(urth) 5HC : Skinner, Turing (fwd)

Dan'l Danehy-Oakes danldo at gmail.com
Tue Feb 8 14:33:06 PST 2005


> And I'd only be rational in ignoring if I was a good logical postivist
> (darnit, confirmation *can* too be support for a theory!  We don't have
> to rely on just disproof.)

Mmmm. Yes and no. Support, yes. Not proof ... except ... Well, 
there are real-world problems with limited domains, and in those
domains, you can sometimes demonstrate that only a certain
number of possible _theories_ exist. But most such situations
concern theories-about-fact rather than theories-about-general-
principles. 

(Ex: Three people are locked in a room. One is stabbed
to death - several wounds. There are only seven possible 
answers to the question "whodunit?") 

But for more general (scientific) problems ... it seems to me that
disconfirmation is the best we can hope for, because there are
always more theories and, while we can generally rank them
in loose probability-clusters,

(Ex: How does gravitation work?
High probability: gravitation is mediated by a particle exchange.
Lower probability: exchange of massive, charged particles.
Still lower probability: exchange of protons.
Extremely low probability: exchange of gumballs.
Vanishingly small probability: massive particles yearn for
    each other.)

and in fact we _have_ to rank them this way, because otherwise
there's no way of deciding which theories are most worth testing.

(Ex: We will probably never test the gumball theory, nor the
theory I've just invented, that gravitation is mediated by 
tiny alien elephants flying around in the noosphere, propagating 
morphogenic resonance and administering random rectal
probes to isolated humans.)

But, alas, now and then theories that look like they should be
ranked very low turn out, with new experimental data, to be
much more likely true than the ones in the former high 
probability grouping. (Ex: the heliocentric theory. Evolution.
Quantum decoherence - well, no, anyone with any common
sense recognized the inherent rightness of that one 
immediately...)

> Consider that every situation can be broken down to a binary choice 

I suspect that this is true only of situations where you are faced
with countably finite numbers of possible choices...

Regarding your "bi-lemma": It only works if you're reasonably
confident of your ability to choose the option that best matches
your desires. Some of us (me) often wind up like the famous
donkey that starved to death because he was equidistant from 
two equally tasty-looking piles of hay.

This example was originally cooked up to prove that animals 
_aren't_ automata, because this would never  happen to any 
_real_ donkey. Which however suggests to me that the real 
donkey is indeed the automaton, which evolution has 
thoughtfully provided with a nonalgorithmic solution to such
situations - "pick one at random." Whereas we, nonautomata,
can dither forever.

--Dan'l

-- 
"We're going to sit on Scorsese's head"
     -- The Goodfeathers



More information about the Urth mailing list