Re: Bad Bayesian - no biscuit! (was A New Year's gift for Bayesians)

From: Eliezer Yudkowsky (
Date: Thu Jan 20 2005 - 01:15:13 MST

Ben Goertzel wrote:
> Well, when I was a math prof at UNLV, a number of my fellow math
> professors liked to gamble on a regular basis.
> These folks had high IQ's, decently long lists of pure-math
> publications, and generally a lot of common sense and rationality in
> everyday life situations.
> I was a bit perplexed at the time that these highly clever and
> reasonable mathematicians were at all interested in gambling, but that
> was before I grasped the deep perversity of human nature as fully as I
> do now ;-)

You're admiring the wrong people! Let them stop buying lottery tickets,
then admire them. It only defeats your own potential if you think that
intelligence has no better use than that.

>> I understand the thriving lottery industry as the present state of
>> affairs, and I ask "How can we train people not to buy lottery
>> tickets?"
> But if people buy lottery tickets because they find the emotional drama
> of it FUN, then teaching them not to buy lottery tickets means either
> a) teaching them not to want to have fun
> b) teaching them to find different things fun than they currently do

People pray because they want their child to not die of cancer. Teaching
them not to pray means (a) teaching them not to want their child to survive
(b) associating different outcomes with the effort of prayer. So too with
the lottery. Even if they claim to pursue the allure of the allure of
unimaginable wealth rather than the allure of unimaginable wealth per se,
both allures depend on a false belief - a subtle false belief, because
while their verbally reported probabilities may be correct, their feeling
of probability doesn't correspond to the verbal report. If people knew
more and thought faster, they wouldn't play the lottery.

If yon mathematicians really understood that winning the lottery was
impossible and that they were absolutely certain not to win - an absolute
certainty I don't bother to qualify with disclaimers, because it is so
enormously stronger than the propositions that people usually believe to be
absolutely certain - then they would feel no drama. I strongly suspect
that "I understand the probabilities, but I pursue the drama" is a post
facto excuse so that they can retain their self-respect as rationalists
without modifying a behavior they know to be irrational. If they
understood the probabilities, there would not be drama. Mathematicians
anticipate that to appear rational they must admit to their fellow
mathematicians that they have no significant probability of winning the
lottery. In the corners of their minds they anticipate they can get away
with claiming that they pursue the drama, so their self-deceiving mind
inserts that justification into their account of themselves. In today's
society their anticipation is correct, for few are strong enough in the
Way, or have enough explicit knowledge of the Way, to explicitly describe
the mistake.

> I.e. it involves changing their motivational structure rather than their
> reasoning methods, it seems to me.

Change the reasoning method, and the motivational structure will follow.

> Personally I have pretty much given up on teaching people not to have
> silly chimp-like motivational structures. It's a damn hard problem,
> harder than creating an AGI, in my opinion.

But explaining rationality is a hell of a lot easier than explaining AGI.
One of the reasons I now concentrate on the art of human rationality in my
writing, is that human rationality is a comprehensible reuse of AGI
knowledge, impressive in its own right. Where AGI itself is nearly
impossible to explain, one can see - in retrospect, at least - that one who
seeks to apprehend the hidden succession technique of the Way should be a
master of its ordinary application, or at least have explicit knowledge
thereof. If I am an AGI wannabe and I have made any sort of significant
progress, I ought to be able to solve problems in human rationality with my
eyes closed standing on my head in a cold shower with my left cerebral
hemisphere removed.

> I try to guide my kids to have motivational structures that I respect,
> but that's an easier problem than the more general case, since they have
> genetic material that's relatively amenable to my taste, and they get a
> lot of personal attention from me.... And it's still not such an easy
> problem, as they've all inherited my individuality and pigheaded
> stubbornness along with many of my other fine and not-so-fine traits ;-)

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:50 MDT