On 2003-07-28, Robin Hanson uttered:

>FYI, our DARPA project (www.policyanalysismarket.com) has just been
>denounced by two senators:

It's still a nice plan. Much like Brunner's delphi pools in Shockwave
Rider. (BTW, if anybody ever tried to patent artificial markets like
these, it'd be highly interesting to know whether sufficiently exacting
scifi qualifies as prior art.)

What I'm wondering, though, is asymmetric information. General contingent
markets are basically complicated markets in risk, and the claims traded a
form of insurance based on risk sharing. So the same problems apply to
them that do to insurance. If we're betting on diffuse, difficult to
influence risks, there's no problem. But if we bet on something individual
market participants can change, we have to consider things like moral
hazard and adverse selection. I mean, because of the efficient market
hypothesis, the ideal contingent market only allows profits to someone
bringing in new information. That's why betting on whether someone will be
assassinated will give an unreasonable edge in information to an assassin,
and will actually spawn assassinations. (Assassinations are probably
cheaper than defending against them if true anonymity is present.)
That's also why the market could turn into a twisted incarnation of
Assassination Politics (http://jya.com/ap.htm).

Furthermore -- unlike the original AP proposal -- these are futures
markets where you can speculate and win without pulling the trigger.
Rational bubbles can form, and those will in case turn the bet on an
assassination into a self-fulfilling prophecy (early investors will have
an incentive to rig the wheel, so to speak). When this happens, the ill
effects of moral hazard will be amplified, and we can't rely on the
resulting assassinations being reasonable even to the degree we can with
garden variety AP.

To someone with a penchant for conspiracy theories, then, all this would
probably count as one of DARPA's tacit aims.
Sampo Syreeni, aka decoy - mailto:[EMAIL PROTECTED], tel:+358-50-5756111
student/math+cs/helsinki university, http://www.iki.fi/~decoy/front
openpgp: 050985C2/025E D175 ABE5 027C 9494 EEB0 E090 8BA9 0509 85C2

Reply via email to