To: WWWWWWWWWW who wrote (3957 ) 12/28/2004 5:18:38 PM From: cirrus Read Replies (1) | Respond to of 362770 I understand what you say, but once a vote is cast there should be zero "margin of error" in counting. For example, every machine ballot should have a slot for "None of the Above" that immediately notifies the voter that a response is required before the process is complete. That would allow a positive reconciliation of the number of voters against the number of ballots cast. (Sheesh. One can't even order dirty pictures over the net unless all required information is there...). In the case of punch cards or paper, the technology exists to scan it immediately for errors or omissions and reject it until properly completed. Once completed, a random serial number is assigned. When ballots are scanned again at the conclusion of voting, the numbers should match the initial in process count. Any variances caused by, a light pencil mark, for example, is identified by serial number which could then be inspected and resolved by the judge of election and representatives of both parties. After that, there should be zero variances no matter how many times the counts are done. There are other examples, but it's not rocket science. One can go back and argue that a particular "John Doe" should not have been allowed to vote - and change the count that way - but that's another story. It's not a problem with the count.From the math side, any counting method will have a margin of error no matter how microscopic, so whenever there's an uber-close election (like that snafu in the state of Washington, what a mess that is) it all comes down to scrutinizing the votes of people who can't navigate a pencil, poke a chad, etc.