January 28th, 2005 ERC DRAFT Minutes

DRAFT

MINUTES OF MEETING OF JANUARY 28, 2005

BOULDER COUNTY ELECTION REVIEW COMMITTEE

1. Call to Order, Roll Call

The meeting of the Boulder County Election Review Committee was held on January 28, 2005, at 1:00 p.m. in the Houston Room of the Boulder County Clerk and Recorder Building. The meeting was called to order by Richard N.Lyons, Chairman, at 1:05 p.m.

The meeting was attended, at roll call, by the following Committee Members: Richard Lyons, Tom Davidson, Linda Flack, Jay Harbour, and Michael Taylor. Also present were William Compton, Director of Elections, Department of State, Howard Harris, President of Eagle Direct, Julie Ellis and Bill Schafer of the Eagle Direct staff , and members of the public.

Paul Tiger arrived at 1:20 and David Leeds at 1:33.

2. Approval of Minutes

Corrections to the January 21 meeting were recorded and the minutes were approved by unanimous vote.

3. Public Comment

There was no public participation at this time.

4. Presentation by Representative by Eagle Direct Printing

Mr. Howard Harris began his presentation by complimenting Linda Salas and her elections staff on their determination to record every vote accurately, stating that Boulder County is fortunate to have the kind of dedication they have exhibited.

Mr. Harris read from a ten page, prepared statement, inviting members to interrupt with questions they might have about any part of the presentation. (EXHIBIT 11)

He began by reading the print project metrics, including Hart Specifications, Production, Project Dates, and Notable Meeting Dates.

In Specifications it was noted that no print specifications and no testing equipment were available.

Project Dates show that shipping was complete on October 4, 2004 and print production took 5-10 calendar days.

Mr. Harris then related the information exchanged in the Notable Meetings:

November 4, 2004 County, Hart and Eagle This was the first indication to Eagle Direct of a potential problem. At a meeting that day Ms. Salas informed Eagle Direct that a printing problem had occurred. She showed them 3 ballots in which Eagle noted printing errors having to do with overtone, undertone and fuser problems. In the examination of ballots, they observed many write-ins, overvotes and undervotes. They also saw damaged ballots, in some of which the reason for the damage could not be seen. There was a significant amount of confusion about the damaged ballots and the tolerance for variation in the Hart/Intercivic voting system.

November 5, 2004 County, Hart and Eagle

The group began by observing the ballots being read on the screens. Many of the ballots being corrected were write-ins. Other rejects were due to very tight tolerances the machine was reading. Ballots with boxes almost imperceptibly off were rejected. The Hart engineer, when asked, responded the tolerances were plus or minus 20% of their calculated expected box position. He said that, probably, neither the county staff nor Eagle knew there were such tight tolerances.

In addressing the software program that read the ballots, the Hart engineer stated that the various sizes of bar codes did not affect the system because their triangulation process would adjust accordingly. In consideration of other perceived errors that might have caused misreads, they were told by Hart that these were not a problem because the software triangulation process would correct for all anomalies.

The Hart engineer who attended the meeting could not answer some of the questions that were asked by the Eagle representative, but he said that a VP was flying in that day and would answer more questions. Hart has not answered the question and has supplied no further details or exact specifications.

The Boulder County representative at the meeting stated that the entire lateness of the ballots were not exclusively a print problem, and she listed other limiting factors that contributed to the lateness of the ballot count.

At the end of the meeting the group agreed to an official statement that Eagle Direct printing was likely a part of the problem but not the entire problem.

November 9, 2004 Xerox Machine Technicians and Eagle Eagle staff met with Xerox Machine Technical and Service Representatives and a Product Development Engineer in order to try to identify apparent print problems. The obvious overtone, undertone and fuser conditions were noted. Xerox Staff was somewhat confused regarding the bar code and box inconsistencies.

In response to a question by Mr. Taylor, he was told that three models of Xerox machines were used. They pointed out that Xerox service specifications stated on 20 lb. stock a plus or minus 0.8mm movement could be expected on 110 lb stock plus or minus 1.2mm. It was calculated that the box has a read tolerance of 1.6mm. The group wondered why the Hart tolerances were so small. They seem too small for the Xerox machine tolerances. Tests showed that the machines’ magnification was within tolerances.

In using a film positive to give a visual test, a “bubble” of magnification appeared to be present only with the Xerox 6145 machine. This bubble effect made some, but not all, of the boxes move out of the approximate Hart tolerance level.

Speculation of the group was that a combination of the known Xerox tolerances and the unknown Hart system tolerances might be too small for real production conditions. They also speculated that storing the ballots for a month before usage may have impacted the sheet size and skew.

December 20, 2004 County, Hart and Eagle A report, given to Eagle showed 12,114 damaged ballots and 27,223 damaged contests which created approximately 14 hours per scanner. 24,44,824 ballots were produced with 4.9% damaged.

Eagle requested, from Hart, a copy of a computer simulation demonstration of the printed box movement, in order to more closely analyze the precise movement. The request was not granted by Hart.

Eagle made a formal request of Hart for the following detailed information: 1. Printing specifications given to Boulder County, 2. A detailed report of Ballot Now data, 3. The quantity of ballots that appear to be defective for specific causes, 4. A sample of each ballot that appears to have specific problems, 5. A sampling of precincts’ defective ballots with defects clearly marked 6. A copy of the industry standards that Hart used in designing their tolerances 7. A written document outlining Hart’s specifications and tolerances. 8. A list of other cities and commissions with which Hart has contracted in order for Eagle to compare printing services

January 20, 2004 First Ballot inspection/testing This was the first opportunity Eagle had to access real ballots and the read system. 54 ballots were selected, that were thought to have a high potential for defects. Evidence of heavy toner was used as the clue to predict defects. The ballot sets were scanned 3 different times on the same Hart machine. Scanning results were recorded as follows: 1st scan, 54% had one or more defects.

