Re: How dark is a mark?

From: Douglas W. Jones <jones_at_cs_dot_uiowa_dot_edu>
Date: Wed Dec 19 2007 - 10:51:38 CST

On Dec 19, 2007, at 10:10 AM, Richard C. Johnson wrote:

> Doug Jones cited "competently made erasure" as a standard for
> ballot scanners. I would respectfully disagree. Erasure is far
> too variable; I would encourage voters who make an error to return
> a spoiled ballot for a fresh one and vote their intentions.

Absolutely! There is a big difference between what we recommend
to voters and how we test the calibration of scanners. We recommend
that voters not erase, we recommend that they spoil the ballot and
get a new one (at polling places) or force an overvote with a big
X to cross out the bad choice and properly vote their intention
(so that the canvassing board can resolve the correction) on an
absentee ballot.

But, when calibrating the scanner, it is a good idea to calibrate it
so it ignores well-made erasures and hesitation marks, while sensing
that single stroke through the voting target using the recommended
marking device.

> Erasure is never a good idea with scanners, and I would not allow
> erasers in a polling place.

You cannot prevent a voter from making an erasure. You cannot prevent
a voter from using whatever pen or pencil is handy. The most you can
do is recommend and encourage.

> The most egregious provision would have voters erasing in the
> case of over votes rather than turning in the spoiled ballot for a
> new one.

We did this in Johnson County Iowa for several years. Each polling
place was equipped with a new gum eraser. Voters voted with number 2
hard lead pencils on ballots scanned by Optech I and II scanners (they
are the same machine). The scanners were configured to kick back
overvotes. If the voter's erasure worked, the scanner would accept
the ballot. If not, it would kick back again and the voter was given
a replacement. There was no override provision on this scanner, it
was either kick back all overvotes or silently accept all.

Hava sent those scanners to wherever old equipment goes after it's
illegal to use in polling places. Probably some sweatshop
disassembly line in China.

> Hesitation marks, as opposed to a stroke of the pen or pencil, can
> readily be ignored by scanners.

Not so easily. The Optech 4C was developed to scan pencil, and then
they re-tuned it to scan ballpoint pen in Phoenix. They ran the
sensitivity up to the point where it could almost reliably sense a
single stroke of a ballpoint, and at that point, any faint pencil mark
or complete erasure was sensed as a vote. Subliminal marks were

Bic black ballpoint ink is not very black when illuminated with a
red LED (the Optech 4C does not use infrared to scan voting targets,
but it does use infrared to scan the index marks, according to the

> Otherwise, it would be nightmare for voting officials to have to
> evaluate one erasure against another, especially where an erasure
> is attempted on a full marked ballot choice.

The scanner's ability to ignore erasures and smudges is a matter of
making a good faith effort. Beyond that, the instructions should seek
to minimize the frequency of these efforts.

Technically, distinguishing between erasures and deliberate marks is
a matter of looking at the derivitive of the intensity signal. You
are looking for edges. Deliberate marks have sharp edges, while
erasures and smudges do not. Some discrete sensor scanners had a
degree of edge sensitivity to help them ignore smudges. To my best
knowledge, the current crop of pixelizing scanners (dating back to
1995) are all based on simple intensity measures over the sensing
area. This will change.

                Doug Jones
OVC-discuss mailing list
By sending email to the OVC-discuss list, you thereby agree to release the content of your posts to the Public Domain--with the exception of copyrighted material quoted according to fair use, including publicly archiving at
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Mon Dec 31 23:17:06 2007

This archive was generated by hypermail 2.1.8 : Mon Dec 31 2007 - 23:17:10 CST