Robin Hanson replied
>>.. speculative market reactions to downsizing ...Subsequent analysis,
>>.. found no significant relationship .... Bre-X ... buying into a
swindle ...
> I'll accept these as specific examples of errors in market
> estimates. But the question is whether some other source has fewer
> errors on average.
You know the joke about predicting market crashes being like a stopped
clock -- it's right twice a day. One might say the converse for
markets. They may *on average* perform the aggregation of (admittedly
constrained) information relatively efficiently. But when they foul up,
they really foul up. I wouldn't necessarily attribute the problem with
markets to markets but to the phenomenon that where there's a system,
someone will find a way to "work" the system. Markets are not immune to
manipulation and may offer an unusual amount of motivation and
opportunity for doing so, not to mention ideological justification.
> Let's say I want a probability estimate right now of the chance Bush
> will win. What specific library or conversation should I go to get get a
> better estimate than I could find at Iowa Electronic Markets?
I would have to say that the Iowa "Markets" could more resemble a highly
structured conversation than an actual market. Unlike stock markets, the
buying and selling activity doesn't feed back into the outcome of the
actual election (for now at least). Also I doubt there's enough money
involved to cause some one to try to fix the election so they can clean up
on the Iowa Market.
Tom Walker
Sandwichman and Deconsultant
Bowen Island
(604) 947-2213
