So what is the point of a post-election audit? Well, one thing an audit is supposed to do is spot check the machine tallies by doing a hand count of a proportion of the ballots, and comparing with the parallel machine tallies. But there are several problems with this idea. As demonstrated before, for the paper trail on a DRE to be considered a "ballot," every voter must check it when they vote to see if it reflects exactly what they intended to vote. This would take massive voter education and reminders by the poll workers and posters prominently placed in the voting room, that say, "Be sure to check your paper trail; this IS your ballot. Check for both omissions and additions." It has been demonstrated that people will find additions ("I didn't vote for that guy!"), but not omissions ("How can I possibly remember if all the races were included on that tape as it scrolls by?") We have already seen this omission of candidates trick on even paper ballots the voter fills out themselves; why not just leave off the whole race on the paper trail?
Don't think this bait and switch tactic hasn't been tried. The numerous reports from many states we get in every election of "vote hopping" on the screens are most likely election rigging gone wrong. Internal switching of votes between candidates is far more serious because there is no outward indication that anything is wrong. I talked to a young woman voter who voted in 2004 on a Diebold touch screen machine, who reported that she did check her paper trail and discovered a candidate she was positive she did not vote for. The summary screen was correct, however. She was allowed to have her vote cancelled and to vote again. But if she had been one of the majority who does not check their paper trail, would the audit have picked up this mistake and others like it?
That depends upon whether the memory card inside the voting machine reflected the votes on the paper trail. Was the internal recording the same as what the tape printed? If it was and only the screen was different, then the audit would have found nothing.
And don't forget that the audit is of a percentage of all the ballots cast. In Ohio's audit the first spot check is of 5%. If that one doesn't pass muster, another spot check is done of 3%. If that one doesn't pass then, all the paper trails and paper ballots must be recounted in the county. The idea is that if the 5%, or 3%, is chosen randomly from all the precincts, any rigging would be revealed. Or probably would be revealed. If I were an elections board, or a citizen, paying for an audit with an already stretched budget, I'd want to know exactly how probable is it that rigging, or even unintentional machine miscounting, would be caught? That is, if we randomly select 5% of the precincts (and they actually were instructed to do this by drawing all the precincts written on pieces of paper, from a box), what is the probability that fraud or miscounting will be caught? Only one race was counted, usually President, unless another race had to be recounted anyway, so does this say anything about the probability of malfeasance in the other races? I am not a statistician, so I'd want a university statistical expert, or two or three, deciding the odds for all of us to see.
Another problem with counting these cash-register-type tapes is that sometimes they jam, when the voters voted, and a whole bunch of ballots get printed on top of each other, and so are lost. This happened in the small county I observed the audit in. About 60 ballots were lost, they estimated. Apparently 60 voters in a row were not checking their paper trails. But no worries, they just ran the internal memory of the machine for that particular tape, something called "ballot imaging" (which of course is not of real ballots filled out by voters, but is the digital recording of whatever the voter touched on the screen, assuming it recorded that correctly. So for this faulty tape, it was proven that the memory card and the internal memory of the voting machine were consistent. A triumph for computerized voting machines. Never mind checking the intention of the finger of the voter.
The 5% randomly chosen tapes in the county I was observing were intact except for the one tape I mentioned. But as I observed the audit of the tapes from the ES&S machines in a large county, Franklin, about six rolls of tapes from the 5% and 3% chosen machines were deemed unreadable; this I estimated to be 5% or less of the total tapes counted. However, in Cuyahoga County in the 2006 election audit, a full 17% of the tapes from the touch screen machines were torn, had been jammed, or were otherwise made to be unreadable. SO if you were going to rig the election in Ohio, it would be best to rig the machines in Cuyahoga, or in crowded precincts where the precinct judges are very busy, or on the oldest most decrepit machines where the paper rolls are the likeliest to jam. Or where there are likely to be the most inexperienced voters who would never think to check their paper trail tape.
I have addressed only the spot check hand count in this paper. A real audit should also include spot checking the signature books to see if signatures match and if the numbers add up to the number of votes cast at that precinct. As an observer I was not encouraged to oversee that part. THis problem is correctable. Absentee and early vote numbers should also be reconciled to the numbers of voters reported. SOS Brunner's directive does include these other aspects, but citizen oversight is crucial, as the election workers perform these tasks. More citizen observers must be recruited if an audit is going to have any real purpose. I suggested to one board member in one county that we observers even be paid, but she said they were already scrambling to even be able to afford the audit.
It is no wonder that the election workers will do about anything to avoid recounting all the ballots, or paper rolls. Plus there is a distinct attitude that, "of course the machines count correctly; if there is a discrepancy, we must find the error in the paper trail." Attitudes toward the audit varied between the two counties I observed. In the smaller county, the Director said more than once that she felt an audit was necessary for transparency in elections, and to check the accuracy. In Franklin County, on the other hand, in the warehouse where they counted the paper rolls from the ES&S machines, the leaders gave the impression that an audit was a waste of time--some of the counters said so also. There was no comprehension of how machines might count incorrectly or ever be rigged.
As for that jumble of paper roll on the floor in the photo, some Franklin County clever handyperson has devised a hand-turned machine with a handle to wind the roll from one spool to another as they were counting. To rewind the tapes after counting, if a second count was needed, a similar machine with a drill attached made short work of that. The smaller county wished for such a device if this auditing business, or recounting, of touch screen machines continues. The voting machine vendors are probably rushing to patent such a device to charge more money to Ohio Counties even if I write this.
Franklin County's Ingenious "Spool Winder" for Counting Votes on the Paper Rolls from Touchscreens
For both counties I observed, I had no reason whatsoever to doubt the integrity of the leaders and people counting. I believe they were doing their best to follow the directions from the SOS office on how to conduct this post-election audit. In one instance during the counting of one tape, the counters kept coming up with different numbers from the official machine count; suddenly someone realized that the voter could go to another precinct's machines in the same polling place, if their precinct's machine had a line. Therefore, each machine's roll would probably have ballots from the other precinct. Fortunately, there were only two precincts in this polling place and the machines had recorded the letter of the precinct, A or B, on each ballot. Separating out the ballots on the same tape by precinct letter made the counter's work more tedious.
I think it the benefits of recounting optical scan ballots versus touch screen ballots are obvious. The primary benefit of the OS ballot is that the ballot is exactly what the voter filled out; there is no "middle man" or "middle software" to change that ballot's choices electronically. There is no need for the voter to check their paper trail with the OS ballot, since they just filled it out, physically. There is no paper to jam; if the OS ballot does not feed correctly, it is obvious and can be set aside for human eye scrutiny. There are real OS ballots to recount, and not missing ballots because the DRE machine did not record correctly. If OS ballots are placed in the wrong precincts, they can be separated out again; you cannot cut up the DRE paper roll to do this. The paper ballots are sturdier and bigger, not likely to be ripped or torn apart.
In short, here are my recommendations for Ohio's audit.
Or, let citizen auditors actually be the ones counting the ballots; swear them in like poll workers and pay them what the election workers, or poll workers, would be making. These recruited citizens are not as likely to expect the results to match exactly, and have no stake in "making the office look good."
2. Phase out DRE's in favor of OS's, for all of the reasons stated above. If counties can't afford to make the change, give them a few years and wait till the DRE's start to fall apart, or the next "upgrade" is just too expensive.