Some Info on the Diebold Machines:

The State of Wisconsin Approval Process - key to this is the following part of the Help America Vote Act (HAVA from here on out):

Section 301(3) of the Help America Vote Act (HAVA) requires that every voting system used in a federal election:
be accessible for individuals with disabilites, including nonvisual accessibility for the blind and visually impaired, in a manner that provided the same for accesss and participation (including privacy and independence) as for other voters....through the use of at least one direct recording electronic voting system or other voting system equipped for individuals with disabilites at each polling place.

That is where the reqirement that we get some kind of 'touch screen system'.

Because we in Winnebago County use the Diebold Systen now, the product that was looked at was the Diebold TSX DRE.

This has been looked at by many different agencies. The one I quoted in yesterday's debate was the California Secretary of State (16 page report).

Some of the items I pointed out were:

From the data we estimate the Mean Time Between Failures (MTBF) of these machines to be approximately 15 hours under the conditionsexperienced during volume testing. It is unclear what failure rate this might imply for a real election.

We found the many software failures potentially more troubling than the paper jams. It seems likely that further changes to the AccuVote TSx software will be required. Under one possible interpretation of the standards, the failure rate observed during these tests was more than 10 times higher than permitted by federal standards (which require a 163-hour MTBF). The failure to detect this fact during the ITA’s testing process appears to be due to serious defects in the testing methodology specified by federal standards. One lesson of this analysis is that the testing performed during the federal qualification process is apparently inadequate to ensure that voting machines will be reliable enough for use in elections.
...

We found that there were 34 failures, spread across 29 distinct machines. We classified each failure into one of two categories: (a) printer jams, and (b) software failures, where the touchscreen machine crashed, froze, hung, or reported an unrecoverable error condition. The 34 failures broke down into 14 printer jams and 20 software failures, with 12 machines experiencing at least one printer jam (2 machines suffered from 2 printer jams) and 18 machines experiencing at least one software failure (2 machines encountered 2 software failures). One machine experienced both a printer jam and a software failure.
...

During a typical election, the polls are open for 13 hours. If the conditions during a real election were comparable to the conditions during the volume test, then we could use the MTBF to estimate the number of failures likely to be observed during an election. For instance, if we assume that we can recover from printer jams, but machines are taken out of service upon any software failures, then such calculations would suggest that almost 40% of machines would experience a software failure and need to be taken out of service, leaving only 60% of machines in working order by the close of polls.
...

Under these assumptions, some polling places would be left without any working machine by the end of the day. Obviously when the failure rate is this high, recovery from failures is a critical issue.

...
In general, we are concerned that the prevalence of software failures during the
June 20th test may indicate software quality problems in the TSx. It is possible that these failures are a sign of a large number of other latent software defects. As far as we know, there has never been another volume test of the TSx that tests the machine under realistic conditions, and generally at best spotty records kept of any failures that may occur during elections, so there is no way to know the extent or magnitude of the software quality problem.
...
Another vexing problem with recovering from software failures is that there is no clear, general way for a poll worker to determine whether the voter’s vote was recorded before the failure. We are very concerned that many of the software failures encountered in the June 20th test violate this crucial requirement. Some of the failures led to a crash after a ballot had been cast; some led to a crash before the voter had begun voting; yet the screens displayed after the crash gave no obvious way to distinguish between these two situations. What can a poll worker do when they are called over after such a failure?
...
Recovering from printer jams generally involves opening up the printer unit and manipulating the paper tape. This is a sensitive operation. It is analogous to opening up a paper ballot box in the middle of an election and inspecting and manipulating the ballots contained therein. If not performed properly, this could endanger vote integrity and privacy.

I will have more as I have time. But this is the gist of what I went over last night.