You provide strictly computer voting equipment and nothing else, with bar code receipts. The equipment sends data to * government voting servers * computers of both parties involved * the press * any other group that feels like counting votes The only way to fraud this system if kiosk voting stations are engineered to send the wrong data. But then you take an exit poll of the voters. In the receipt they gave the text displays who the user voted for (so that an attempt at fraud can be detected immediately), and in the bar code there's some cryptographic data that cannot identify the voter, but can identify the vote as it was later counted. The receipt is then hard evidence that cannot be frauded. You don't need to go through the expense of counting all the receipts, you just take a random sample of the receipts, and you should be able to identify 100% match with what was counted on the servers. Any attempt to fraud can be detected and a real recount of the hard-printed evidence be done. That would be real proof, from real hard evidence. ''That sounds like a smart idea (seriously). What are you doing with it besides posting here?'' * I don't have time or energy to do anything with this idea. Should I send it to the patent office ? I think not. I am puzzled though because either I am missing something essential, or there's too much talk and too little action for a problem so easily tractable. This idea came to me while driving with some friends of mine to work in the friendly carpool lane, and hearing on NPR all the debates about electronicc voting . Micro-printers are ubiquous, robust and cheap (just pay a visit to your grocery store), bar codes are trivial technology. Why in the world some fellow engineers at Diebold or wherever cannot print receipts for votes, while other groups shout and whine that electronic voting is unverifiable, that one is beyond me. ''Counting votes is easy. Counting secret votes is harder. (However, the Indian electronic system is worthy of note, given its simplicity and robustness.)'' * It is not harder, all you need is to provide an unique (and cryptographically secure) identity to the vote. You don't need to tie the vote to its voter. In the "verify the ellection" poll you draw a rather small random number of receipts from all over the country, you feed them through optical scanner again, and you should be able to have a 100% identification with the same votes as were counted nationally on the ellection day, of course, the scanner can confirm that the bar code that encodes the vote uniquely across the system corresponds to the name displayed in the clear print. The clear print is important for the voter to verify, otherwise the machine can display one name to the voter and encode a different name in the receipt (and send that other name to national counting servers). But if such a machine is engineered for this fraud it will be detected when the latter is verified - that the print receipt doesn't match the bar code with the clear name - or when the user sees that the printed name on his/her ballot doesn't is not that of the candidate s/he voted for. In such cases, even erroneous votes can be rectified just-in-time: say an old grandma thought she voted for a particular candidate but she pushed the wrong button on the screen; she'll see the discrepancy in the receipt and will ask an election official to help her correct the problem). ''But what about the husband that demands to see the wife's recept? She is no longer allowed to lie so she can be coerced.'' * If this is a real problem (and I'm not saying if it is or it isn't), the wife can put the receipt in a receipt box, at the polling station. just like she does currently with the voting chads. The solution tries to reconcile two opposing forces: hard copy votes are verifiable, while electronic votes are much more economically and much faster to count, but without a hard print receipt electronic votes are unverifiable. The hard copy votes then can serve as a control or safety net against someone stealing electronical votes (if they do, the chances to get away with it, at a poll of hard-print votes are statistically negligeable, so nobody will try to do it). ---- IANAL, but, one issue with the idea of "receipts" is a legal one. In the event of some sort of manual recount a "receipt" has a different legal status than a "ballot" that was actually handled by the voter. In other words, without actual paper "ballots" a manual recount is potentially not legally valid. ''I don't speak legalese very well, but I can imagine a scheme in which we make the ballot really be the receipt and the electronic votes be just a speedy way to count the ballots. Unless problems are detected in the QA phase (poll receipts and see that they match what was counted) and challenges are brought, the hard copy of the ballot will not be counted. Voters can put the receipt in a box at the station, just like they do now. Or they can have two receipts, one to put in a box and one to take home. Under such a scheme I do not expect manual recounts to be necessary at all, they are just a safety mechanism for worst case scenarios.'' "...I can imagine a scheme in which we make the ballot really be the receipt..." This is approximately what the Open Voting Consortium advocates: http://www.openvotingconsortium.org/modules.php?name=FAQ&myfaq=yes&id_cat=9&categories=The+Ballot, http://www.developerpipeline.com/trends/170702562, http://gnosis.cx/publish/voting/electronic-voting-machine.pdf ---- I was thinking just this last week of a voting proposal similar to this, not knowing this proposal was here. In my thinking the voting machine would be the first repository of the voting record, containing the records of votes, and voters (while maintaining the secrecy of their choices). when the voter enters the selections, and registers them by pressing the "Enter Vote", the vote goes immediately into its memory and is independently transmitted to two official vote tabulation sites, an official independent monitoring site, and a publicly available site. The voter receives a two part record of vote, with plainly observable choices marked with X's with which the voter can compare the choices made, with the choices transmitted. If they are agreeable, the voter retains one copy, deposits the other in a locked box, opened only if necessary in a manual recount, and both are separately retained until the next election available for secondary and other necessitated confirmation of electronic counts. Should the boxed count and the electronic count not agree, the voters receipt would be used to discover any fraud or innaccuracies. 20050328 -- DonaldNoyes ---- See also: WikiWikiGovernment, VotingMachineDiscussion, TechnicalSpecificationForVotingMachines ---- CategoryVoting