From: hal@finney.org
Date: Wed Jan 19 2000 - 14:52:21 MST
Robin Hanson, <rhanson@gmu.edu>, writes:
> 1) Have courts enforce a rule that policy makers must always (except for
> specific exceptions) prefer a market estimate over other sources when
> estimating matters of fact.
This seems the most reasonable approach to me, although as I wrote I don't
see why it has to be enforced by courts. Ultimately, if IF proves its
value the system should be self enforcing.
> 2) Have the rule be that decision D must be enacted whenever a market says
> that there is more than an X (=75%?) chance that a random jury, chosen
> twenty years later and paid twice the average wage to study the question
> for a full year, would agree that D was better than the decision that would
> have resulted had this market rule not been followed. Now I can see some
> problems with this, but it relies a lot less on courts to work.
I suspect it would be hard to find this much certainty on most
controversial issues. So this rule would probably not be used much.
> 3) Choose some official national objective function, something like GDP,
> only more carefully measured and trying to include more contributions like
> the value of leisure, etc. Have some independent and well monitored agency
> continually update this estimate of the state of the nation. Then have the
> rule be that decision D must be enacted whenever a market says that among
> all the decisions markets are considering (including letting policy makers
> do what they want), D gives the highest conditional estimate for this
> national objective. I think this is the most promising approach.
Choosing that objective function would be a massive political battle.
Chances are people would want to change it from year to year, too, as
their concerns change. Education is the big issue one year; the next
it's crime, the next it's taxes, then it's growth. People wouldn't want
to turn their political decisions over to a formula that they couldn't
change as needed.
But aside from these issues, the more important first step is to
have some way for IF to be tested. I don't think the online game at
www.ideosphere.com has had very meaningful results (only a few dozen
players are really active at any one time), and my sense from playing
the game is that the predictions haven't been noticeably more accurate
than those from other sources.
If there is to be significant support for using IF in the political
process then it must be given an opportunity to prove its worth. I don't
see that happening any time soon.
Hal
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:26:19 MST