Re: Left off the ballot?

From: Douglas W. Jones <jones_at_cs_dot_uiowa_dot_edu>
Date: Thu Apr 15 2004 - 14:11:40 CDT

On Apr 14, 2004, at 12:10 PM, charlie strauss wrote:

> Historically, this fact has been used for rigging. Punch cards can
> be "rigged" at the printing plant by filing down the dies that
> pre-perforate the chads, making the chads for one candidate slightly
> harder to dislodge than another.

I have seen this allegation repeatedly, but have never seen any
evidence that this attack has actually been done. I strongly doubt
that it has, for the simple reason that punched-card ballots are
ordered in lots of 1000, while the average precinct has only 500
votes in a presidential election year, and ballot layouts vary from
precinct to precinct. In general, you can't know, at the time a
card is perforated, what state or county that ballot will end up
going to, much less, what precinct.

The way to rig a punched-card voting fixture (aka Votomatic punch)
is to use some scotch tape to attach a piece of 2 mil steel shim
stock to the face of the T-strips behind the voting position you
want to disable.

> 1) systematically misallign the touch-screen and image so, for
> example, democrat buttons are harder to actuate than republican,
> leading to missed votes.

The neat thing about this hack is that you set it up using the touch
screen calibration routine. No software changes required. Just put
in your supervisor card, type in 1111 (or whatever is required to get
into supervisor mode), select the "calibrate touchscreen" menu, and
then bias all your touches to the left, right, top, or bottom. If you
bias all your touches to the left, you'll make it really hard to
vote for candidates on the left side of the touch screen, while making
it really easy to vote for those on the right.

Do the "audit trails" on the touch screen even record a record of the
recalibration of the touchscreen? I don't remember seeing records of
that the time I tried this on an ES&S E-slate. The results were
pretty neat, thought.

> 2) Systematically, making changing the votes say 2% of the time so
> that votes toggle to another candidate before the ballot is printed
> out. Sure the voter can verify on paper this happened but what
> fraction actually will.

This requires research, but for example, if 50% of all voters check the
paper copy, and 2% of the votes are change, you'll need about 100 voters
before there's a 50% likelihood that someone will catch the problem.

If the software messes with 1% of the votes and 50% of the voters check,
you'll have to have 200 voters to have even odds of catching the
problem.

Given that 400 voters is typical, if the machine is dishonest with 0.5%
of the vote and only half the voters check their ballots, half the
precincts won't see a problem. Half will see one or more error reports.

My guess is that if only one voter out of 400 complains, people will
write them off as mistaken. In 1/4 of the precincts, though, two voters
will notice the problem, and given that there are 10s or 100s of
precincts in a typical county, some precinct will have enough complaints
that the precinct officials get upset.

So, the effectiveness of voter self-audits on the system depends
critically
on the frequency of voter's checking the paper. It also depends on the
instrucitons given to the polling place workers about how to respond to
voter complaints. If the polling place workers are told to reassure and
help, but not to make a record of the complaint, we'll never know about
low-frequency problems. If those records are never examined, we'll
never
see how frequent those problems really are across multiple polling
places.

                                Doug Jones
                                jones
==================================================================
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
==================================================================
Received on Fri Apr 30 23:17:07 2004

This archive was generated by hypermail 2.1.8 : Fri Apr 30 2004 - 23:17:29 CDT