The computer programs that tell electronic voting machines how to record and tally votes are allowed to be held as "trade secrets."
Cryptography guru Bruce Schneier has long maintained that the worst form of security in cryptography is in hiding the algorithm. It gets no review, it gets no test, and as all too many real-life situations show (including the cracking of the Nazi Enigma machine, which though being quite powerful had a subtle weakness amplified by certain usage patterns), it's the things the developer doesn't foresee that become the weakness. (The most superior key in the world is still worthless if you continue "hiding" it under the front door mat.)
In my position as a humble but vaguely technically aware techie, this voting system cannot be considered secure or safe unless *everything* except the keys (passwords) are made completely public and subjected to intensive review in an open forum.
Any piece of code that remains hidden -- that could be the back door that isn't supposed to exist. But how would you know unless you could examine it, and rebuild the identical executable software yourself from the source?
Slight disagreement: a voting system should only be considered secure and safe if its integrity would not be compromised by publicizing everything INCLUDING all encryption keys [actually, I don't see much use for encryption, though I do see some use for secure hashing].
In a normal encryption scenario, the guy who creates the keys is presumed trustworthy. In a voting system, by constrast, nobody is deemed trustworthy.