X-Message-Number: 19809
Date: Thu, 15 Aug 2002 11:04:11 EDT
Subject: more on probability

Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit

Thomas Donaldson insists that probability assessments are impossible or 
useless in dealing with unprecedented situations, such as revival after 
cryostasis. This whole discussion is of very marginal relevance, but I'll say 
a little more anyway.

I was amused recently to see a 1996 book, Data Analysis, by Oxford's D.S. 
Sivia. Apparently the "ideological" split, regarding the foundations of 
probability theory, still exists that I noted--and resolved to my own 
satisfaction--about 50 years ago. 

On the one side are "traditional" or "establishment" statisticians, using 
mainly Fisher, Neyman-Pearson, etc., and headed in theory by Richard von 
Mises. To these people, a "probability" is a relative frequency in a sequence 
of experiments or observations that is infinite, or at least very large.

On the other side is the notion of probability as a degree of rational 
belief, and which can include unique events--past, present, or future--as 
well as unknown states of nature. Related to this is the Bayes/Laplace 
formula for *a posteriori* probabilities, which depend in part on *a priori* 
probabilities which the frequentists would say are unknown or subjective.

I have demonstrated--and it's on our web site--that they are both wrong, in 
part. A correct probability (not "the" probability) of an event is its 
relative frequency of occurrence in an actual, known, historical sequence, 
which is another way of saying that it depends on the information available 
to the observer. The sequence is necessarily finite and the resulting number 
necessarily approximate. 

The point is that one can ALWAYS find an appropriate sequence of known 
experimental results, by sufficiently broadening the criteria, and therefore 
one can calculate an OBJECTIVE probability for ANY event whatever--even 
though, in the worst cases, the vagueness or uncertainty makes the exercise 
nearly useless.

An example. Dr. Bob-a-Loo, the shaman, claims he can make it rain in the 
desert by dancing and chanting. So he dances and chants, and it rains. What 
are the odds that dancing and chanting can make it rain?

The "establishment" types are embarrassed. By their usual methods, since the 
outcome was very unlikely on the basis of chance, there should now be high 
confidence in the shaman's power. Of course, the statisticians will try to 
squirm out of it by demanding much longer than usual odds against chance. But 
the Bayesians will point out that this is EQUIVALENT to acknowledging a very 
small *a priori* probability for the shaman's claim--even though the 
statisticians and the von Mises coterie deny the validity of any such thing 
as *a priori* probability.

Where does my interpretation come in? We don't depend on guesswork or on 
dogma. We find a suitable, historical sequence of experiments into which the 
present instance can fit. This could be any of many. An obvious choice might 
be claims of "paranormal" power. How many such claims have been made (a great 
many), and how many validated (none). Hence the *a priori probability* was 
extremely close to zero, and we are not even interested in the outcome. Of 
course, if the shaman could do it many times in succession, on demand, that 
would change the picture. 

Into what sequence might the cryonics question fit? One of them is on our web 
site--the sequence of technological goals and the results. Look at all the 
historical goals or projects that might be considered reasonably similar, by 
sufficiently broad criteria, and the record of successes, continuing efforts, 
failures to date, and acknowledgements of failure. Try it--you'll like it.

Robert Ettinger
Cryonics Institute
Immortalist Society


 Content-Type: text/html; charset="US-ASCII"


Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=19809