Although I have some doubts about the concept in general, I think that election software is one area where open-source software development is absolutely compelling.
Diebold's software was leaked on the Net a few years back, and the people who took a close look reported a lot of bugs, some of them security hazards, and were pretty critical of the overall design and quality. A lot of the problems would've been spotted in the first day of intense scrutiny under an open-source regime.
Some people think it's nuts to put sensitive software out in public, but it's worth noting that when it comes to encryption software, professionals won't use proprietary encryption algorithms no matter how cool the author claims they are--the algorithms have to have been published and examined for flaws (not to mention backdoors). The gold standard for encryption is that an encryption algorithm has to be so bulletproof that it doesn't matter if hackers have the source software, the encryption still can't be broken.
The same standard needs to be applied to electronic voting machines. The software should be developed as open source, picked apart by everybody with a political axe to grind, proven to be unhackable, and only then should it be put into service. Private companies could still build the hardware, specified as tightly as needed to ensure that the software will run, but local officials would load the software themselves.
Tampering with the hardware's a whole separate problem, and things like open USB ports are just too stupid for words. The thing is, if the software can be controlled, it provides a way to prevent hardware tampering. Good voting machine software would record all accesses to the hardware, perform continual diagnostics, provide access control, etc.; and be able to report problems to a supervisory system. If the software is in firm control of the hardware, then the only way to defeat it is to load new software, and that can be detected after the election by doing a checksum of the hard drives of the voting machines.
Seems to me that Geek-Americans have a duty to help along democracy by taking the design of voting machines into our own hands. We're the people best equipped to get it right, and we sure as heck shouldn't be leaving this to government bureaucrats sucking up to private vendors. Just let the geeks loose to do what comes naturally, and we'll get it solved right quick.
Off Topic » in reply to Voting security
open source
