X-Message-Number: 8251
Date: Wed, 28 May 1997 19:31:18 -0400 (EDT)
From: "Perry E. Metzger" <>
Subject: falsification

> From: 
> 
> I can't resist one quick remark on the alleged nonfalsifiability of the
> "invisible bunny" hypothesis.
> 
> 100% proof is not available--but it seldom if ever is for
> anything. Yet it is extremely easy to show that the assertion is
> false to a near certainty. We need merely think about the
> context--someone inventing an example of a nonfalsifiable hypothesis
> for the sake of argument.

I hate to say this, but the notion that a hypothesis should be assumed
false because it was brought up as an example is one of the most
idiotic abuses of logic I've seen here on this mailing list -- and
I've seen a number recently. By your reasoning, I should discount each
and every gedankenexperiment you have asked us to think about. Even
worse, by this reasoning, I have to assume that the truth of a
statement has something to do with the intent of the speaker -- one of
the worst abuses of the Ad Hominem fallacy I've seen.

> In light of our total experience, it is overwhelmingly more probable
> that the assertion arose in this way, or as a joke, rather than as a
> result of the actual existence of the bunny.

"Probability" is not the way to judge things like this, because they
have no "probability". Either the assertion is true, or it is
false. "Probability" is a statement of the percentage of times an
event is observed in multiple trials. You cannot reason about things
like this using probability theory because there are no multiple
trials -- the bunny is there, or it isn't, there are no multiple
observations.

I must say that I'm really incredibly sick of these "Clinton has an
80% chance of winning the election", "Joey has a 40% shot at getting
the job", etc., abuses.

The reason people have so much trouble reasoning clearly about odds
and probability is the constant abuse made of the language of
probability theory -- and the continuous promulgation of "fake"
probability theory like the "law of averages". The fact that so many
people on this list had trouble with the Monty Hall paradox shows the
dangers of falling into the trap of discussing "probability" loosely.

Beyond this, however, I think you've totally missed the point. A
non-falsifiable hypothesis has the flavor of stupidity about it
FOR A GOOD REASON. The "God will put you in the pit of hell if you
don't believe in him" hypothesis, the "You will be reincarnated
depending on how nice a guy you are" hypothesis, and the "There is an
undetectable bunny following Robert Ettinger" hypotheses all have the
same flavor about them. They might be something someone made up, or
they might be something someone read, but none of them have any
conceivable way of ever demonstrating them one way or another.

Similarly, I contend that you, Robert Ettinger, along with a large
number of other people like Penrose, Searle and others, have a
religious belief which may be summarized as "we are too important for
mere computers to be able to simulate us" -- similar, I will point
out, to the 19th century "What! I couldn't be the descendant of an
ape!" arguments against evolution. The reason that we find there to be
a strange non-falsifiable flavor to Searle's arguments is precisely
because they originate not in solid thinking about the problem, but in
the embarassment that these people and you feel in the idea that you
might be a finite deterministic process and not some sort of
miraculous special thing.

The whole discussion has the irrational feel of religion precisely
because it is an argument about the existance of the soul. You are
basically arguing for the existance of a soul, Mr. Ettinger -- a
strange, ineffable thing animating you that could not be built into a
machine. Being mostly a rationalist, you don't use the term, because
it would be embarassing to do so, but in the end, what you, Penrose,
Searle, and the rest of the "YOU CAN'T SIMULATE *ME*!!!" party are
arguing for is a soul.

Well, I don't believe in one, and even worse, I can't find a
non-falsifiable hypothesis hiding in there anywhere. Searle's
arguments are the worst of all -- becuase once you argue that you
could build a walking, talking android that none the less would not be
"conscious", there is no reason to believe that any sort of
falsifiable hypothesis of any sort is floating around at all. The
differences Searle argues for are entirely non-falsifiable. He is
right that I cannot prove by his criteria the consciousness of the
robot -- but neither can I prove Searle conscious by these invisible
unusable criteria.

At least Mr. Donaldson is being sufficiently reasonable that he is
willing to substitute the "recognises visible objects" test for the
Turing test, thus giving us a test that at least permits falsifiable
hypothesis building.

Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8251