Re: Infinite universe

From: Eliezer S. Yudkowsky (
Date: Tue Apr 29 2003 - 19:21:43 MDT

Psy-Kosh wrote:
> Well, I was talking level 4 "all mathematical/subjective experiances"
> exist, not level three "many worlds interpretation" but the same
> thing is here too.
> Let U be the grand universal wave function. ie, the actual (level 3)
> multiverse as a whole.
> P(a1|U) has a certain value. P(a2|U) has a certain other value. U
> simply is what it is. U can be viewed as the sum of a1 and a2 and b
> and so on and so on, and P(a2|b) may be less than P(a2|U), but that
> in no way changes U itself.

You're *within* U. *You yourself* are part of the play and flow of
probabilities. In the moment when you decide whether to help or hurt
someone, that decision is embodied in the flow of the probability
amplitudes, and the future correlates with your decision. Of course the
future is determined. It's determined by the probability amplitudes.
It's determined by you. It's determined by your decision. All these
statements are equivalent.

The timeless form of the decision rule is that given choices A vs. B, you
should consider the state of the multiverse given that your decision is A,
versus its state given that your decision is B, then pick whichever choice
corresponds to the more desirable state. Only correlation is necessary -
not "time" or "causality". Note that this resolves Newcomb's Paradox.

Making "meaningful choices" means that the computation embodying your
decision is what determines the future. To make meaningful choices, the
future must be determined by "mere" physics, and we ourselves must be
"mere" physics, because only in this way can *we* determine the future.
The future is determined; it's determined by the physical process that is
your choice, now. Knowing in the abstract that the future is determined
is not the same as knowing *what* the future is determined to be; the
future is determined by you, yourself, making a choice under uncertainty.
  If you don't yet know your choice, you can't possibly know the
deterministic future, because what the future is determined by, is your

Your choice at T+1 is determined by the physical process at time T that is
your mind, but at time T, that physical process doesn't yet internally
represent ("know") what its choice will be, so it doesn't internally
represent ("know") the outcome at time T+2 of any processes determined by
that choice. The outcome at time T+2 is determined by the computations of
a cognitive process at T which does not yet know its own decision at T+1.
  That's what we call "free will". Note the noncoincidental resemblance
to the halting problem.

Now you can get messed up if you know the abstract fact that the future is
determined by physics, and you think of "physics" as something outside
yourself, rather than a system that includes you as a part. You may get
the idea that the future is dependent on something other than your
decisions, or that your decisions are determined by something other than
your own cognition under uncertainty.

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:42 MDT