On Wednesday, October 19, 2005 Diebold and Fidlar were in Madison to show of the Diebold DRE, GEMS and Diebold optical scanners.
The display was impressive and I am confident the Election Administration Council will foolishly vote for certification of these machines.
The accuracy question was question brough up up prompted by me. She asked: "How can I sign off on the check box for accuracy when we did not do that this morning?" The staff admitted to the council they were not able to get the machine tallies to agree with the expected results of the test deck. The staff was willing to write off the discrepancy as "Tester input error". The staff was asked why the test deck was not reentered. Apparently entering the test decks was so arduous the staff was unwilling to "waste time" and actually test the accuracy of the DRE software properly.
Unfortunately, what I saw not only supported but deepened my concerns regarding testability of these machines under 5.84 and accuracy and correctness of these 2 systems (GEMS and optical scanner) under 5.91(11).
The screen, user interface, 4 point stand, and ergonomics are phenomenal on the DRE. Unfortunatey, my concerns about the machines complying with 5.91(11) and 5.84 have little to do with these flashy, but superficial externals. As much as these features improve accessibility for the disabled (blind, deaf and wheel chaired) they do for accuracy or correctness. The underlying security defects from version GEMS 1.18.17 exists in GEMS 1.18.24. The security defect in AccuVote OS Precinct Count firmware 1.94w is still present in firmware version 1.96.6. The same defect of the precinct count 1.96.6 firmware is present Central count firmware 2.0.12.
The security defect on the optical Scanners (OS) is IF
the memory card can be swapped with an altered memory card, the optical scanner can do nearly nearly anything under programatic control when the zero totals or tallies total reports are invoked. The details of the security defect are described in the Hursti Report
I am writing a report to the State Elections Board recommending against certification of the system as presented on Wednesday, October 19, 2005.