Boulder Elections Review Committee

[Joe Pezzillo <joe@xxxxxxxxx>]

I attended the first meeting of the Boulder Election Review Committee 
yesterday at the Courthouse. Here are some of my notes.

The meeting was about two and a half hours, started with some 
introductions, Linda Salas gave a lengthy presentation, the by-laws 
were reviewed and adopted, and finally the three members of the public 
got to speak.

The makeup of the committee has been published in the paper, and I'll 
say that I was impressed with the quality of some of the questions, and 
the willingness of the committee members to make sure their questions 
were answered. One detail that might not have been known in advance is 
that Drew Durham has been appointed by the Secretary of State to fill 
the seat on the committee that Paul Danish requested for her office.


Some Information from Linda's Presentation:
Linda said that among other circumstances, the 2 days of LAT testing 
and the 2 days in court contributed to their delay in counting, as did 
a statewide delay due to the Nader inclusion lawsuit.

According to my notes, they had 13,000 damaged ballots, with 27,000 
damaged races.

There is a suspicion that a "toner" issue may have been involved, faded 
printing was apparently visible on some ballots that were inspected.

All undervotes were autoresolved.

The counting was finished Friday afternoon.



Things the Committee Asked About:

The Paper Ballots & Printing
	Often in detail, e.g., "Did the postscript files contain individual 
ballot images?"
	Questions about pre-election testing and quality control sampling 
procedures for the ballots.
	Were the ballots printed on a printing press or a high speed photocopy 
device?
	Was a specification provided by the system vendor for their ballots?

Precinct Optical Scanning
	Much talk about Larimer County and "what went right" there.


Other info from questions and answers:

A Hart representative (Chuck Logan) present stated that there is 
supposed to be a 20% tolerance around the box for the mark on the 
ballot.

There was some discussion about how many times the ballot stock was run 
through a printer, 3: watermark, signature and ballot text + barcodes, 
with ballot text and barcodes both during the same pass.

Errors began to be noticed after about 1,000 ballots had been run, and 
then on all systems simultaneously, but there was no indication at 
first, or from absentee/early counting.

A Kodak representative was called in on election night and determined 
there were no scanner errors.

Discussion of paper weight and postscript file preparation as well as 
other printing specs.

Hart had never previously experienced a barcode failure.

Discussion about the printing vendor (Eagle Direct) and that they had 
called Xerox representatives, although they had also discussed 
Heidelberg printing presses and "multiple printers" (interpreted as 
multiple printing devices by the Clerk's office, but may have meant 
multiple subcontractors).

Questions about how the ballots were serialized, the number of ballot 
styles (322 styles for 227 precincts), individual ballots in postscript 
files, large PDF files that could not be FTP'd to the printer.

The ballots cost $160,000, or approximately one dollar each.


the Committee asked for a sheet of definitions for the various acronyms 
and other terms that are used




Assertions that Should Be Examined:

Precinct Scanning is a Good Solution
Central Scanning is Always Easier than Precinct Scanning
An advantage of precinct scanning is that the voter can have the 
scanner verify that the ballot is counted correctly (this was said, I 
think what was meant was that the verification is that the ballot is 
marked correctly, i.e., no overvotes).

There were no problems in the Primary Election.



Some other interesting observations:

Scanning was centralized in Boulder, despite the earlier Commissioner's 
direction to spend additional money to have a scanning/processing 
center in Longmont.

There was little or no questioning of the issue of the appropriateness 
or Constitutionality of the barcode or serialization, but there was 
plenty of discussion about the barcodes themselves, the sequential 
balloting, and so on as they related to the issue of the 
damaged/rejected ballots.



The committee worked on its by-laws, probably the key details of which 
as far as CVV listmembers are concerned is that public comments will be 
limited to 3 minutes (although they would probably entertain giving 
time to another), and they expect written materials to be submitted at 
least 24 hours in advance.


Peter Richards spoke, introduced and provided some background.

Beat Fehr presented information on the Swiss Voting System, including 
an estimation that counting 160,000 ballots could be done in about 2 
hours in the 227 precincts.

I presented a copy of the Jan 20, 2004 Clerk's correspondence about the 
evaluation materials, called attention to the Daily Camera editorial, 
and referenced materials submitted to Boulder City Council about the 1 
to 1.5% error rate that has been documented in the system, and said 
thank you!

Please let me know if you want to know more about any of this. I'm sure 
I've overlooked something, but I hope nothing of major consequence.

Joe

Joe Pezzillo, Citizen Activist
Boulder, Colorado USA
jpezzillo@xxxxxxxxx