Re: Accuracy and repeatability in opticalballotscanners.

From: Ed Kennedy <ekennedyx_at_yahoo_dot_com>
Date: Wed Feb 02 2005 - 23:20:21 CST

Hello Charlie:

I'm trying to get a handle on your 'specified difference' by looking at
specified or measurable accuracies of optical scan machines. While optical
scan machines aren't perfect, I think that they are a good intermediate
steps towards our easier to read, by machine, bar coded ballots. As we are
probably going to be stuck with optical scans as an effective standard for a
decade(?) we might as well use them as our basis for accuracy, reliability
and repetability IMHO.

As you your suggestion about games of chance for settling ties, I know that
there's a town in Arizona that's called Show-Low. It seems that the
inhabitants had an election over what to name the town and it tied. The
name of town was decided by a game of cards. Rather than being stuck with a
sore loser, the winner named the town after the play that won the card game.
Also, I believe that most numerical election ties are specifically decided
by a game of chance such as a coin toss. I'm sure that the fine folks in
the State of Washington might find your suggestion quite worthwhile.

If we decide to live with commerical optical ballot scanners as an
intermediate step, this might give new life to the Markamatic ballot printer
concept. D'ya think that printing the optical scan timing lines along with
the votes would be viable? I'll note that the Diebold scanners I used last
November have to use card stock especially for recounts because they'd just
destroy anything lighter.

Thanks, Ed Kennedy

charlie strauss wrote:
> My modest proposal is that all election returns that lie within a
> specificed difference are declared ties and settled by a game of
> chance. The logic here is that in addition to machine accuracy there
> is a more fundmental limit. A perfect machine can tell you precisely
> how all the voters voted. But it cant tell you the uncertainty in
> assessing either the voter's intent or the intent of the large
> citizenry. Voter intent gets screwed up because voters make mistakes
> when they vote. Judging from the only measure we have, undervote
> rates, this is possibly as high as 1% per question on the ballot.
> Citizen intent measurement gets screwed up because things like long
> lines, snow, traffic, flu, polling place locations, election judge
> stupidity..., greatly influence outcomes in pathological ways that
> sometimes are reflected in the people who were prevented from voting
> not those who voted.
> At some point any election is a tie in the sense that both candidates
> are equally qualified and desired in the eyes of the voters. That is
> a large fraction voted for each so by definition they are qualified
> regradless of other measures of their intellect and ability. Thus
> selecting one of them by a game of chance is not an affront to
> democracy.
> Since this would eliminate almost all challenges to an election and
> elminate any need to recount on suffieintly accurate machines it
> solves a heck of a lot of issues in voting and promotes public
> confidence that no swindle took place.
> -----Original Message-----
> From: Joseph Lorenzo Hall <>
> Sent: Feb 2, 2005 5:45 PM
> To: Open Voting Consortium discussion list
> <>
> Subject: Re: [OVC-discuss] Accuracy and repeatability in optical
> ballot scanners.
> On Wed, 2 Feb 2005 15:53:00 -0800 (PST), Edmund R. Kennedy
> <> wrote:
>> I'm thinking that whatever the error rate is
>> discovered for optical ballot scanners that these be
>> used as the threshold numbers for setting accuracy
>> standards for voting equipment and for recounts. I
>> know that there was a discussion on this a month or
>> two ago but I think that there was always the problem
>> of how to set a numerical base level of error
>> tolerance. Seeing as optical scan equipment seems to
>> be the best pragmatic standard hardware I could see
>> this as a good point to nail down error tolerance.
>> Does anybody have any comments or ideas about this?
> There are some documents that have come to light in FOIA requests that
> the BlackBoxVoting crew have made... let me see if I can find them:
> Here we go... there's some indication of error rates with Diebold OS
> machines here:

OVC discuss mailing lists
Send requests to subscribe or unsubscribe to
= The content of this message, with the exception of any external
= quotations under fair use, are released to the Public Domain
Received on Sun Feb 27 17:17:02 2005

This archive was generated by hypermail 2.1.8 : Sun Feb 27 2005 - 17:17:13 CST