mentioned this, but it's worth repeating far and wide. Diebold electronic voting machines may be (are) very insecure
. In short, an insider or intruder can very easily modify vote totals while leaving little, if any, trace.
Conceptually, I'm a big fan of electronic voting. Florida proved the amazing error rate of the scads of paper-based voting systems deployed across the country. Electronic voting is potentially much simpler, safer, and more accurate. It can allow for safe same-day registration, among other things. However
, improperly done, it can facilitate significant fraud. Electronic voting systems should be extensively evaluated by a wide spectrum of security experts of varied political stripe (partisan and ideally nonpartisan). The ideal system would be open source so that citizens can insure fair election practices. Closing a system's source for fear that someone might crack it is the wrong idea -- if someone can crack it, it shouldn't be used. (Paper-based voting systems are all open-source.) Finally, multiple vote-counting systems should be in effect. Electronic systems could produce a paper ballot which is verified by the voter and then placed in a traditional ballot box. Then, both electronic and paper votes are counted. Significant (and certainly meaningful) discrepancies should be investigated.
Voting is one of the few available checks on corruption in a democracy. It should be as robust as possible.