From: "Bruce McNaughton" <@earthlink.net>
Date: July 24, 2006 4:17:30 PM MDT
Subject: [GPOC] Report -- Voting machine tests, Denver, 7-20-06
I received an emailed invitation from the Denver Election Commission to
"be on the Test Board for the Public Voting Machine Logic and Accuracy
Testing scheduled for Thursday, July 20, 2006." I accepted.
By agreement, the four-citizen volunteer Test Board met Wednesday at
the DEC warehouse to select the machines to be tested the next day.
We were told the statute required testing of at least one of each type of
machine to be used in the election, but no more than five machines total.
Denver has 1180 Sequoia 'Advantage' direct recording electronic (DRE)
voting machines, in use since 1997. Sequoia was also the vendor selected
by DEC to supply 240 new DRE machines, the new Sequoia 'Edge' model,
to comply with "access for disabled persons" provisions of the 2002 fed-
eral Help America Vote Act, and new state laws requiring a voter-verifiable
paper record.
The Test Board selected one of the old Advantage machines to test, and
four of the new Edge machines.
On Thursday 7-20 the Test Board gathered in the conference room of the
DEC, 303 W. Colfax, to spend a long day at the "Public Logic and Accuracy
Test." Our instructions were as follows:
1. Each Test Board member (TBm) will fill out 34 ballots, one for each
ballot style that Denver will be voting on for the Primary Election.
2. Each TBm shall secretly vote their position and retain a record of the
tally of their test votes.
3. Each TBm must vote two of their ballots as Audio Votes.
4. Each TBm must mark every vote position for every candidate on every
race.
5. Each TBm must include overvotes and undervotes in each race.
Each TBm having marked and tallied the thirty-five ballot sheets (seventeen
Democrat, seventeen Republican, and one Unaffiliated), we were herded
to the voting machines across the room and asked to witness the Zero Proof
Report process, which is a piece of paper produced by each machine when
the machine is turned on to open the polls. The report lists each race in the
election with a zero tally after it, "proving" that all counters in the machine are
set to zero before voting begins.
The TB then decided to eliminate two of the Edge machines from testing
because of time constraints -- as each TBm would have to vote each of her/
his 35 ballots on each of the machines tested, and it had been agreed to try
to wrap up by 6pm if possible.
Each TBm then proceeded to punch in all votes on all ballots on each mach-
ine. The Advantage test machine was familiar to each TBm, having been in
use in Denver elections since 1997. It accepted all votes as cast on the touch-
screen and tallied correctly at the end of voting.
The Edge machines had not been used by any of the TBm before this, but
weren't hard to use. The machines are much smaller than the Advantage
models -- about the size of a big suitcase when folded up. (When Election
Commissioner Susan Rogers commented that one of her great objections
to paper ballots was the ease of stealing and disappearing ballot boxes, I
mentioned that the new Edge machines are smaller than most old ballot
boxes and have a convenient handle for a thief.)
The Sequoia Edge voting machine has a small touchscreen (approx. 15"x15")
on which ballots and instructions appear as selected, much like an ATM.
To the left of the touchscreen is a smaller (approx. 4"x15") plexiglass screen
behind which the printed ballot selections appear on a scrolling paper when
the "print" button is selected. The paper record is not accessible to the voter,
and can be voided by the voter. The paper record remains in the machine
and becomes the official record in case of recount, according to new state law.
Of the two Edge machines tested, one had printer problems after just a few
ballots were cast, and required fifteen or twenty minutes to fix.
The Audio Vote components of the Edge machines were very "cludgy" and
difficult to use. The audio instructions received via earphones were confusing
and very time-consuming. None of the TBm lliked the Audio Vote system. In
fact all TBms accepted a Sequoia employee's offer to vote the required
ballots in the TBms stead. "Needs work", admitted DEC Communications Dir-
ector Alton Dillard.
After all TBms had entered all ballots on each of the test machines, we were
taken to the back room where absentee ballots are counted in one of two
DEC computer-based optical scanning machines. Having flipped a coin to
determine which of the two machines would be tested, the operator demon-
strated the machine's self-testing and diagnostic routines, and then fed all
140 (35 x 4) ballot sheets into the machine, which whirred and flipped and
popped and delivered itemized totals in about thirty seconds, having sepa-
rated out all ballots that had overvotes on them for individual attention.
DEC operations coordinator then compared TBm tallies with machine tallies
and pronounced the machine tests "near flawless", after sorting out a "human
error" that happened when one TBm voted the same ballot twice, somehow.
"You happy now, Bruce?" asked Election Commissioner Susan Rogers,
with whom I've been having a friendly but distinctly adversarial discussion
about electronic voting machines for several years. "Happier than I deserve
to be, Susan. Thanks for inviting me."
The "Public Logic and Accuracy Tests" proved that Denver's voting machines
can do arithmetic. I'm glad, but we knew that. We also know that the trade-
secret software that runs these machines is capable of much more sophistic-
ated manipulation of numbers than simple arithmetic. And that's a problem.
because we're not allowed to find out what manipulations might be going on
with our votes.
B McN
_______________________________________________