Uncle Diebold's Clubhouse

> Uncle Diebold's Clubhouse
> (http://www.nukefreezone.net/archives/000140.html)
>
> A first person's account of the Diebold electronic voting machine and
> its
> associated problems.
>
> On March 2nd, I was a poll worker for this year's California primary
> election. More specifically, I was a Systems Inspector in San Diego
> county,
> whose problems with voting machines and procedures received some
> coverage
> in the national media.
>
> First, a summary of my personal opinion: I think that current
> electronic
> voting systems are better than the traditional systems in terms of
> security, and also in terms of usability for most people. However, I
> share
> the opinion of many bloggers that major security issues remain in the
> new
> machines and implementations, and that these issues should already have
> been fixed.
>
> More details below…
>
> This was San Diego's first election using electronic voting machines -
> specifically, Diebold AccuVote-TSx stations. Previous elections in the
> county used punch cards. The county failed to make the mandated upgrade
> prior to the last election, and a federal court ordered that it be
> done for
> this primary.
> Previously, precinct boards in the county were made up of an
> Inspector, an
> Assistant Inspector, and clerks. As of this election, a Systems
> Inspector
> and an Assistant Systems Inspector have been added at each precinct.
> According to the Registrar of Voters, this is because a four-hour
> training
> session would have been required in order for Inspectors to learn both
> the
> general procedures and how to operate the machines. Instead, most of
> the
> technical details are left to the Systems Inspectors.
>
> I was contacted and assigned as a Standby Systems Inspector, meaning
> that
> if necessary, I would stand in for a missing Systems Inspector or
> Assistant
> Systems Inspector in my part of the county. The standby system is
> apparently not used very much; they forgot to handle some details, like
> sending me a copy of the poll worker's manual, or notifying me that the
> location for the mandatory training had changed. Fortunately, I'm
> fairly
> resourceful, and the classes were running late anyway.
>
> In the class, we were introduced to how the system works. Along with
> the
> usual paperwork and supplies, each precinct has:
>
> * A Precinct Control Model (PCM).
> * A number of voting stations (either four, six, or eight).
> * Two Voter Access Cards (VACs) per station, plus one or two extras.
> * Two Supervisor Cards.
>
> A poll worker (usually the Systems Inspector) sits in front of the
> PCM. One
> poll worker has each voter sign the roster, while another checks the
> voter's address on another list. That second worker points to the
> appropriate line on the address list, and the PCM operator sees which
> party
> to program a ballot for - with the party name never said aloud.
>
> The PCM operator then selects the party on the PCM's touchscreen, and
> inserts any one of the Voter Access Cards (VACs) for programming. The
> VAC
> is then given to the voter, who inserts the VAC into any one of the
> stations, and is then presented with the ballot for their party. After
> casting their ballot, the voter's VAC is ejected, and the voter is
> instructed to give it back to the poll staff. The VAC itself is not a
> ballot at all - it just authorizes a voting station to bring one up,
> and
> tells it which party's ballot to display. After a ballot has been cast
> using a VAC, it must be reprogrammed on the PCM prior to being used
> again.
> We were warned that some voters might try to cheat by claiming that
> they
> received the wrong party's ballot. We were advised that, should this
> happen, we should insert the card in a station to make sure that it
> had not
> been used to cast a ballot already; then, add one to the tally sheet of
> programmed but uncast ballots, and reprogram the VAC after checking the
> voter's registered party on the street address list.
>
> That was about it. We were shown the startup and shutdown procedures
> for
> the machines, and cast a few sample ballots with them. The regular poll
> workers were noted on a list, and some paperwork or other was handled.
> I
> asked about getting ahold of a poll worker's manual, and was promptly
> given
> one from a large box that was sitting on the curb. It contains
> operating
> passwords for the machines, and voice certification codes, all
> laser-printed and mass-produced in binders.
>
> The passwords were obviously not chosen for security. Nevertheless,
> I've
> blacked them out in the complete copy of the manual that I have posted
> for
> your downloading enjoyment. NFZ's research suggests that it's legal
> for us
> to do this. If you don't already have it, grab a BitTorrent client
> (such as
> the official one), and then start downloading! The archive is 11 MB.
> After
> downloading, please leave your BitTorrent window open for as long as
> possible. BitTorrent works by having every user both download and
> upload,
> taking most of the strain off of our server. If more users continue
> uploading, things will be faster for everyone, and we won't have to
> pay as
> much to host the file.
>
> So, how did the election itself go? Avi Rubin mentioned that his fellow
> poll workers were dedicated to making things work right, and I can say
> the
> same thing. Few people without a strong sense of civic duty would work
> a
> fourteen-hour day for $60 or $100. We all did a good job, with only
> minor
> exceptions. Of course, those exceptions are what I'll be concentrating
> on
> here. Almost all of the voters were very understanding, and loved the
> new
> machines after they voted. We only had two people who went out of
> their way
> to be difficult, although a few more did voice concerns about paper
> receipts.
>
> At 6:00 AM on March 2nd, I was dispatched to a precinct two suburbs
> south
> of here, where their Systems Inspector was unavailable due to a case
> of the
> flu. Solana Beach is a small and affluent suburb, although it was
> originally a grungy surfer colony. I don't know if it has anything to
> do
> with the fictional community of the same name in Tentacles. The
> precinct I
> worked at was one of the more affluent: the poll was held in a
> resident's
> garage on a ridge perhaps half a mile inland, with an excellent view
> of the
> Pacific.
>
> My arrival was practically Messianic. It was ten minutes prior to
> opening
> time, the PCM was having trouble, and one of the stations wasn't
> working.
