Free will (was: How hard a Singularity?)

From: James Rogers (
Date: Thu Jul 04 2002 - 18:22:16 MDT

On 7/4/02 2:44 PM, "Samantha Atkins" <> wrote:
> This looks tantamount to claiming that only an utterly unlimited
> being could have free will.

"Free will" is a relative term. Computational machinery can only
deterministically model other computational machines smaller (re: AIC) than
themselves. Therefore, it is impossible to deterministically self-model. I
define "free will" as not being able to deterministically predict the
outcome of your own computational processes. All finite state machinery has
this property, and not being able to perceive and perfectly evaluate your
own processes doesn't mean nothing else can either -- that would be the
logical equivalent of "I can't see you, so you can't see me". If you have a
different (but meaningful) definition of "free will" or assume non-FSM
computation, then maybe that might change the discussion.

Of course, this doesn't preclude using external machinery to
deterministically model yourself, though it puts a different spin on things
(and creates a feedback loop).

> Oh, wait, we are, afaik, capable of
> going beyond our biological limits. Either way, you lose.
> Actually, it is not a question of mathematics at all. If you
> can choose among alternatives, even if the alternatives are
> finite, you have free will. Espcially if you can modify your
> value and goal structures to some extent.

That is an embarrassingly superficial analysis. When you choose among
alternatives, how do you know that an SI with access to your computational
state would not be able to perfectly predict which alternative you would
pick? How do you know the modification of your values and goal structure
are not the result of a deterministic process?

> Ah, now any questioning of your bald assertions will be taken as
> "irrational backlash". Sigh.

Spare us the melodrama. Your insistence on reading things into what people
write is a reflection of your own issues, not everyone else's. I've
discussed this topic, rigorously, in other forums with individuals who are a
lot farther back on the curve than most of the people on this list (though
technically competent), and was relaying their reaction. The rigor of the
actual technical argument was questioned in other forums but held up under
scrutiny -- there is no magic/difficult mathematical concepts involved.

I didn't think I would have to go into it too far into detail, as there are
quite a few people with sufficient background in the relevant areas of
mathematics on this list that it wouldn't surprise them. I was only
pointing it out to refresh memories and keep things on track. I have
limited time and don't want to spend it elaborating on things that should be
obvious to someone with sufficient background and/or can be verified in
rigorous detail elsewhere.

How about actually contributing something useful? Something vaguely
approximating a rigorous refutation would make for a nice discussion.

-James Rogers

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT