little bit.
If the code is open, it is easy to make sure that it is doing what it is supposed to do. This is a positive point. Now this code will be installed in a voting machine. It is also necessary that the design of this voting machine is open.
We must go one step further in the definition of the voting machine. It must use free software and its hardware design must be open. By this statement we are already beginning to touch the limits . But however, it exists, there is for example the OpenRISC 1000 a processor with free specifications.
We can therefore obtain a voting machine whose software and hardware specifications are free.
How are we going to guarantee that the voting machines built according to these specifications will comply with them? Who will be responsible for this certification? Then admitting that this point is also raised, it will be necessary to ensure that the machine delivered to the voting point is the one that has been checked and that it has not undergone any alteration.
Then comes the moment to enter the machine to vote the data about the voting point: the list of voters, candidates. It will be necessary to find a solution so that this operation does not allow to introduce falsified data or any other program likely to modify the official program.
All this seems very complicated to replace a simple urn and avoid the chore of stripping. This manual procedure is not flawless either. But his control is accessible to anyone and not just computer scientists.
Richard Stallman had declared himself opposed to the principle of voting machines, whether free or not:
Democracy is too important to take preventable risks. If a digital voice counting system seems sound, it would take twenty years of experience to trust it.
All this to show that in this case , the transparency of open source does not necessarily solve all the problems. Of course it looks better than a proprietary voting machine. But in this case it only changes little in the end. The question is rather: do we really need a voting machine?