Thanks for the replies,

On Fri, Feb 29, 2008 at 4:44 PM, Ben Goertzel <[EMAIL PROTECTED]> wrote:

> I am not so sure that humans use uncomputable models in any useful sense,
> when doing calculus.  Rather, it seems that in practice we use
> computable subsets
> of an in-principle-uncomputable theory...
>
> Oddly enough, one can make statements *about* uncomputability and
> uncomputable entities, using only computable operations within a
> formal system...
>
>
What I'm thinking is something like this: the uncomputable 'ideal' calculus
serves as a motivation for the many computable subsets. Similarly, the AIXI
model is supposed to act as a motivating concept for many possible realistic
implementations. It is the ideal that the computable implementations
approximate. In that way, it "compresses" them; it is supposed to be the
pattern behind them.

(Of course, since at the moment I don't endorse Solomonoff Induction, it
doesn't act as such a motivating force for me, but it could if I did endorse
it.)

So I think what I'm saying is that uncomputable models are useful because we
can compute useful statements *about* them. But I wish I could give a more
solid argument!


On Fri, Feb 29, 2008 at 8:10 PM, Matt Mahoney <[EMAIL PROTECTED]> wrote:

>
> There is evidence that the universe is Turing computable, as opposed to
> being
> computable only by a more powerful machine or uncomputable.  In particular
>

 [...]

>   Likewise, all other conserved properties are
> quantized, such as electric charge, baryon number, etc.
>
> We also observe that Occam's Razor works in practice, which is predicted
> by
> AIXI if the universe and our brains are both Turing computable.
>
>
> -- Matt Mahoney, [EMAIL PROTECTED]
>
>
Accepting uncomputable models does not mean throwing out Occam's Razor, at
least not the way I intend it. We can have longer or shorter differential
equations. (It seems convenient to keep using calculus as the main example
of an uncomputable model.)

If we do discover a computable model of the universe, then the previous
models will be demoted to the position of approximations. But wouldn't they
still be *useful* approximations in some cases? In fact, Newtonian mechanics
will probably still be what's used for everyday-scale physics. (Again, I'm
trying to assert that the uncomputable can usefully abbreviate the
computable.)

-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription: 
http://www.listbox.com/member/?member_id=8660244&id_secret=95818715-a78a9b
Powered by Listbox: http://www.listbox.com

Reply via email to