2nd scan, 43% had one or more defects.

3rd scan, 25% had one or more defects.

22 ballots after 3 scans, showed no defects.

32 ballots showed defects in one or more scans. Defects seemed to appear randomly throughout the three scans, in different places on different sides of the same ballot. One contest that should have been counted as an under vote was accepted by the system as a valid vote.

January 25, 2004 Eagle Telephone Conversation with Hart InterCivic General Manager Mr. McClure, Hart General Manager pointed out that scanning ballots that did not meet Hart’s specifications, run through multiple times, would show varying results. He explained the difference occurred as the scanner pulls the sheet through differently each time the sheet runs through the scanner. The boxes may be read a pixel or so different at each scanning, causing different results. (A pixel at 400 dpi scan would be 1/400 of an inch tolerance.)

Mr. McClure was questioned as to the sheet and box tolerance. He could not answer the question, but promised to get the standards information along with the source of industry standards and a copy of the standards to Eagle by January 28. The information has not been forthcoming.

It was deemed by both parties, that the conversation was not productive and it was discontinued.

In conclusion, Mr. Howard Harris stated that there were some legitimate errors in printing. He believes that print errors were only one of the contributing factors in those 12,000 of 252,000 ballots in error. He suggests other contributing factors were:

Too tight tolerances within the system design The lack of manufacturing specifications

No equipment to test ballots through the production run, leaving some aspects of quality control to be performed during the live election process.

During and after the presentation there were questions and discussion about the paper used. Eagle responded that the paper used was appropriate for this project.

Clarifying the size and placement of the boxes, it was stated that the box size and placement were in the postscript software. Boulder staff had control only of font.

Ms. Flack questioned as to subcontracting of some of the print contract. Mr. Harris replied that a portion of the project was subcontracted to John Phillips Company. They were under Eagle supervision at all times. Mr. Harbour asked if errors could be identified by printing machine. Eagle suspects that problems occurred on one (6135) machine.

Specifications to Eagle were for paper only. Hart was to determine all others, prior to election. Post election, Hart has not been forthcoming with specifications.

Mr. Tiger questioned the cause of the duplicate numbers on ballots. There was some speculation about the occurance, but the issue remains a question. Mr. Harbour requested information from the Clerk’s office regarding the number of duplicate ballot numbers that occurred. Eagle staff will request this information from the Clerk.

Ms. Flack asked about Eagle contracting with other Colorado Counties. Ms Ellis responded that they have, and she will provide further information in that regard.

Mr. Lyons requested background information about Hart InterCivic. Mr. Harris responded: Hart InterCivic, Inc, is a 50 year old organization. They started printing high tolerance work in the seventies, for the Denver Post, where printing could not be off more than 100/1000 of an inch. They were approached by a balloting company about 10 years ago, and went through their certification process. This is their first contract with Boulder County, and first experience with Hart. They expect 100% perfect printing. Realistically they have less than 1/2% error, anything more than that is unacceptable.

Mr. Schafer presented the 33 test ballots that were rejected in the test run and pointed out the discrepancies in designated errors. Members then examined a small number of these ballots.

The Chairman, Mr. Lyons asked if the committee wants to enter some of the ballots into the file. Mr. Taylor asked that the committee be allowed to keep the 33 ballots for further examination. Eagle will retain custody for now and will turn them over to the committee at a later date.

5. There was no presentation by an expert in printing

6. Other business and committee members’ comments

Mr. Lyons stated that the next meeting, on February 4, will be held in the Houston Room. Hart/InterCivic staff is tentatively scheduled to make a presentation. Mr. Lyons has not been able to contact Tom Halicki to appear on January 11. He will try to set a meeting with Kodak representatives. An evening meeting for members of the public to address the committee will be held on a Thursday, possible dates for that meeting are February 17, 24 or March 3. Members were asked to bring their calendars to the next meeting so that a date can be agreed upon.

Regarding the wording on the ballots, it is understood that the entity holding the election provides the wording. This should be clarified with the election office.

Mr McClure, of Hart, was in attendance at the meeting, he stated that he will appear before the committee next week to answer technical questions. Mr. Taylor and Mr. Tiger will send questions to Mr. McClure.

Mr. Compton stated that he appreciates the opportunity to attend the committee meetings because the State Election Office has some of the same concerns.

Documents provided by Ms. Flack were entered into record. They are the 2004 election calendar and chronology of voting procedures from both the County and the State and election terms and definitions, (EXHIBIT 12)

Also entered into record was State Electronic Voting System Implementation, Boulder, Colorado, Variance Analysis Questionnaire. (EXHIBIT 13)

7. Public Comment

Mr. McClure stated that he has a number of comments related to the meeting today, particularly in reference to the 0.8mm tolerance. He will explain next week, an important misuse of information.

Joe Pezzillo thanked the committee for their thorough work and asks that the committee make all documents in exhibits available to the community. Mr. Lyons pointed out that all committee proceedings and relevant information is public record. Mr. Tiger offered to scan his materials onto a web page. He commented that Colorado Voter.net should and would put all documents on their web site.

Neil McBurnett asked that all documents be provided in the most easily accessible format.

Mr. Pezzillo aproached the topic of duplicate serial numbers on ballots. Mr. Compton stated that the State is revisiting the waiver.

The meeting was adjourned by unanimous vote at 4:45

___________________________ Recording Secretary

Return to ERC Agendas and Minutes