Experiences with Sequoia AVC Edge with VeriVote Printer as Precinct Inspector in Santa Clara County

From: Arthur Keller <arthur_at_kellers_dot_org>
Date: Thu Jun 08 2006 - 11:08:36 CDT

On Tuesday, June 6, 2006, I served as Precinct Inspector at Precinct
2112 in Palo Alto in Santa Clara County, California. This precinct
was combined with Precinct 2118, but there was a single roster book
and a single precinct board consisting of 6 poll workers, including
myself. I had previously served as a Precinct Inspector in the
November 2005 election using the Sequoia AVC Edge Direct Recording
Electronic voting machines, but without VeriVote printers. A
Precinct Inspector is the poll worker in charge of a polling place.
The VeriVote printers were added for June 6, 2006 as a result of
California state law that went into effect on January 1, 2006
requiring that Direct Recording Electronic voting machines have a
voter-verified accessible paper audit trail.

During poll worker training a few weeks before, some longtime poll
workers complained about the added work involving the use of the
VeriVote printers. These poll workers felt that the added security
was not necessary.

On Friday afternoon before the election, I went by the polling place,
a church, to obtain a key for opening the building on Tuesday
morning. A cart with the 5 voting machines had just been delivered.
The voting machines are strapped onto the cart, but there is no lock
or seal on them. On Saturday morning before the election, I went to
the regional satellite voting headquarters, another church parking
lot in Palo Alto, to pick up my inspectors materials to take home
with me. These materials included the activator and cards, the voter
roster, and the street address rosters.

On Monday evening, the day before the election, our precinct board
collected at the polling place to set up the voting machines. The
voting machines were extracted one by one from the cart, arrayed in
place, plugged one to another daisy chained and then to an electrical
outlet. We checked that they were all on. Some of them had the
yellow light indicating their backup batteries were being charged.
We also connected the VeriVote printers and turned the voting
machines on to test that they started up properly. We checked that
the voting cartridges were secured with numbered tamper proof seals.
We checked that the cover that is used to start and stop voting was
also secured with numbered tamper proof seals, but we did not open
that cover. We tested that the card activator turned on and
initialized. We turned off all the voting machines. We turned off
the lights and checked that the voting machines were still powered
on. We all left to return on Tuesday morning.

Our precinct board converged at the polls again on Election Day at 6
a.m. We checked the tamper proof seals, turned on the voting
machines, and enabled the start of voting. The VeriVote printers
printed a zero proof report, which also served to test the printers.
Unlike the November 2005 election, the start up process now involves
calibrating the touch screen. A few people were waiting patiently
for the opening of the polls at 7 a.m.

The sequence of steps for casting a ballot had changed from last time
because of the addition of the VeriVote printers. When the voter
reviews the ballot selections made, the voter may change one of the
selections by pressing on that contest. The voter is then shown the
original selection screen with that contest. Once the voter changes
the selection for that contest, as the voter presses next repeatedly
to get to the last selection screen, the voter can make additional
changes if desired. The voter then presses next to go the review
screen. Pressing next again gets to a message telling the voter to
press the screen to print the selections or review again. I was
asked by quite a few voters how to proceed if they did not want to
print their selections. I patiently explained that they had to print
their selections because the printout is used for recounts and we
need to be able to recount their vote, too. The entire list of
contests did not fit on one window's worth of paper audit trail. So
the voter was presented with a screen that gave the choices
"continue" or "review"; the "continue" choice printed the next
window-full of audit trail (the second and last batch) while the
"review" button voided the audit trail and displayed the review
screen again. Perhaps the "review" button should have said "make
changes" instead. After the final window-full of audit trail was
printed. The voter was presented with the choices "make changes" and
"cast ballot"; "make changes" voided the audit trail and brought back
the review screen, while "cast ballot" caused the "press to print
ballot choices" screen to reappear while the VeriVote printer printed
a bar code. Then a message appeared indicating that the ballot was
being recorded, while the VeriVote printer advanced the paper to
conceal the voter's choices. Once the vote was recorded, the
activation card was ejected, but that was before the voter's choices
had been completely concealed.

I think the voter could get to the review screen directly from the
choice for that contest, but I believe the voter could not continue
from that review screen to print/cast the ballot from there.

