----- Original Message -----
From: Hal Finney <[EMAIL PROTECTED]>
Sent: 11 January 2004 17:57
Subject: Re: Peculiarities of our universe
> The question of why we live in a sparsely populated universe, then,
> comes down to a comparison between the measure of a typical universe
> with many observers versus the measure of a typical universe with few.
> The former universes would get a large bonus factor for their many
> observers, while universes like ours don't have that.  So for our
> observations to be consistent with the AUH, it must be that universes
> like ours have much larger intrinsic measure than universes with many
> observers.  And since, as far as we can tell, our universe is not
> just sparsely populated, but extremely so, the measure differential
> in these two classes of universes must be extremely large.  That is
> (turning to the Schmidhuber interpretation) it must be much simpler
> to write a program that just barely allows for the possibility of life
> than to write one which makes it easy.  This is a prediction of the AUH,
> and evidence against it would be evidence against the AUH.
> On the face of it, this prediction doesn't seem too plausible to me.
> Of course, no one has ever written a program which evolves intelligent
> life, so we don't really know.  But our initial explorations towards
> artificial life seem to indicate that it's not particularly difficult
> to achieve model universes just swarming with tiny and unintelligent
> replicators.

My own take on this problem is that if under AUH we take all possible
programs, or alternatively all possible formal systems, then because there
are so many ways that extra complexity could be reflected in the existence
of invisible entities within a universe, entities in other physical
universes (specified by the same bit/axiom based string), or just nonsense
bits/axioms, then *if* we are in the simplest type of (visible) universe
consistent with the existence of SAS's, strings with the *particular* net
extra complexity that is needed to specify a SAS-abundant universe should be
swamped in number by those of the same amount of total complexity but
including the above mentioned invisible/disjoint/nonsense
entities/elements, all of which end-products will be visibly
indistinguishable from our own universe. In other words, even though the
additional size of the program (or formal system elements) necessary to
specify a SAS-abundant universe (as compared to a SAS-sparse one) might be
quite small, the increase in measure provided by the vast variety of
invisible changes/additions logically available with the same additional
size will, it is plausible to suppose, more than compensate for the
increased measure due to SAS abundance.

Similar arguments can apply to longer bit/axiom strings, and further
support can come from the intuitive idea that a universe of minimal
complexity compatible with SAS's is unlikely to be teeming with them.
Related details in the first URL mentioned below, section 2h.

Alastair Malcolm
For participation in discussions on the fundamental problem of existence:

Reply via email to