Re: A generic best practice document for New Mexicolegislators

From: Ken Pugh <kpughmisc_at_pughkilleen_dot_com>
Date: Wed Dec 29 2004 - 09:33:13 CST

At 02:38 PM 12/28/2004, you wrote:

> > At 12:41 AM 12/27/2004, you wrote:
> >>3) Innocent anomalies will occur. Without open systems, errors, fraud,
> >> and innocent anomalies can appear indistinguishable; for elections to
> >> be trustable we have to be confident we can distinguish these.
> >
>I'm not sure who wrote this (and I may be taking it out of context):

This statement was taken from the best practice document.

What the remainder of my message was is to request estimates on what is a
reasonable expectation for accuracy of a vote counting system and for the
accuracy of a recount.

For example, if ballots were not secret (they were identified by id) and
voters could examine their ballots in a recount(say in an on-line
spreadsheet), could one expect a 100% match between voter intent and the
actual vote? Any mismatch may be due to human error, scanning error,
failure of the voter to examine their ballot, or other cause. I don't
believe that 100% accuracy is possible. In this instance, I might estimate
that .001% of the votes on the ballots would be incorrect. I have no data
to base my estimate on, just a gut feel.

The estimate of error in any voting procedure should be significantly less
than the margin of victory in order to assure that the candidate who
actually received the majority of votes was declared the winner. For
example, if the estimate of error was .001%, then the margin of victory
might have to be .01% in order to have a 99.999% probability that the
correct candidate was declared the winner.

I'm not a statistician, so I leave it up to others to calculate what those
percentages are in reality. But I strongly feel that any specification for
a voting system should state percentages for accuracy. Those percentages
can be tested for any system.

[Okay, so you are left with the possibility that the margin of victory was
significantly closer to the estimated error. Recounts may keep switching
the victor every time. This mess is now being played out in some elections
today. The expectation of absolute accuracy is not reflected in the
reality of voting. Perhaps any vote where the margin of victory is so low
should be decided by cutting for the high card, like they do in ties in

Some messages on this mailing list have suggested that OVC take a look at
the wider issues in voting. Let us suppose that whatever counting system
was used (DRE, optical scan, paper) were almost perfect (99.99999% accurate
in its counting). But the voter sign-in numbers numbers were different
than the vote count. That is going to introduce errors (or at least
questions) in the votes. That error may swamp any error in the counting
system. If one does not address the wider issues, there could exist large
discrepancies between the real vote and the reported vote.


>Any statistically significant anomalies should be "explainable" by some
>decernible factor IMO and never assumed to be innocent.
>Distinguishing what causes anomalous patterns may not always be possible
>even with detailed investigations, but noticing them may be important and
>cause good things to happen nonetheless.
>OVC discuss mailing lists
>Send requests to subscribe or unsubscribe to

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 Fri Dec 31 23:17:21 2004

This archive was generated by hypermail 2.1.8 : Fri Dec 31 2004 - 23:17:22 CST