Is Bruce Bueno de Mesquita's predictive forecasting work for real?
Gaetan
who has made a name for himself by developing mathematical models
based primarily on Game Theory and somewhat on Rational Choice Theory
to forecast rather complex outcomes mainly in the field of political
science (foreign policy outcomes, national elections, etc...). He
often refers to a CIA study that confirmed that his predictive models
were 90% accurate.
I would be really interested in the assessment of BBM's methods from
mathematicians who use Game Theory and are familiar with BBM's work.
Thanks in advance for your input.
Econotron
news:2821aa79-a65a-4187...@x6g2000prc.googlegroups.com...
Rich Ulrich
I haven't seen much response in the sci.stat.* groups, so I will
toss in a comment. I'm not expert in those areas, but I did see
the author on Jon Stewart's comedy news show.
The Subject line asks whether his work is "for real."
I would say, Yes, if that means, "Does he do something that
can be taken seriously?"
It is more of an open question whether his work justifies special
hype as "90% correct" or "better than the CIA estimates".
Who was judging? ... what sort of questions? ... and who selected
'CIA estimates' that were being compared?
On Stewart's show, he described a rational basis for his work
which is rather better than where a lot of people start -- He
says, he collects a small group of experts about an area and
asks them (as individuals, meeting in a group) to tell him which
*people* matter in making a decision. For instance, the president
of Iran does not have a very important opinion, because other
actors in Iran are much more powerful and influential.
- That might suggest that his best predictions would concern
issues where the outcomes are *deliberate* rather than
accidental or determined by imprecisely described resources.
I got the impression that, somewhere down the line, he was
using mathematical models. Whatever he is doing, it ought
to help that he is making use of data that is probably too often
ignored. The fact that he is publishing bias-free results might
make him a grade more reliable than what might be gleaned from
selective and leaked "predictions" that enter the public domain
in order to sell a pollicy.
I was reminded of what we learned about "secret government
estimates" in the wake of the Pentagon Papers -- Private
parties making use of only the information available to the public
had been as good and comprehensive and valid in commentary
as what the military could conclude with their extra, secret data.
I haven't seen any of his predictions, etc., but I thought that he
was probably offering a sensible voice on how to look at issues.
--
Rich Ulrich
Gary
> On Wed, 30 Sep 2009 11:10:55 -0700 (PDT), Gaetan
>
I haven't seen the show you refer to and didn't understand the
Original Posters question. But reading your answer I wondered whether
he was using a form of the Delphi method?
See
http://en.wikipedia.org/wiki/Delphi_method
Lance
Rich Ulrich
wrote:
He happened to mention Delphi by name, as something that
he is *not* doing. That is, he says, his projections are not
taken as any consensus of the experts.
It seems to me that (so far as I remember) he could have
left room for a sort of Delphi assessment of "who is important"
Now I've done a quick search, and I see that he has been around
and somewhat controversial - at least, for making overly-strong
claims - for a couple of decades. Here is a skeptical site with some
comments and references.
http://decision-making.moshe-online.com/criticism_of_bueno_de_mesquita.html
--
Rich Ulrich
Les Cargill
Kinda sounds like it, doens't it? I am very skeptical of Delphi. It's
another Hap Arnold/RAND/Macnamara device...
--
Les Cargill