Some voters asked how they get "their" printouts. I told them that
we keep the printouts for recounts, so they don't get a copy. Most
people don't know what an "voter-verified paper audit trail" is, or
even an "audit trail"; it's jargon. I suggest that we refer to it as
a "printed recount record" as then people instantly understand why
the Registrar of Voters keeps it rather than the voter. The ATM
receipt analogy is misleading as there is no monthly statement to
reconcile with your receipt.

The contests were "paginated" into numbered screens (1/8, 2/8, etc.).
There was a feature to change the display type size. The result was
a magnified display of the same numbered screens but with scroll bars
added. The voters I helped seemed befuddled with the scroll bars, as
they required one to scroll down, then to the right and up, and then
down again to examine the entire numbered screen. You also had to
scroll to get to the "next" button. Repagination with one column and
a "more" button would probably have been less confusing. Some voters
who tried the magnified screen got frustrated and returned to the
regular screen.

The sample ballot in the county voter booklet matched a hand-marked
paper ballot used for absentee voting. More than a dozen voters got
confused with the organization of contests on the touch screen as
compared with the sample ballot. The sample ballot was printed on a
series of pairs of facing pages. The sample ballot was printed so
that you had to rotate it 90 degrees to turn the pair of facing pages
into "portrait" mode. The sequence of contests on the voting machine
followed column by column of the sample ballot. The problem is that
you had to go from one facing page to the other to read an entire
column and then back again to get to the next column. Some voters
got disoriented when changing to the next sample ballot page as they
either turned the booklet to the wrong sample ballot page or switched
the pair of sample ballot pages at the wrong time.

Santa Clara County now requires support for five languages.
Unfortunately, all of the voting machines except for the one with
auditory support only had space for English plus two other languages.
The auditory voting machine supported more languages. Our combined
precinct polling place had about 1350 registered voters. About a
third of those were marked as absentee voters, some of whom were
permanent absentee voters. About 370 of the somewhat over 900
non-absentee voters cast ballots at our polling place at election
day. Several dozen absentee ballots were dropped off at the polls.
None of the voters at our polling place used the auditory voting
feature. Some of the voters did vote in Chinese. Fortunately, we
had a Chinese-speaking poll worker, because (as I understand it) the
voter could not actually cast a ballot in Chinese. Apparently, there
was an error in the sequence for Chinese of printing the review copy
and casting the ballot. I recall the problem being that there was no
"continue" button on the Chinese screen. I personally find this
astounding for these reasons. (1) I expect the logic to be
language-independent, with the text being slotted in for each
language into the right locations in a template, (2) I expect that
the logic for Chinese-language multi VeriVote-window ballot casting
should have been tested by the ITA and the California Secretary of
State prior to certification, and (3) the Santa Clara County
Registrar of Voters has a Logic and Accuracy Testing process that I
expect should have caught this problem.

We had about 30 voters per hour throughout most of the day.
Occasionally, none of the voting machines were in use. Quite often
all 5 of the voting machines we were allocated were in use. Several
times (particularly in the evening) as many as 5 voters were waiting
with activator cards in hand to use the voting machines. A longer
ballot (we had only 2 statewide propositions, not the usual 10 or
more) would have taken longer, for both making selections and
printing the paper recount record. Voters will be more familiar with
the machines then, somewhat balancing the increase in time. If the
expected voter turnout is more than 400 voters per polling place or
the ballot is longer, then I would expect that more than 5 voting
machines would be required per polling place.

While my precinct did not have hardware problems with the VeriVote
printers, I understand from my Field Inspector that some other
precincts did have hardware problem. I understand that a nearby
precinct had two faulty printers, one that jammed and one had a
flashing light with an undetermined error. I don't know the overall
hardware problem rate for the county.

Arthur M. Keller, Ph.D., 3881 Corina Way, Palo Alto, CA  94303-4507
tel +1(650)424-0202, fax +1(650)424-0424
OVC-discuss mailing list
= The content of this message, with the exception of any external 
= quotations under fair use, are released to the Public Domain    
Received on Fri Jun 30 23:17:03 2006

This archive was generated by hypermail 2.1.8 : Fri Jun 30 2006 - 23:17:12 CDT