Reading the comments on the electronic voting story today was interesting. After reading them, here's what the kind of electronic voting system I would want:
1.) Voter selects a candidate from a menu (including write-in option).
2.) For write-in votes, the voter can enter the name from a keypad or touchscreen.
3.) A paper ballot with a unique ID is printed out for the voter to verify. (The ID or serial number could be printed on the paper in advance, therefore having nothing to do with the individual machine, and no correlation with an individual voter.)
4.) The voter can choose to accept or discard the ballot. If accepted, the ballot is fed back into an optical scanner. The machine uses the serial number to confirm that the ballot being submitted is the same as the one most recently printed, and checks that the selection of candidates is the same. If the serial number or selection does NOT match, the vote would not be counted, and the voter would have the option of starting again.
5.) If the voter rejects the ballot, he or she puts the ballot in a shredder. The vote would not be counted, and the voter would have the option of starting again.
6.) A separate optical scanner, preferably made by a diffent supplier, would scan and count all the ballots in the ballot box. This would provide a quick way to check whether the voting machine is keeping an accurate tally. If the numbers don't match, a hand recount would be required.
I think voting machines have a lot of potential if done right. They can be made more accessible by adding braille keypads and audio output. If there were a well-tested system that worked like this, I would probably trust it more than random people counting by hand...
I wonder what it would really take to build and market an electronic voting system? Also, if everyone is so gung-ho about open source voting machines, why not start an open source project to develop them? Funding would be nice, I guess... Any takers? :P