[EM] Observations of a poll worker

Alex Small asmall at physics.ucsb.edu
Wed Mar 6 12:56:13 PST 2002

Hello everyone.

Yesterday I was a poll worker in my precinct in Santa Barbara California.
We talk so much about theory on this list, I'd like to talk about voting in
action for a bit.

I was very pleased with the AccuVote machines that we use.  We have
optically read ballots where voters simply fill in ovals.  If the machine
detects an over-vote the ballot is rejected and the voter receives a new
ballot.  If we should ever have a race as close as Florida's we might have
to worry about under-votes, but not over-votes.  This isn't true in every
county, but a ballot initiative passed to fund the purchase of better
voting machines in counties that need them.

(According to one poll worker SB County used to have the punch-card
machines, but those machines were sold to Florida several years ago.  I
don't know if the story is true, but it was not presented as a joke!)

The potential for rigging the election is quite small.  At the beginning of
the day the machine gives a printout of the totals for each candidate, and
we verify that it's zero, so even if somebody tried to run ballots through
it in advance it would be detected.  You can't reprogram the machine at the
polls.  I don't recall if the printout was time-stamped (I wasn't in charge
of the machine) but if there's no clock in the machine it would be even
more difficult to rig the election, because you couldn't program the
machine to give a zero total in the morning, but a rigged total in the

The voter lists were well-maintained.  We issued only 3 provisional
ballots, and those were for people who lost their absentee ballots.  We
don't check ID (against the law) but since we keep track of who already
voted (we have a list of all registered voters) anybody posing as another
voter would be detected if the actual person tried to vote that same day.
Since we get signatures from each voter and the county has a signature on
file with the voter registration the fraud would be detected in due time.
I don't know enough about the voter registration procedures to assert that
they are immune to fraud, but if they are as careful as the election day
procedures I have confidence in the process.

The weakest link in the process is low voter turnout.  All of our
procedures are pointless when less than 15% of the precinct even cares
enough to vote or request an absentee ballot.  That is sad when you think
of how many people today would sacrifice a great deal to vote (and how many
already have in the past).  The head of our precinct is an Army Reservist
who said that (a) he doesn't jump out of planes for the money, and he
doesn't work the polls for money either and (b) if you can get thousands of
people to protest something why can't you get them to vote?

Also, the low voter turnout is the greatest window for fraud as long as the
laws forbid checking ID.  If hardly anybody votes, posers are unlikely to
be caught since the odds are slim that they person they impersonated will
show up.  I personally see little harm in checking ID, and I would
encourage the lawmakers to consider it, but that's a separate issue.

Finally, to get back to election methods, the AccuVote machine could easily
accomodate Approval Voting.  In one county race (electing party officials
for the county, since it's a partisan primary) I had the option of picking
up to three candidates out of eight or so.  I'm sure it could be programmed
to handle up to eight of eight.

I encourage everybody hear to vote whenever there are elections in your
area, and volunteer at the polls.  We can discuss the best way to resolve
cyclic ambiguities until our fingers fall off from typing, but somebody
still has to keep the polls running.

Alex Small

More information about the Election-Methods mailing list