X-Message-Number: 7430
From: 
Date: Mon, 6 Jan 1997 11:00:49 -0500
Subject: probability

Although Anatole Dolinoff's English is much better than my French, he says
his understanding is not good enough to cope with my booklet on probability
and cryonics. And he clearly did not understand my recent comment that the
chance or probability of revival of today's cryonics patients is certainly
NOT zero. For a few readers, an attempted capsulation of a couple of points
may be useful.

The basic point is that a "probability" always reflects the state of
knowledge of the observer, and therefore varies from time to time and from
one individual to another. There is no such thing as an "objective"
probability (leaving out of account, for the moment, such things as gas
statistics, quantum statistics, etc., which are not of immediate concern).

Classic "frequency theory" proponents, e.g. v. Mises, did not understand
this, and regarded any true "probability" as the fraction of occasions in
which the event occurs, out of all the times it might occur, in an infinite
series of experiments. For example, the probability of throwing a six with
one die is 1/6, since a die has 6 faces and, in a fair toss, each side has an
equal chance to land up. He emphasized that certain "events" are ruled out of
the theory, and these specifically included past events or "states of nature"
as well as such statements as "The Iliad and the Odyssey had the same
author."

Clearly, although he was a great mathematician, v. Mises was a naive
logician. For example, what about the "probability" of a six being up on a
die ALREADY TOSSED BUT NOT YET EXAMINED? This is a past event and a state of
nature (the six is either up or it isn't); but from the standpoint of the
OBSERVER who might make a bet, there is no difference whatever between a die
about to be tossed, or one already tossed but not yet inspected. The
probability is 1/6.

I'll leave Iliad and Odyssey for the reader, in light of the following
example that I have used to illustrate the nature of probability calculations
in many questions of real life.

A football game is scheduled between Michigan State U. and Wayne State U.
Bettor A reads the Associated Press sports writer polls; the AP picks MSU,
and their record over recent  years shows that their choices have won 65% of
the time. Therefore, for bettor A, the probability that WSU will win is 0.35,
and refers to the following experiment: Pick the team chosen by the AP poll;
in the long run, you will be right about 65%of the time.

Bettor B is a visiting Bantu who knows nothing at all about American
football, and doesn't read the papers or talk to sports fans, and can only
pick a team in some arbitrary way, maybe the team whose uniform colors he
likes best; or he could toss a coin. For such a bettor, the probability that
WSU will win is 1/2, and refers to the following experiment: If you know
nothing about the relative strengths of the combatants, pick one by an
arbitrary method, such as coin tossing; in the long run, you will be right
half the time.

Bettor C is the Coach of WSU. He rates his own team two touchdowns worse than
MSU; and his record for predicting outcomes in similar cases is 80% correct.
For bettor C, then, the probability of a victory for WSU is 0.2, and refers
to the following experiment: whenever the Coach rates his team a 2-touchdown
underdog, bet against it, and in the long run you will be  right about 80% of
the time.

Probabilities are ALWAYS approximate, never exact. (In something like coin
tossing, the probability is very close to 1/2; but coins are not perfect and
tosses are not unbiased.) (Again, for lack of space, I leave out of account
such things as quantum probabilities and statistical mechanics.)
Probabilities are generally different for different observers, reflecting
different states of knowledge; they also differ, for the same observer, from
time to time as the situation changes or information changes. It is ALWAYS
possible to calculate a probability (or more than one) for any event of
whatever sort, past or future--although often only with large uncertainty or
variance. 

There are three crucial points: 

First, in the current discussion, the probability of revival of today's
cryonics patients is NOT zero, because to say "the probability of revival is
zero" is the same as to say "We know for sure the patient cannot be revived."
This reflects the MEANING of probability. 

Second, the rational person will be guided by probability--EVEN WHEN THE
PROBABILITY IS A VERY ROUGH ONE. The RELEVANCE of a number is more important
than its accuracy. For example, one can point out that, to restore today's
cryonics patients, a large number of molecules might have to be moved or
repaired. Perhaps that number could be calculated fairly accurately, and it
might be very large. Or one could look at the sweep of history, at the
astounding advances in technology and the rising rate of advance, at the
conjectures of Richard Feynman, etc., and conclude that feats of comparable
subjective difficulty have often been accomplished; whereas extremely few, if
any, comprehensive and specific negative predictions have held up. (There is
absolutely no known law of nature forbidding revival; the negative
considerations are entirely of the practical variety, involving estimated
degree of difficulty, not matters of principle, in general.) 

The sweep-of-history argument is hard for some scientists to accept, because
their own work seldom involves it. But their everyday lives DO involve it,
and they should remember this. How do you know you can trust your wife? If
you can and do, it is not from explicit calculations, but from your overall
impression of accumulated experience, which might be rough and vague but
still extremely relevant and useful. You could still be wrong, and in some
situations you might want to think things over, but in general you are well
served by an educated intuition, which is an implicit evaluation of
probabilities.    

The third point to re-emphasize is that probabilities change to reflect new
conditions or new information. The probability of revival of our patients is
not a fixed number, but is subject to feedbacks from our own activities.
Advances in research and growth of the cryonics organizations can obviously
make important positive contributions.

Robert Ettinger
Cryonics Institute
Immortalist Society


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=7430