Finally, some approaches other than PM fetishism (fw)

[Bala Pillai]

---------- Forwarded message ----------
From: David Brin <david...@gmail.com>
Date: Wed, Aug 6, 2008 at 4:29 AM
Subject: Finally, some approaches other than PM fetishism
To: Prediction Markets <Predictio...@googlegroups.com>



As resident crank-skeptic toward Prediction Markets, I retain a belief
that they are far too much like "Delphi," suffering from some of the
same faults, like tendentiousness, herd mentality, and far too little
attention being paid to aspects that would make a prediction-tracking
system actually useful to society.

Particularly tracking those who make grand forecasts for polemical or
persuasive reasons, and who do NOT want to participate in any kind of
betting or tracking.  These politicians and CEOs and newsletter
publishers and so on never participate in PMs, yet theirs are the
blithely persuasive auguries that sway policy, investment and society.

Okay, well. I would like your expert opinion of the following. At
last, a stab in the direction of a predictions registry:

<i>"Think you've got the gift of foresight?   The Washington Post
<http://www.washingtonpost.com/>  has partnered with Predictify
<http://www.predictify.com/> , an online polling service, to create a
"Prediction Center" that allows readers to vote on possible outcomes
for selected stories. Users will  be able to leave their predictions
and discuss their beliefs on an  integrated comment thread.
Predictify, which launched in 2007, goes beyond basic polling systems
by integrating discussion features and monitoring a users' accuracy
score across the entire service. While there isn't currently a way to
weight  one question more than another, the site's algorithm does take
into account  the type of question and the accuracy rate of
participants. To offer an incentive for users to take part in the
polls, the site has  also implemented a premium program that allows
companies to sponsor a poll  and reward the most accurate participants
with cash. In return, these  sponsors are entitled to the demographics
data that the service asks for with  each vote. "</i>

Note, this lacks most of the added features that could turn something
like Predictify into a truly useful tool for accomplishing what
society really needs:

- systematic ways to appraise predictive success/failure.

- ways to overcome natural feigning, backdating and retro-disavowal.

- sufficient attractiveness to draw in a large critical mass of
participants.

- a widely-accepted way to "out" those who claim predictive acumen,
but refuse appraisal or accountability.

- discovery of "3-sigma" forecasters, so that scrutiny can be applied
to their methods.

- rewarding "2-sigma" people with credibility and greater access to
those in power, thus making the powerful "right more often."

Still, Predictify appears to take some baby steps toward a few of
these desiderata -- baby steps that could be so much more.  In
contrast, for all of the hype that has recently been given to
"prediction markets," they in fact make almost no efforts toward
achieving these goals.  Ideed, their entire drive is in other
directions.

 With cordial regards,

David Brin
http://www.davidbrin.com