Re: Are errors in elections acceptable/inevitable

From: Edward Cherlin <cherlin_at_pacbell_dot_net>
Date: Mon Jun 20 2005 - 03:07:24 CDT

Ed Kennedy wrote:
> Hello Teresa and All:
> So you believe there should be zero tolerance for errors?

You're fighting a strawman. "We of the profession are not
accustomed to speaking in such inexactitudes." The point is not
to have zero error, but to be able to account for the known
errors, and to determine whether there are enough of them to
compromise the outcome. (More than the margin of victory, for
example.) There will be errors. For example, voters will "flee"
with uncast ballots, leaving an electronic record of a vote with
no paper. If we count that vote, we open the door to tampering
with the machines. If we don't count that vote, we open the door
to tampering with paper ballots. So we don't decide in advance
either way. If there are enough errors to put the outcome in
doubt, we do a full audit, and go to a judge to get a
determination on how to apply the law. If the outcome is not in
doubt, we do a full audit anyway. In both cases we need to
determine the likelihood and extent of deliberate tampering, and
if necessary redesign the system to prevent it next time.

> Would you advocate that any election with errors be rerun?

Under some circumstances, this might be the only way, but only if
the number of errors is greater than the margin of victory, and
perhaps not even then. Every effort is made to avoid this
result, since it requires that someone not elected to a position
carry out its duties until the new election is certified. We
have seen something similar in the state of Washington, where
the court challenges from last November were concluded in May.
It turns out that the Governor is the Governor, but this
conclusion was in doubt for six months.

> I
> can imagine how a determined spoiler could deliberately
> introduce errors in every election and call attention to them
> if results were not what they desired.

Then we must make sure that we can catch such spoilers.

> I'm a civil engineer and I've been taught that random error in
> inevitable. Engineers are directed to look for systematic
> errors and compensate for them. For example, there is only a
> finite precision in surveying equipment. Yes, before anyone
> says it, the potential error is pretty small with the newer
> equipment but it is still there. When you dump your total
> station's data into a computer it always finds an error and
> redistributes it along the results typically by a least
> squares method.

I am familiar with "error of closure".

> I suspect this is old news to you with your
> major statistics background but I'm mentioning it for the
> benefit of everyone else. Also, before anyone mentions it,
> obviously, the absolute theoretical limit of accuracy in
> voting is one vote while survey equipment can continue going
> for finer and finer divisions of distance and angle.
> Do you or anyone else believe that any error beyond one vote
> is unacceptable? I'm bringing this up (again) because I've
> never gotten a satisfactory answer to what is the acceptable
> level of error (say as a percentage of total votes cast) in
> the voting process.

"Any measurable level of error is too high."--Bill Godbout

No, there is no "acceptable" level of error in voting, any more
than in air transportation or data backup, but there is a level
at which the outcome of an election is not in doubt, and we can
carry on.

> I know that there are rules and laws
> specifying automatic recounts which could be used as a legal
> basis for maximum error. I believe these vary substantially
> from state to state. Does anyone have a summary of these?

It isn't that simple. The laws are so vague and
self-contradictory in many locations that only the courts can
make a decision that is generally accepted--sometimes only the
Supreme Court.

> Thanks, Ed Kennedy

Edward Cherlin
Generalist & activist--Linux, languages, literacy and more
"A knot! Oh, do let me help to undo it!"
--Alice in Wonderland
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 Thu Jun 30 23:17:09 2005

This archive was generated by hypermail 2.1.8 : Thu Jun 30 2005 - 23:17:11 CDT