Bear with me here, this is a phone conversation that I will try to document to the best of my ability.
Last night I called Brett Rapp, President of Triad GSI. That is the company which tabulates punch card votes for a number of counties in Ohio. It is also the company which David Cobb references in this now well known quote:
"A representative from Triad Systems came into this county's Board of Election's office unannounced, that is on this Friday. He said he was just stopping by to see if they had any questions about the upcoming recount.
"He then headed into the back room where Triad supplies tabulators, that is the machine that counts the ballots, is kept. This Triad representative told them that there was problem with the system, that the system had a bad battery and it had `lost all its data.'
I left a voice mail, and incredibly, he called me back at 6:15 eastern time this evening. I spoke with him for about 25 minutes.
Much more below the fold
I expressed concern that there is a perception that one of his employees tampered with a machine and also gave instructions to the poll workers which could be construed as tampering. Here is the gist of his response. There is a lot of process in this, so look carefully at what this process entails and might mean:
The Sec of State has specific guidelines on the way in which a recount is to be conducted. Follows is how the process relates to the punch card tabulators.
- The tabulator machines must be adjusted to report votes counted ONLY for president during the recount.
- Triad employees must go to each county where they are responsible for counting, and enter commands into the machine to suppress reporting for all but president.
- These commands are entered into a DOS based system. So for you older guys out there, you know it would be a series of commands at a c:\ prompt.
- When the reporting is suppressed, the employee then runs a test deck of cards through the machine to ensure that indeed, the reporting is suppressed, and indeed the correct number of votes have been counted.
- The test is repeated in a by-precinct manner, and a report of all votes, by precinct, is produced. (Evidentially, for the Triad Counties, there is only one tabulator per county, so all cards are counted by one machine.
- When the actual recount takes place, 3% of the cards are hand counted. Totals are noted. The same cards are run through the machine to ensure that the machine has matched the hand count. This will lead all to believe that the machine is indeed working.
- If the machine tally matches the hand count, all cards are then run through the machine for totals by precinct. If the 3% test doesn't match, or the total group doesn't match the Nov 2 totals, a hand count is needed.
Now, if observers verify that the hand count matched the tabulator count, then it should be safe to assume that a count of all ballots should be accurate. The observers need to be incredibly diligent to ensure that the hand count is not "rigged" in any way, but I don't believe that a count can be pre-loaded into this card counter.
The big question I have is this: Are we absolutely sure that regardless of the outcome of the 3% test, ALL cards are re-run through the machine counter. This is where the potential for fraud exists. Because if the central tabulator totals were hacked, then without a full recount to match up, we wouldn't know.
Now on to the actions of his employee:
First of all, he claims that the county David Cobb referenced was NOT one of his counties. So two possibilities:
- David Cobb was wrong about the county, and indeed, the incident he references was for a triad employee with punch card ballots, or
- The county he references is correct, and the company doing the counting differs. So this could be not a Triad punch card issue, but a Diebold, or ESS, or whatever.
He did relate that for one county the process related above was performed on Friday, and indeed, when the employee got there, the battery had gone dead. As a result, certain info was lost. That info related to the TYPE of hard drive in the system (Help me here, I think he said something about CMOS). The employee had to update that info, and replace the battery. The Same hard drive is in the system.
It seems that the resident software for this type of machine may or may not hold totals. The commands entered via DOS may or may not change the totaling process. I just don't see how malfeasance can happen.
Hang with me:
If the original tabulations "skimmed" every 10th vote, or something of the sort, the 3% hand recount would catch that.
If in the process of suppressing reports for all but the presidential election, the skimming program were changed, then the full recount wouldn't match the Nov 2 tabulation.
So again, it seems it would be hard to do anything funny here.
I get back to the original:
If the 3% hand count is only used to verify that machines worked, and no full recount is ever really performed to match to Nov 2 tabulators, then THAT is where the Triad employee or any other voting machine employee could take advantage of the process to reset counting methodologies to ensure accurate counts.
Also, if full recounts are not somehow matched to central tabulator counts, the field test could work fine, but the totals could still be off.
In that last instance, hacking of central tabulators could have happened, or skimming at the field level on Nov 2 would not be caught.
It all gets down to this 3% test, and what the next steps are. Do they do a full machine recount or not???