[extropy-chat] US election prediction markets failure

[Brent Neal]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


On Nov 13, 2006, at 5:10, BillK wrote:

>
>
> The problem with predicting trends, whether by computer software or by
> markets, is that a trend is only a trend until it isn't.  Drastic
> changes are notoriously difficult to predict.
>

I wonder if it was really a failure to "predict a drastic change" or  
whether it was a problem with an ill-posed question.  The question  
that Orlowski used in his article was posed as something like "GOP  
retains Senate."   That's a binary proposition and one that we  
understand now (and I believe we understood before the election) to  
have been a very close thing.   I wonder if the question had been  
posed as "Number of Seats lost by GOP"  if the idea market would have  
given a more accurate result.

My experience with the ideas futures through the Foresight Exchange  
is that many questions are posed in binary fashion - all or nothing.   
I would be interested in seeing how accurate those are as a category  
compared to questions that have a range of answers.

Off topically, I would be intrinsically distrustful of Orlowski's  
articles, as it has been shown that his intellectual integrity will  
often take 2nd place to his desire to write sensationalist stories.

Brent


- --
Brent Neal
Geek of all Trades
http://brentn.freeshell.org

"Specialization is for insects" -- Robert A. Heinlein


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFFWILu0Sh4Y5m/F5kRAiTDAKDfcoKDCvMh6oTs1YAEIk85xNeCAACgyrXc
cxnUUOXibaZkOV5vvw1g5cE=
=/WHU
-----END PGP SIGNATURE-----