Re: Ballots for elections with many races

From: Douglas W. Jones <jones_at_cs_dot_uiowa_dot_edu>
Date: Wed Apr 28 2004 - 08:51:33 CDT

On Apr 27, 2004, at 4:42 PM, David Mertz wrote:

> I spoke with new OVC subscriber Kevin McDermott last night. He
> raised the issue of elections--like those apparently held in Cook
> County, IL--with a very large number of contests in them.

Ted Selker has been raising the same question. He has a sample general
election ballot formatted in cash-register receipt format that runs on
for about a meter. On a cash-register type of printer, this would be
very likely to take over a minute to print.

> But I don't see how 100 Judicial Confidence votes can fit on one
> piece of paper. Maybe with a really small font.

There is no doubting that such ballots pose big problems. This is why
Ted Selker doubts the effectiveness of voter verified paper and has
been proposing an alternative model of voter verification (headphones
with an audio feedback given at the time the voter selects a candidate
or yes-no position, and an audio recording of the entire voting session
from polls open to polls closed used for audit.

(Selker's model doesn't preserve voter anonymity because of the
real-time
nature of the record, but let's assume that problem can be solved.)

After arguing this point with Ted Selker, we agreed that what we need
are
two things:

1) Experiments. Build a Fraud-O-Matic version of the voting machine
and see how many voters catch the fault.

2) More experiments. Study reward structures that encourage voters
to take the audit step in the process seriously. (As one behavioral
psychologist said, when asked whether people could effectively verify
their ballots, it all depends on how you reward them.)

3) Yet more experiments. Study variations in printing style, ballot
layout, etc in order to find what form of voter verification works
best.

Every voter-verified paper trail proposal should be given this level of
study. Until this is done, most of the objections to such proposals
are hot air, but so are most of the arguments in favor of such
proposals.

Unlike all those closed-source voting systems from the industry, our
open-source prototype is a perfect platform for conducting such
experiments.

> Or a creative use of space that e.g. groups together the Yes, No,
> and No Preference Judges within different boxes on the ballot.
> But overly creative typography and layout makes voter verification
> less simple and reliable.

Exactly why we need to experiment.

> So do we just spill over to as many pages as are needed? Is it legal
> to only record the relatively rare "No confidence" votes, and just
> leave off the usual "Yes" or "No Preference" votes?

The answer to all such "Is it legal" questions depend on state laws,
which vary all over the map from state to state. In most states,
though,
I'd guess that it wouldn't be legal to summarize things this way.

And, of course, there is the issue of leaving someone's name off the
ballot. If a touch-screen user interface for the California recall
election had left off a candidate, would anyone notice. This argues
for parallel testing to check this aspect of the system.

                Doug Jones
                jones@cs.uiowa.edu
==================================================================
= 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 23:17:20 2004

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