> Fixing the station was a fairly simple matter - the stations are
> connected
> in a daisy-chain configuration, and one had been plugged into itself
> rather
> than its neighbor. The PCM was simply very confused, and kept giving us
> dialog boxes with various runtime errors. Power-cycling the PCM made it
> work for about 25 ballots, but it then reported more runtime errors
> related
> to the smart card reader. I was worried that data might be lost, but
> the
> Inspector went ahead and restarted the PCM once again. It worked for
> the
> rest of the day, with no apparent data loss.
>
> I found out later that a cable allows the PCM to be plugged into line
> power
> while in its sealed case, and that there had been an extended power
> outage
> the day before in many areas. This is apparently what caused PCMs in
> several places to show a Windows CE desktop instead of the voting
> software.
> I heard from one source (which I forget, unfortunately) that this was
> because part of the boot loader was stored in volatile memory. I have
> no
> idea if our problems were related to power.
>
> We plugged in the station and fixed the PCM, then opened the polls
> about
> five minutes late. Fortunately, we didn't come close to turning anyone
> away; a number of precincts in the county apparently opened an hour or
> more
> late. As instructed, we asked the first voter to inspect the "zero
> report"
> from each machine before closing the printer doors for the rest of the
> day.
>
> About half an hour into the poll, one voter had apparently gotten
> halfway
> through her ballot before the station began going to the previous
> page. I
> had her go forward to the last page on which she had been able to
> vote, and
> we both stepped back and watched the machine page backwards to the
> instructions. I'm guessing that this was a broken touchscreen. It
> worked
> well enough for me to cancel the ballot and move her to a different
> station; I closed the broken station for the rest of the day, and our
> Inspector called the troubleshooter hotline. There were supposedly one
> hundred Diebold employees and some county troubleshooters covering
> about
> 1,200 precincts, which seems generous to me, but they didn't dispatch
> anyone. Our problems were apparently minor compared to those at many
> of the
> other precincts.
>
> Most of the day was fairly routine. We tended to have patches of heavy
> traffic, lasting for maybe twenty minutes, and then quiet periods
> before
> the next rush. The busiest time was around 5 PM, when we had six people
> voting and another six in line - still not too much for us to handle.
> We
> did have some problems figuring out what to do about absentees and
> exceptions. San Diego lets you vote in a precinct other than your own
> using
> a "provisional ballot," which is also used in some other situations.
> With a
> provisional ballot, your vote is tied to your ID. After Election Day,
> the
> Registrar of Voters reviews the provisional ballots to make sure that
> each
> voter is, in fact, entitled to vote; their vote is then counted. The
> system
> makes things complicated for everyone, and I'm not sure whether we
> actually
> did a few of the provisional ballots properly. (For more details on
> provisional ballots, take a look at the poll worker's manual.) – [I
> downloaded a copy if anyone wants it, 11 meg - Ivan]
>
> Another procedural problem, which I understand was the case in a
> number of
> precincts, was that many voters' parties did get said aloud. This was
> especially problematic for voters who were registered as Non-Partisan,
> and
> were allowed to pick which party's primary they wished to vote in.
> Explaining how this worked usually took several minutes per voter. One
> voter was quite upset after voting; he had chosen a Non-Partisan
> Non-Partisan ballot, but had wanted to vote in the Democratic primary.
> After that happened, we told Non-Partisan voters to ask for help before
> casting their ballot if they didn't see the expected options.
>
> We had another technical problem later in the day, when a voter
> reported
> that his summary screen (the last step before casting the ballot) was
> blank. I confirmed that it was; everything else seemed normal, and the
> boxes were checked next to each candidate on the ballot, but there
> wasn't
> anything on the screen where the votes should have been listed. I
> moved the
> voter to another station, chalking it up to a card programming error,
> and
> kept checking on that station. Another voter had the same problem
> shortly
> afterwards, and I closed the station. It appeared to have recorded all
> of
> the votes properly, but I can't be 100% certain. If it hadn't recorded
> some
> data, there wouldn't have been anything that we could do. Again, the
> troubleshooter hotline didn't send anyone out; we only had a few more
> hours
> to go, and still had four machines working.
>
> Closing went fairly smoothly. I noticed that the ballot memory cards
> are,
> in fact, standard SanDisk 128MB PCMCIA flash cards; Avery labels on
> them
> identify the precinct and machine number. As Avi Rubin mentioned, the
> memory cards were behind locked doors, although the keys were fairly
> simple
> and identical county-wide. (We didn't check on slow voters as much as
> in
> his precinct, but I did develop a Pavlovian response to the sound of a
> VAC
> ejecting.) Another poll worker accidentally ejected one of the cards
> before
> powering off the machine, and contrary to my expectations, no warnings
> appeared. As other poll workers have undoubtedly complained before,
> there
> are far too many materials that have to go in far too many different
> pouches.
>
> We drove to the local collection center in order to drop off the
> ballots
> and equipment, and upon arrival, the Inspector was very unhappy.
> Normally,
> a staff of five or six quickly unloads everything, and a uniformed
> Sheriff's deputy signs for it. Instead, we waited in a long line.
> There was
> one person loading machines into a rented truck, an
> authoritative-sounding
> guy in a hat who turned out to be the deputy, and two volunteers
> helping
> the precinct officers unload their own cars.
> As the Inspector was driving me home, she asked: "So…how many ways
> did
> you think of…?"
>
> "Quite a few."
>
> Posted by jrenken at March 16, 2004 05:46 PM
>
>
> Ivan Handler
> Networking for Democracy
> [email protected]
>
> Version: 6.0.581 / Virus Database: 368 - Release Date: 2/9/2004
>
>
>
>
_________

Lee A Wells
mobile: 917 723 2524
studio: 718 349 7951

[email protected]
http://www.leewells.org