I've been putting some thought into this lately. I've come up with a few features that could improve things a great deal.

Since I haven't done any research, it's likely that a few of these will turn out not to be original. Except for this feature from India, which I favor as well:

- India uses a system that does not connect the voting machines to any server. Votes are recorded on a memory chip, and removed from the machine at the end of the day. Chips are collected, and their information tallied at a central location. This removes the possibility of hacking into the voting machines.

To this change, I would add my thoughts below. The general idea is to add subsequent verification of the voting record to the process. Let me know what you think. APOLOGIES IN ADVANCE FOR TECHINESS!

- Give the voters the ability to see how their individual vote was counted. Put the raw data into databases on the net and permit voters to view their own votes. Requires the user to 1) have a record of some kind of id# generated by the system (so their vote can be identified without having to mark the vote with the identity of the voter), and 2) enter (and later remember) a password at voting time. To look up their vote, the voter must log on to a website, present the ID# of the vote, and present the password. Then they are shown the official record of their vote.

- The feature above begs the question: so a voter calls up and says he looked up his vote, and it is not accurate - what do we do? The following features address this question:

* Every precinct has some designated volunteers who vote there, and check their votes as soon as they are posted. If they find no discrepancies, then it is one indication of a "personal problem".

* If few complaints or only the one complaint come from this particular precinct, then it is one indication of a "personal problem".

* Very fancy solution: add a way for the voter to verify how the vote was entered on election day (as opposed to how it was counted at the central location - see above). Goes like this: 1) an additional record is made of their vote, encrypted so that the election authorities cannot decrypt this record. 2) user has to dream up and remember another password (would not want to use the same password, because the authorities record the first password). This password-encrypted voting record is absolutely unbreakable because the voting record consists of a list of Ys and Ns (1 and 0 bits). Part of the encryption should be to insert noise bits at locations determined by the password. 3) this callback procedure should probably only be performed while directed by election official? not sure. Any way it's a simple lookup only with a different password (this time the password is used to decrypt the voting record - not to ID the voter). Could require the other password to ID the voter here, as well. Record is decrypted and voting record is displayed. If discrepancy between this voting record and the one on the database, then somebody spoofed the database that the count came from!

* This begs the question: what if the voter screwed up their encryption password? Simple solution: at the voting machine do some kind of hashing on their password and record the result in the record. Then when the voter keys in the password at verification time, verify the entered password against the hash code (similar to the check digit that comes at the end of your credit card number). So if the voting results do not agree with the voters recollection, and the hash disagrees with their entered password, we can tell them "that's not the password you keyed in at the machine, sorry."