Study their behaviors. Observe their territorial boundaries. Leave their habitat as you found it. Report any signs of intelligence.

Loading Table of Contents...
 
 
 
 
 
 

Thursday, December 18, 2008

Bostrom's Simulation Argument

The only time I've touched on this topic is in the exchange below with Bostrom back in 2002.  I soon learned that the idea I was asking about is called modal realism, which encompasses all of the most interesting philosophical implications of the Simulation Argument.

From more recent reading on quantum physics, I no longer have such a firm intuition that a non-zero Planck Constant makes the universe easier to simulate, especially in light of the hacks and optimizations that Bostrom describes. However, I'm still fond my insight -- perhaps true, perhaps even original -- that classical physics should allow in principle for infinite information density.

As a technologist, I tend to think there isn't an interesting possibility of our sort of physics being able to support a simulation of a universe of our sort of physics. The hacks and optimizations that Bostrom talks about -- monitoring a simulation to see what its inhabitants "notice" -- can be recognized as nigh-impossible by anyone who's tried to debug their own software (let alone the simulated mental operations of minds that nobody programmed).

So I think that element (2) -- simulations won't happen -- of Bostrom's disjunct is the most probable, but as a modal realist I already feel sort of like how I'd feel if I believed we were in a simulation.

-----Original Message-----
From: Brian Holtz [mailto:brian@holtz.org]
Sent: Wednesday, March 13, 2002 2:36 PM
To: nick@nickbostrom.com
Subject: anthropic reasoning re: "why is there something rather than nothing?"

Hi. Two quick questions from someone who's enjoyed for several years your work on anthropic and transhumanist topics:
1. Has anyone ever applied anthropic reasoning to the perennial philosophical question of "why is there something rather than nothing?"?
2. Has anyone ever noticed that Planck's Constant being non-zero (i.e. that our universe is quantum rather than classical) could be construed as evidence that our universe is a simulation?
---------------------------------------------------------------------------------
Some details:
On the latter question, the point would of course be that simulating a classical universe to arbitrary precision would be much more computationally expensive (indeed, perhaps impossible) compared to simulating a quantized universe.
On the former question, I notice of course that it is closely related to anthropic cosmological reasoning like that discussed in Ch. 2 of your upcoming book. I'm just wondering if anyone has ever applied applied anthropic reasoning to the logic-motivated "multiverse" of logically-possible universes, as opposed to the quantum-theory-motivated multiverse of physically parallel "universes".
Here is a relevant excerpt from a book (see http://humanknowledge.net) I'm writing:
A possibly meaningful (but unparsimonious) answer to the Ultimate Why is that the universe exists (more precisely, is perceived to exist) roughly because it is possible. The reasoning would be as follows. Absolute impossibility -- the state of affairs in which nothing is possible -- is itself not possible, because if nothing truly were possible, then absolute impossibility would not be possible, implying that at least something must be possible. But if at least one thing is possible, then it seems the universe we perceive should be no less possible than anything else. Now, assuming that physicalism is right and that qualia and consciousness are epiphenomena, then the phenomenology of a mind and its perfect simulation are identical. So whether the universe we perceive existed or not, it as a merely possible universe would be perceived by its merely possible inhabitants no differently than our actual universe is perceived by its actual inhabitants. By analogy, the thoughts and perceptions of a particular artificial intelligence in a simulated universe would be the same across identical "runs" of the simulation, regardless of whether we bothered to initiate such a "run" once, twice -- or never.

An earlier exploration of this idea is this:
Consider gliders in Conway's game of Life.  Even if nobody ever wrote
down the rules of Life, gliders would still be a logical consequent of
certain possible configurations of the logically possible game of
Life. It has been proven that Life is rich enough to instantiate a
Turing machine, which are of course known to be able to compute
anything computable. So if mind is computable, consider a
configuration of Life that instantiates a Turing machine that
instantiates some mind.

Consider the particular Life configuration in which that mind
eventually comes to ask itself "why is there something instead of
nothing?".  Even if in our universe no such Life configuration is ever
instantiated, that particular configuration would still be logically
possible, and the asking of the Big Why would still be a virtual event
in the logically possible universe of that Life configuration.  The
epiphenomenal quality of that event for that logically possible mind
would surely be the same, regardless of whether our universe ever
actually ran that Life configuration. So the answer to that mind's Big
Why would be: because your existence is logically possible.

So pop up a level, and consider that you are that mind, and that your
universe too is just a (highly complex) logically possible state
machine.  In that case, the answer to your Big Why would be the same.

Note that, while the Life thought experiment depends on mind being
computable, the logically possible universe (LPU) thought experiment
only assumes that our universe could be considered as a logically
possible sequence of (not necessarily finitely describable)
universe-states.  The LPU hypothesis also depends on the thesis that
physicalism is right and that qualia and consciousness are
epiphenomena. The LPU hypothesis is of course unparsimonious (sort of
like the many-worlds interpretation of quantum theory), but parsimony
is perhaps inconsistent with *any* answer to the Big Why.  The LPU
hypothesis is incompatible with strong free will (which itself may be
incoherent), but is compatible with weak free will (perhaps only if we
assume there are rules governing the transitions among
universe-states).

The idea that the world might be a dream is of course not new.  But I
don't recall ever hearing that the world might be just a logically
possible dream for which no dreamer exists.


-----Original Message-----
From: Nick Bostrom [mailto:nick@nickbostrom.com]
Sent: Wednesday, March 20, 2002 1:26 PM
To: Brian Holtz
Subject: Re: anthropic reasoning re: "why is there something rather than nothing?"

Hi Brian,

Hi. Two quick questions from someone who's enjoyed for several years your work on anthropic and transhumanist topics:

1. Has anyone ever applied anthropic reasoning to the perennial philosophical question of "why is there something rather than nothing?"?

Derek Parfit touched upon this topic in some lectures he gave in London a few years ago. There is also a mailing list, the everything-list, where this topic has been discussed extensively.

2. Has anyone ever noticed that Planck's Constant being non-zero (i.e. that our universe is quantum rather than classical) could be construed as evidence that our universe is a simulation?

Yes (I think Hans Moravec might have been first, but I'm not sure). My view (see the Simulation Argument paper) is that it is not good evidence for that because the apparent ultimate physics of our universe could easily be an illusion if we are living in a simulation. That is, our simulators could create the appearance that our physics is quantum or classical without actually having to go to the trouble of simulating our world down to such find detail.

All best wishes,
Nick