Never mind my previous posts about the voting machine problems at Downtown West. After trying to sort through all the confusing media reports about the exact nature of the problem, I finally watched the WATE video. A picture is worth a thousand words, as they say. Some of the mystery is solved, but new questions are raised.

Read more after the jump...

The "voting machine" in question is not a voting machine at all. It is the "Jury Booth Controller" (JBC) that controls the voting machines as we originally thought.

The first clue was that "the machine’s primary data card is apparently corrupted" as stated in early reports. That same report, however, talked about a "malfunctioning voting machine". The eSlate machines themselves do not have a data "card", but rather on-board non-volatile memory chips where backup votes are stored.

The next clue was the explanation that a "pin that acts as a hinge for the printer paper cover" had fallen into the machine and shorted out the motherboard. The voting machines used in Knox County do not have printers.

WATE confusingly reported that there was an "error with a voting machine" and that the "voting machine" was taken to a local company for repair. But watching the WATE online video, you see technicians working on a controller, not an actual voting machine. Why this was not made clear in the media reports, and why they continuously referred to it as "voting machine" is a mystery. Most likely it is just lazy reporting.

Regardless, it answers many of our previous questions. Namely, why weren't the votes simply counted from the controller's internal memory or the controller's flash memory card, which is where the "vote of record" is stored.

The reason they couldn't count the votes from the controller's internal memory is now obvious. How and why the controller's MBB flash memory card was corrupted is another question, but presumably this was related to the controller being "shorted out" and "smoking".

So here are the new questions.

What happens when a controller fails like this? This would seem like a pretty dramatic event that would have been reported. Local media reported that the machine was "smoking" so they simply "unplugged it." It wouldn't have been quite that simple. No voting can take place on the eSlate voting machines without a controller. A new controller would have had to be installed with a new, freshly initialized and serialized MBB flash memory card inserted.

Hart InterCivic operating manuals do not appear to have a procedure for this, so unless there it is covered in another manual something had to be improvised. How it would affect the eSlate voting machines previously attached to the failed controller or the votes contained in their internal memory is not clear. Maybe this is why they weren't able to simply go to the individual machines and count the votes there.

The Hart operating manuals describe a procedure for obtaining a "recount" from individual eSlate voting machines using the Hart SERVO software. This data is obtained from a backup downloaded from each machine as it is checked in to the warehouse after an election.

The Hart operating manuals describe a procedure for rebuilding a corrupted MBB flash memory card, which is the "vote of record" recorded on the JBC controller. In this procedure, the MBB flash memory card is rebuilt from the controller's internal memory. This was not possible in this case for obvious reasons until the controller's internal memory chips were transferred to another motherboard and installed in a new controller.

This seems to be a flaw in the eSlate system architecture and design. If two of the redundant vote data backups are on one machine, in the case of a complete system failure it is not possible to recover them. The fact that the machine and its two redundant vote data backups can be rendered unusable by a ten cent part falling into the motherboard would also seem to be a design flaw.

Even with such a system design there is a third copy of the data on the individual eSlate voting machines. The question then is, why wasn't the Hart SERVO software eSlate recount function used to retrieve the votes? This would seem much simpler and more reliable than flying in two technicians from the vendor and doing major surgery on the controller.

And finally, why the media can't just explain what is going on in a straightforward manner is also puzzling.

UPDATE: As I was posting this, Knox County Administrator of Elections Greg Mackay answers on another thread the question of why the dSlate counts were not used. They weren't backed up using the SERVO software:

Some of you have wondered why we had such trouble reading votes from the Downtown West JBC.

Apparently a loose printer pin damaged some internal components. Fortunately the memory chips that serve as a back-up were not damaged and when they were moved to another board we were able to count them. But, you ask, surely we did not buy a system that holds all the votes in one box.

We did not. Due to the learning curve on the machines the technicians failed to back up the early votes cast on the eSlate machines. Had this been done, if a JBC failed, it would have been a simple matter to retrieve the votes in question by reading the backed up totals from the eSlates. This is the third back up you have heard about.

The important point is, even though one component was damaged, and one procedure was not followed, we were still able to get an accurate count. The Hart eSlate system and Hart Interactive Inc. stood up to the test and passed with flying colors.

Greg Mackay
AOE

OK, then. With the new information that the eSlate machines were not backed up, everything adds up now.

I would still like to know the procedure for replacing a failed controller. Do the connected eSlates also have to be replaced? And in the case of a "smoking" controller, why wouldn't the first thing you do be to obtain a backup of the votes from the eSlates connected at the time? Live and learn, I guess.

At any rate, the actual machine vote records do not appear to be available, and according to Hart's security analysis report this is one of the components used to cross-check the controller and flash memory card "votes of record" for inaccuracy or tampering.

In a case involving a corrupted flash memory card and a "smoking" motherboard and extraordinary intervention and handling of the controller internal memory chips by the vendor and by an employee of a third-party local company to retrieve the votes that decided a $50 million ballot issue, I'm guessing the Election Commission would have liked to have had the eSlate voting data for audit and verification purposes.