First I would like to express my deep appreciation to all
our County Council members, Ann McFall and her staff. You have all put in many
months of hard work to understand and employ evolving technology which
challenges even the most knowledgeable "techies" among us. You have
all shown courage, wisdom and great patience and, as your constituents, we are
deeply grateful.
Having made my own presentation to the County Council based
upon the evidence of others, the Annual Conference of the Florida State
Association of Supervisors of Elections at which many different voting machines
would be open to the public, seemed like an ideal opportunity to learn more.
With our wives in tow promising to help, Dr. Jerry
Axelrod and I drove to Wesley Chapel
Florida yesterday, June 7th determined to learn more about touchscreen
voting machines and their application for handicapped voters. Jerry¹s wife,
Paton is vision impaired and agreed to be the handicapped voter while my wife Dee ³voted² as a non-handicapped voter. The exhibit hall contained many different voting machines
and related products. The only three being represented as handicapped-friendly models were the Diebold
Accuvote TSX system, The ES&S Automark and the AccuPoll AVS1000.The first machine we evaluated was the Diebold Accuvote TSX
assisted by Wes Krivanek who, most generously, gave us several hours of his
time, and later Mark Earley. Two machines were used. S/N 202010, contained Georgia software and a disabled
voter interface. After several unsuccessful attempts to boot the system, the
disabled interface was moved to the 2nd machine, S/N 201267 which we were told
was programmed with Florida Certified Software. My wife and I then ³voted² on 202010 (sans interface) while
Paton, voted on the disabled configured machine, 201267. With the screen
blanked off, a synthesized voice led her through the ballot.
My wife had a problem that it took 5-7 screen ³pushes²
before any of her actions registered. Wes observed that and postulated that
perhaps her nails (which were slightly longer than mine) may be causing the
problem. Even with her repeated pushes, her vote took just over 3 minutes. I
had no problems and my fat fingers got a response on each touch, completing my
ballot in just under 3 minutes. Paton¹s vote using the handicapped audio interface to
outline the ballot through headphones took 31 minutes, much longer than I had
thought it would.. The handicap interface was a ³telephone keypad² style with
12 keys to be selected than pressed. To select the appropriate key number required sightless
touch-counting of the keys to locate the correct one before it could be
pressed. (Think of placing a call on a telephone in the dark)Wes felt 31 minutes
to be a reasonable time period for each vote and suggested it might be reduced
by not listening to all the ballot choices.
I also asked Wes about the printer option. He did not
believe 60 days was a realistic estimate and was very surprised that such an
estimate was given to the County. He said the specifications, description,
information contained, format, hardware and software engineering had not yet
been completed and could give no date certain when this would occur. In any event, the printer option would in no way provide a voter-verifiable
paper ballot nor any way of doing a meaningful recount should an equipment or
memory card fail since no voter-verifiable ballot was generated.I asked why ³bar code² (which puts another computer level
between the ballot and the voter) had been chosen. He said he wasn¹t aware that
it had been (See Contract - Pg. 3 of 46, 1.2) Another major problem surfaced in further discussions with
Mark Earley. I had the previous apparently erroneous impression, that the
Diebold TSX would electrically interface with the County¹s Diebold optical
scanning units. Mark explained that no such interface was now possible nor
was it planned. The TSX memory card would have to be separately outputted via
modem. This would require massive reprogramming of the County¹s Central
Tabulators which I understand are now programmed for a single output per
precinct.
Diebold plans to address this problem with new programs
being prepared over the next several months to accomodate an adapter interface.
This planned adapter interface is described on Diebold sales literature titled:
³Mixed Optical Scan / TSX Precinct² which I believe applies to
our County. The ³solution² creates many more obvious and serious problems. The Diebold procedure requires poll workers to remove
memory cards from the optical scan machines and place them into an Optical Scan
Accumulator Adapter which is then inserted into the PCMCIA slot of the TSX. The
newly designed interface programs would then consolidate all the data for transmittal
to central collection. Removal and reinsertion of small cards (roughly credit-card
size) carrying huge voting files by relatively untrained temporary poll workers
would seem to openly breach any semblance of security procedures.
