There is a widespread belief that 60,000 ballots were audited in Nevada after the 2004 election. In fact, I heard Mr. Alfie Charles of Sequoia make that statement in a demo of the Sequoia VeriVote on July 14, 2005. However, only 1268 ballots were audited in Nevada — ballots in Clark County.

On Friday, July 8, 2005, I spoke with Larry Lomax, Registrar of Voters in Clark County, Nevada. Mr. Lomax said he believes no other Nevada counties did a manual audit, due to a misunderstanding about what was required. I confirmed this with the election offices in both Washoe and White Pine Counties. The audit they did consisted of comparing the results printed on the machine tapes with the results reported by the central tabulator.

Mr. Lomax described to me the audit process used after the 2004 General Election to compare the voter-verified paper records produced by Sequoia VeriVote voting machines with the results reported by the machines. I have written the details of his description below.

Twenty machines were randomly selected using a computer program that generates random selections. Five teams of four people each tallied all the paper records produced by the 20 selected machines. The counting process took a total of two days; it took approximately half a day for a team of four people to complete the records for one machine. The total number of man-hours was 5(teams)*4(on a team)*8(hours)*2(days) = 320. Observers were welcome, but there weren't any. However, when the county previously audited the primary, a television station had filmed the counting. Here's how they audited the records from one machine: The paper roll was removed from the machine and taken to a team of four people. One of the four people read all the selections on each ballot in order as two people, sitting apart from each other, manually tallied the votes on pre-printed tally sheets. The fourth person of the team observed in order to catch any mistakes as they occurred. Each tally sheet contained a list of all candidate in all races. A vertical mark was made for each of the first four votes for each candidate, with a diagonal mark through each set of four to form a group of five. Mr. Lomax said there were between 20 and 25 races on the ballots (he couldn't remember for sure). Previously his office had told me there were 21. The printing was in 10 point font at a normal aspect ratio. While a few people commented that it was small, it didn't present any problems. During voting, a voter can use the magnifying glass which doubles the size of the print. Some of the records were for Spanish ballots. I asked if the names of the contests were in Spanish, and he replied that, of course, the names of the candidates were the same on English and Spanish records. Records were separated from each other by about 6 inches of blank space, so, since there were about 70 records on each machine, the used part of each roll was about 70 feet long (out of a total of 300 feet). I asked if he thought it was possible for the record of a ballot to be longer than the length of the viewing window. He said it was not longer in this election, but that it might potentially be a problem if there were lots of races in an election. However, when I was in Olympia watching a demo of the Sequoia VeriVote, Alfie Charles said that if there are too many races to fit in the viewing window, the paper will scroll so the voter can see the additional races. Perhaps this is a different printer or later version. After the selections on each paper record were read and recorded, the paper roll was unrolled loosely into a box to expose the next paper record. When all records had been tallied, they rolled the paper onto the roll again. Since the paper was about 4" wide and approximately 70 feet long, it was difficult to keep the sides of the roll even as they rolled it back up, so the final roll was not tidy — about 8" wide. When the tallying process for a machine was complete, they compared the two manual tallies with the machine tally for that machine. About half the time, there was a discrepancy — usually just in one race. If a discrepancy occurred among any of the three tallies, the process was repeated for only the race that had the discrepancy. In each case, it was discovered that the machine tally was correct. Mr. Lomax said that the level of difficulty of counting the paper records was comparable to counting optical scan ballots.

Everything should be made as simple as possible,
but not simpler.
~ Albert Einstein