Reply #36: more clarification

[Neil B Forzod]

Unless the source code is digitally signed, you can't tell if it's running on a given machine on election day(s) anyway.

Actually you need the compiled stuff digitally signed to do that. Signing the source code doesn't help. Ideally you'd have the ITAs review the source code, hash and/or sign it, build the target executables and libraries and what-have-you, and then hash and/or sign those. You could check the signature of the installed applications against the signatures/hashes published by the ITAs, so you'd at least know that it's the version the ITA built from the source code they actually reviewed. (Whether you think the ITAs are otherwise useful as reviewers is a separate discussion.)

I don't think it would be that hard to do using public/private key encryption. A web browser could authenticate the stuff, but of course you'd need an Internet connection to the certificate authority and that might be a no-no. Authenticate the code via dial-up maybe? But you'd have to train poll workers to do it.

The voting machines don't have a web browser or an internet connection. (Although you don't need an internet connection to verify a certificate's authenticity, you just need the certificate authority's own certificate.) I think many people would object to any kind of dial-up model -- I'm not sure I want poll workers to call up some remote computer using the modem in the voting machine, it sounds like a pretty serious security risk.

At the moment it's kind of a free for all. I don't think anyone has reviewed all the source code. RABA said there were 285,000 lines of Diebold and only a fraction was carefully studied.

That may in fact be true (I don't have any of the numbers at my disposal right now). But -- and it's a pretty big "but", really -- the observation that "they haven't reviewed all the source code" is a much different statement from "nobody's had the opportunity to review all the source code" or "Diebold refuses to let anyone look at it". Clearly they do and have let people look at it, and there's no credible reason to think they'd object to releasing their source code to an NC-designated escrow agency... something they pretty much do all the time in other states.

That said, there are enough ways to rig this junk without having access to the source code anyway and this is the real threat.

Totally agreed on the alternate rigging opportunities. That's why I personally don't understand the general obsession people have over source code.

Neil