From: Hal Finney (hal@finney.org)
Date: Sun Jun 01 2003 - 14:51:57 MDT
Harvey writes:
> I've never understood this. Doesn't the anthropic principle also lead to
> the Doomsday Argument <http://www.anthropic-principle.com/> that argues
> humans DO go extinct in geologically short time? It seems to me that the
> Anthropic Principle argues against the first assumption in the Simulation
> Argument.
This is a good point, but there are a couple of counters. First, among
the vast amount of literature on the Doomsday Argument (DA) there do
exist various counter-arguments, which I must confess go somewhat over
my head. Nick talks about the "Self-Indication Axiom" which can work
against the DA. So you don't necessarily have to accept the DA even
though you might believe the Simulation Argument (SA).
Second, even if you accept the DA, we might still squeeze in quite a bit
of simulating before the end. The DA basically says that we're "typical".
Well, as I understand it, approximately 10% of all humans are alive today.
Put another way, about 10 times as many people have died as are alive.
If we are in the middle, then we'd expect history to last until another
~60 billion have died. At current population and longevity levels,
that would be perhaps 500-1000 years in the future. That is a very
rough prediction of our future longevity based on the DA, to within an
order of magnitude or so. That's enough time that our technology will
probably advance to the point where we have super-advanced computing
and can perform enough simulations to give the SA some force.
And third, the SA is actually an independent counter-argument to the DA.
Why are we here? The DA says it's because we're roughly in the middle
of our species' lifetime. But it could be that we're here because this
is the era that everyone likes to simulate the best - the Transition,
when humans went from animal to god. It's plausible that this would
be an era which would be heavily investigated, and so it would be
over-represented in terms of probability of observer-moments. In that
case our presence here and now could be consistent even with an enormously
long human future, which is otherwise unlikely given the DA.
> The statistical counting of the universes to see which is most likely seems
> weak to me. There would be more dreams than realities and simulations. If
> this process made sense, shouldn't we conclude that we are in a dream
> instead of a simulation or a real universe?
This is a commonly discussed question on the everything-list.
The consensus resolution is that simpler universes must be inherently
more probable. To provide an overly concise explanation, this can be
justified by imagining that universes correspond to computer programs that
describe the "laws of physics" and initial conditions for that universe.
Finite length programs can be thought of as prefixes of infinite length
strings. The measure of a program will be greater if it is shorter,
because then it is a prefix of proportionately more infinite strings.
Therefore shorter programs are of higher probability, and therefore
more lawful universes are inherently more probable. Dreams and such
are chaotic and lack the internal consistency which the real world has,
hence such worlds would be less probable.
> I also question why different universes should be given equal weight in
> statistical probability that we would find ourselves therein. Since
> realities exist for billions of years before the first simulations appear,
> wouldn't that give realities a magnitude of more probability in these
> calculations?
That makes sense, although I think you might only want to count the
realities where observers like us exist. Billions of years of existence
of a sterile universe would play no role in determining how likely an
observer is to appear in one place or another. But if a real universe
held intelligent life for billions of years, that would count for a lot.
You might rate a simulation or a reality in terms of person-years of
conscious existence. A "real" universe like this one, which holds many
conscious observers for a long time, would have to be balanced by either
a sim which held as many people for as long, or by multiple sims which
had fewer people for a shorter period.
Hal
This archive was generated by hypermail 2.1.5 : Sun Jun 01 2003 - 15:06:08 MDT