Notes from Feb. 15th Unilect re-examination in Harrisburg:
(As observed by Mark McPherson, representing Mercer County Citizens for Better Government)

I personally arrived at the North Office Bldg. at about 8:10 AM.

I met the press person (Lynne Landes) and Computer specialist (Rebecca Mercuri) representing the person from Beaver County who filed the initial petition for re-examination (Sheila Green) at 8:20 AM.

The meeting was delayed due to Dr. Shamos's (the Examiner) flight being delayed. Mona Accurti, the head of the Bureau of Elections, was picking him up at the airport. He arrived about 9:45 AM and we were ready to start at 9:50 AM, but Commissioner Michelle Brooks was late.

We started the examination at 10:10 AM. (Mrs. Brooks arrived about 5 minutes later.)

The PA Dept. of State was represented by:
Dr. Michael Shamos (Examiner)
Albert Masson
Mona Accurti
Jonathon Marks
Brian McDonald
+ 6 others.

Unilect was represented by:
Jack Gerbel
George Mitchell

Dr. Shamos asked Mr. Gerbel to run through the entire process from developing the appropriate infopack for an election to closing the polls election night and transmitting the relevant information to the county board of elections.

Mr. Gerbel, prior to starting the process, reiterated his stand that the primary problem in Mercer County, which brought about the re-examination of his system, was based on Mr. Bennington's coding error. (This does not take into account the fact that it was Sheila Green's petition from Beaver County that actually brought about the re-examination.) Mr. Gerbel stated that he felt that if James Bennington of Mercer County had not made the error that we would not be here.

Mr. Gerbel also stated that the error on Mr. Bennington's part was compounded by the fact that he did not test each and every machine to verify that they were working appropriately. This was further compounded by the fact that our county commissioners did not follow through and re-test the machines to verify that they were in good working order. He stated that he felt that had this been done that the problems would have been quickly detected and easily corrected prior to the election.

Mr. Gerbel went on to state that he also felt that part of the problem was due to lack of training on the part of the county with regards to election workers. He told those present that he had spoken with people in Mercer County and had been told that election workers had only received about 15 minutes of training just prior to the election.

Mr. Gerbel also stated that he felt that a substantial portion of the problem in those precincts with undervotes was due to the fact that voters need to be trained on how to use the Patriot system.

Dr. Shamos jumped on this statement. He said that the system is supposed to be easily understandable and easy to use for voters. He commented that it was rather untenable with regards to training every voter in Pa.; the system should be simple enough and with adequate safeguards to ensure an accurate and straightforward election.

Mr. Gerbel went on to state that there were only 13 precincts in Mercer that he was aware of that had the undervote. This is incorrect-he was told by Kathleen Paul during the meeting he had with the counties' appointed investigation panel that there were an additional 28 precincts with a greater than average undervote, for a total of 41 out of 100 precincts in the county.

Dr. Shamos was very concerned with the testing capability with regards to the touch screen portion of the system. He inquired about the ability to test them prior to the election to verify they are all in working order. Mr. Gerbel looked surprised and dumbfounded with the question. He looked at his system specialist and asked if there was that possibility and they both hesitated and then said that there was no real way to do that and that there was an extremely low failure rate on the touch screens. When asked by Dr. Shamos what that rate was, Mr. Gerbel stated that they got 15 units a year for repair, then immediately retracted the number and said it was more like 10 per year. When Dr. Shamos asked how many touch screens were in use around the country, Mr. Gerbel said that there were about 800 units in use. Dr. Shamos then stated that in relation to the number of hours these units are in use each year that that was an unacceptably high rate of failure.

Dr. Shamos then proceeded to run several tests of the touch screens and processor systems to simulate possible voter problems and was able to repeatedly cause the system to have problems recording data correctly. He stated that he could see a number of areas where there were statutory issues as well as reliability and usability issues with the system.

During the test the system dropped the write-in candidates that had been entered into it. The system also failed multiple times during the process and had to be corrected/fixed by Mr. Gerbel, often with little or no explanation by him as to the cause of the problem or the fix.

Dr. Shamos stated that he was concerned that the flexibility of the PC being used to download the infopacks after closing the polls gave users with access to the system the opportunity to change the data. He felt than manual manipulation of data with regards to the final vote count was possible as the system was now configured. Authorized PC users can modify data and can hide that fact if no use log is in place.

Dr. Shamos was also very concerned over the lack of a paper trail for provisional ballots. As demonstrated by Mr. Gerbel the system only shows provisional ballots on the PC used to download data post-election.

Mr. Gerbel said that the PC used for the Patriot system was not designed to be able to be connected to a network or to the Internet so that it could not be corrupted or manipulated by an outside source. He also said that any modification of the PC or attempt to do so would violate the manufacturer's warranty. This is one of several security levels provided by the Patriot system. Mr. Gerbel stated that currently any locality interested in purchasing these systems are required to purchase the PC to be used with the Patriot system from Unilect.

When asked by Dr. Shamos about the type of encryption used on the programs running the unit, Mr. Gerbel said that the programs are not encrypted, just obfuscated. Dr. Shamos seemed stunned by this info and said that if that were the case, if someone were to figure out the logic pattern regarding that program, the system would be wide open to outside manipulation.

When asked by Dr. Shamos about the seals used on the Patriot system on election day, Mr. Gerbel said that they were not tamper-proof, just tamper-evident. Dr. Shamos seemed to be somewhat concerned with this issue.

Mr. Gerbel said that this re-examination of the Patriot system was the first in the units' history.

There was a short break from 12:20PM to 12:30 PM.

Dr. Shamos was quite obviously upset with the report on Mercer county's problems submitted to him by our commission that studied the issue. He specifically asked Mr. Gerbel several questions based on info from the report, to which Mr. Gerbel responded, "I don't know." He also asked pointed questions given to him by Kathy McPherson (of Mercer County Citizens for Better Government) to which Mr. Gerbel responded, "I don't know."

Throughout the re-examination process Mr. Gerbel made repeated bad jokes and dumb comments that did nothing to reassure us that he felt the gravity of the issue for those county representatives present.

During the portion of the re-examination where data was entered in the system to verify accuracy of recorded data, the second touch screen attached to the voting unit failed immediately at startup. (It locked up.) Shortly thereafter, the printer jammed and Mr. Gerbel had to print out the starting data a second time.

After entering all the pre-selected data, the system was shut down and the results tallied. Both sets of data matched 100%.

The meeting was adjourned at 2:20 PM.

After the re-examination was over, Dr. Shamos allowed a question and answer period that ended up running 30 minutes. I personally pointed out to Dr. Shamos that I kept hearing them go on and on about the 13 precincts in Mercer county, and that that was inaccurate. I told him that there were 28 other precincts (all of which were in a different Congressional District) that also failed and had nothing to do with the previously mentioned coding error. I told him that as yet that there had been no explanation for it.

During the Q&A session, Dr. Shamos stated that, historically, from 1948 to the first use of the DRE machines the undervote has never exceeded 3%, nor been less than 1%.

Dr. Shamos suggested that there may be an ability to check whether de-selection of candidates was the chief problem in the election. He said that if one were to print out the anonymous record from a unit used outside of the 13 precincts, one may look to see if there is a high record of de-selection. (This is noted in the record if there is a higher than average rate of de-selection in an election.)

We finished at this point and I left the re-examination room.