These cards could be exposed to stray magnetic fields from
numerous sources ie: polling machines, air conditioners, fans, blowers, lights,
etc. The card could be easily corrupted as were the cards in the Daytona Beach precinct during
both previous Presidential elections. In those cases, the ballots were re-fed
into the optical scanner which saved the day for that precinct. Removed from the safety of their locked chambers, they
could be easily mislaid, misplaced, dropped, stolen or even substituted
with other preloaded cards which would then nullify the entire precincts
election return.Since the TSX produces no ballots, once the output of both
memory cards are combined, within the TSX program, the entire precinct could be
rendered useless by a corrupted or defective or damaged or substituted memory
card. In short, it leaves the County with no effective Plan B at
the precinct level when any single component of the thousands comprising Plan A
County-wide fails - as they have done to some extent in every previous election. Paton Axelrod also tested the ES&S Automark system for
handicapped voters, S/N ENG 023) as our seriously sight-impared voter. With the
screen darkened (as was the Diebold) and going through a similar audio
interface, Paton listened to the complete ballot and voted on all the choices. The voting took
only 9 minutes, less than one third the 31 minutes the Diebold required. The through-put of 6
sight-impaired voters per hour on the AutoMark vs only 2 per hour on the Diebold seems extrememly
advantageous. The sight-impaired AutoMark interface consisted of a large
round central button surrounded by four large triangular arrow shaped buttons
at the 12:00, 3:00, 6:00 and 9:00 positions. The points of the triangles pointed Up, Right, Down and Left
respectively in a manner similar to many TV remote controls. Paton reported she found the AutoMark Control interface
easier to use and more intuitive than the Diebold as it had larger and fewer
buttons and did not require searching the keypad for specific numbers. The AutoMark is also an optical scan unit. In contrast with
the Diebold unit, It produces a voter-verifiable ballot identical to the
present ballot now in use. After voter verification of the ballot, it is placed
in the existing optical scanner for tabulation. In the
event of any failed component, this ballot could be
re-scanned by the county¹s existing optical scanners thus providing the County
with a reasonable plan B to rescue a precincts election rersults.
Note: Mark Earley of Diebold could find no reason why the
AutoMark generated ballots would not scan just as accurately as the ballots
now being used by the County.
Another big advantage for the County in the AutoMark is its
optical scan system is very similar to the County¹s existing optical scan units
and require little additional training. The Diebold TSX with its obscure
virtual voting system is completely different requiring additional complex procedures for testing, trouble-shooting, verification, repair, data transmittal and card handling be taught to large
numbers of temporary poll workers - and repeated at considerable expense for
each election.
We also looked at the Accupoll handicap units which
produced a less desirable standard letter size ballot summary - but I felt
this machine, still under development, was too immature for consideration at
this point. None of the other machines I could see were displaying handicap
accessible interfaces. The Diebold Machine uses a Windows CE operating program
while the AutoMark can use either a Windows 2000 or Windows XP operating
system. Windows systems are attacked thousands of times daily by hackers
employing viruses, worms, spyware, etc. Both systems have been easily hacked
and remain subject to malicious or inadvertent programming errors.
Our impressions were neither scientific nor carefully
planned. None of us have the slightest scintilla of financial interest in the
outcome. Although registered Democrats and members of the Democratic Executive
Committee, this report represents no one but ourselves. The truth is that voting machine issues are and must be
Non-partisan. The rhetoric on both sides should be cooled down. What is
required is more light and less heat. We are all Americans and share the same
values. I believe a voter verifiable paper ballot is essential to restore faith
in our voting system and provide a ³Plan B² to rescue elections when the
systems we rely on fail - as they have done in the past and will surely
continue to do so in the future. Most repectfully, Spencer Lane
Addendum: After reading the above report, Paton Axelrod, our
sight-impaired handicapped voter, has requested I add an additional observation
which she found troubling. In the audio review of her ballot after it was cast on the
Diebold TSX Touchscreen unit with Florida approved software, the synthesized
voice says, "Your choice has been selected" without specifying just
what that choice was. Without audible verification in her headset she had no way
of knowing if the votes she cast were recorded correctly. This appears to be a violation of Florida Statutes, Title
IX, Chapter 101 , Section 101.56062 titled "Standards for accessibile voting
systems" subsection (1) paragraph
5. which reads:
5. "The voter must be able to review the candidate
selections that he or she has made." We again appreciate your consideration of our report and
hope it has been useful in your deliberations.
Spencer Lane
| 
