Based on my experiences with computing punch cards (they were on their way out when I was on my way in) I'd estimate the card readers should be better than 99.999% of detecting a hole (ie a vote) if there is one, and accurately keeping count of the number of cards that were in a precinct and comparing the numbers of cards counted would allow the one real problem, feeding through the machine, to be caught and the batch redone. "Swinging door" chads would only be a factor if there was no vote on the ballot for the race being recounted, and I've no problem with the fairness of those being manually inspected, even given that voters are instructed to check their ballots themselves before turning them in. That is not what is being done, and there do not seem to be any independant checks.
There seems to be at least one charge of a coverup today; a Republican observer made a public charge that cards that had been identified as Bush votes (the claim was about 100 were involved) were put in the Gore pile for counting and the count supervisor (a Democrat) tried to cover up the error. The manual process in that place was identify who the vote was for, and to place it on the correct pile. It might be easy to accidentally mess that up over 10,000 times even if you weren't trying to make one pile bigger. I can't find it in print just yet.
Here is the LA Times description of the even handedness of process the Gore team are championing: "Our goal," said one lawyer as he patiently lectured his new charges, "is to preserve the Al Gore vote." The volunteers nodded. "It's very, very important that if you see any kind of mark--a scratch, a dent, a pinprick in Al Gore's column--that you challenge." When someone then asked what they should do if they found a Bush ballot with an indent, the lawyer said: "Keep your lips sealed." It was hardball that the Bostonians and other imported pros understood well." http://www.latimes.com/news/politics/decision2000/lat_dems001117.htm
Finally, on the Fox News, Paula Zahn seemed to be confirming that a candidate for a countywide judgeship in a recent south Florida election was denied a manual recount because she lost by 11 votes and the cutoff for a manual count was 10. They used to believe the machines were very accurate indeed.
The Florida Supreme Court will probably be an interesting place this week.
-----Original Message----- From: Howard L Hansen <firstname.lastname@example.org> To: email@example.com <firstname.lastname@example.org> Date: Friday, November 17, 2000 8:42 PM
>All, > >I find it interesting that the discussion of mathematics involved in the >current election should focus on alternatives to the pluralistic method >of determining elections when the real issue is statistical. Since the >voting machines count ballots with a margin of error (does anyone claim >that they are 100% accurate with or without chad considerations?) it >seems perfectly clear that if the difference between the candidates is >within that margin, the election must be declared a tie. Most of the >pundits make pronouncements as if the voting machines are absolutely >accurate which is completely ludicrous. Of course the same statisitical >analysis holds when hand counting is considered. Therefore, since the >margin of error is greater than the apparent difference between the >candidates the election in Florida should be declared a tie. Of course >state legislators in Florida and elsewhere are virtually mathematically >illiterate and so we will continue to have an unresolvable dispute. > >H^2 >________________________________________________________________ >GET INTERNET ACCESS FROM JUNO! >Juno offers FREE or PREMIUM Internet access for less! >Join Juno today! For your FREE software, visit: >http://dl.www.juno.com/get/tagj. >