X-Message-Number: 8647
Date: Tue, 30 Sep 1997 23:16:20 -0700
From: Peter Merel <>
Subject: The Tinkerbell Test

John Pietrzak writes,

>In other words, it is only after you have chosen
>a well-defined recursively enumerable algorithm that you can ask if it
>will halt or not.  

I don't think that's a showstopper - as mentioned before, the burden of
proof is on those who would like to show that brains, or whatever it is
that brains do, cannot be emulated by a computer. Until that burden is
hefted, we're free to consider brains as Turing-equivalent. Or do you mean
something else by "well-defined"?

>there may in fact be no absolute criterion for anything anywhere.  

That's taking John Clarke a little far I think :-) Certainly, once you 
define a context, you can define absolute criteria within it to your heart's
content. Without this trick mathematics would be no fun.

>I'd like to think [...] that intelligence actually exists.

Okay, let's suppose it does. Now, you remember my relativist definition? 
So long as you accept it, if an absolute criterion for intelligence 
exists, then there must be some class of games, GI, such that 
every unintelligent processes loses games in GI more often than any 
intelligent process. 

I think that characterizing GI is the problem you're trying to solve, but
there's plainly a simpler problem that you need to solve first: can you
think of any game that is a member of GI? If not, you have no evidence
to suggest that GI is not empty - that intelligence does not "actually 
exist". Without evidence, you may as well believe in fairies as intelligence.

Peter Merel.

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=8647