Re: Shamos Rebuttal, Draft 3--error versus fraud, other items

From: Ron Crane <voting_at_lastland_dot_net>
Date: Wed May 11 2005 - 13:42:40 CDT

On May 11, 2005, at 11:11 AM, Edward Cherlin wrote:

> On Tuesday 10 May 2005 18:34, Ed Kennedy wrote:
>> Hello All:
>> On the incompetence versus fraud issue I think a good test
>> would be to examine in who's favor these 'errors' trend.
>> Being mindful of Jim March's feelings, I'll just say that I
>> get the impression that these errors tend to more favor
>> Republicans than Democrats. One could argue, in my partisan
>> way, that a consistent error in favor of one party or the
>> other or possibly even just in favor in incumbent office
>> holders, would suggest fraud is a more likely explanation for
>> these errors. If incompetence were a more likely source, I
>> would expect the errors to be more randomly distributed....
> Actually, this cannot be taken as evidence for intentional
> cheating. It could be unconscious, like women teachers
> unconsciously favoring boys in their classrooms.

It could be unconscious, though that is still cheating. But it could be
conscious. You are applying the "assume incompetence before malice"
razor. Though it is a very useful tool for personal relations (e.g.,
it's better to assume one's mate runs over one's foot in error than
because she hates you), it's a poor guide for public policy, since it
gives cheaters a much broader potential field of play. It places the
burden of detecting cheating on the public, rather than placing the
burden of proving the absence of cheating on the vendors and on the
governments who contract them.

> In the ESP experiments at Duke, believers scored higher than
> chance when counting their own results in one particular
> experiment, and non-believers scored lower than chance.
> Photographic evidence showed chance results on both sides. There
> was no evidence of people trying to fudge the results, but they
> were more likely to correct errors that went against their
> preconceptions.

Which is just another form of cheating. This experiment just shows that
people easily take a "stake" in outcomes, and are willing to "correct"
results in order to support their stake, even though there are no
tangible benefits in doing so.

> Similarly the IRS reports that arithmetic errors in tax returns
> significantly favor the filer.

And I daresay that many filers probably make a conscious decision
carefully to review unreasonable results that favor the IRS, and
carefully to avoid reviewing unreasonable results that favor
themselves. That some also do this unconsciously doesn't mean it's not

> In voting software, the presumption then is that more effort
> would be put into fixing bugs that might harm one's favored
> candidates, simply through unconscious bias rather than
> conscious malice.

I don't see much support for this presumption.


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 Tue May 31 23:17:33 2005

This archive was generated by hypermail 2.1.8 : Tue May 31 2005 - 23:17:52 CDT