Re: The Simulation Argument again

From: Dan Fabulich (dfabulich@warpmail.net)
Date: Mon Jun 02 2003 - 21:39:17 MDT

  • Next message: Adrian Tymes: "Re: The good ship Extro 1"

    Harvey Newstrom wrote:

    > But what if we don't conclude that there are more simulations? What if
    > simulations are rare?

    Uh, you mean option (2)?

    Remember, Bostrom is simply arguing that "at least one of the following
    propositions is true: (1) the human species is very likely to go extinct
    before reaching a posthuman stage; (2) any posthuman civilization is
    extremely unlikely to run a significant number of simulations of their
    evolutionary history (or variations thereof); (3) we are almost certainly
    living in a computer simulation. It follows that the belief that there is
    a significant chance that we will one day become posthumans who run
    ancestor-simulations is false, unless we are currently living in a
    simulation. A number of other consequences of this result are also
    discussed."

    Your claim is in perfect agreement with Bostrom's argument. Bostrom says
    "1, 2 or 3," and you say "2." If I believe you, then I am of course
    licensed to conclude that at least one of 1, 2 or 3 are true.

    > What if perfect simulations of reality are just as boring as reality and
    > nobody makes them? What if perfect simulations are so wonderful that
    > only people from reality universes migrate into them and very few
    > simulated people get created? What if simulations are almost always
    > programmed to prevent discovery of the simulations, such that
    > discovery-possible simulations are extremely rare? By assuming that
    > simulations are rare, then our circular logic would have to conclude
    > that statistically speaking we are likely not to be in a simulation.
    >
    > Both of these conclusions are circular and lead people to believe what
    > they want to believe. There is no data either way to lead anybody into
    > either conclusion. Whatever people already believe, that is where their
    > circular logic will lead them. There is no statistical support in
    > either direction.

    It seems to me that all you're saying here is that we don't have enough
    evidence to pick between 2 and 3.

    I see this argument as a criticism of a crucial background assumption of
    the argument: that the key unknowns (f[P] and f[I]) can be estimated,
    within perhaps an order of magnitude (or even just a few orders of
    magnitude).

    For example, you ask: "what if simulations are just as boring as reality
    and nobody makes them?" and I answer: well, what are the odds of that?
    What's f[I]? .5? .25? .1? I think you'd have to be pretty confident to
    assume that the answer is even one in a hundred, say nothing of
    one-in-a-million or one-in-a-billion, and I don't think we have that kind
    of confidence.

    To the best of our information f[I] isn't much smaller than 0.001, if
    that. Are you going to try to argue that f[I] is on the order of 1/N[I]?

    > Furthermore, I believe their are other counter-examples to the simulation
    > universe:
    >
    > - Occam's Razor suggests that simpler, more direct explanations are more
    > likely to be correct. In the absence of data, there is no reason to
    > postulate unfalsifiable, hidden, mystical worlds that cannot be detected or
    > tested.

    Here, I think, you have missed the point of this argument. Many
    transhumanists still seem to believe that *we* will create simulated
    environments, in which it will be impossible for our simulated subjects to
    falsify the claim that they are in a simulation. This is a solid
    prediction that makes claims about what kinds of worlds there will be:
    namely, that there will be some simulated worlds that we'll make. Are you
    suggesting that we should reject THAT kind of optimistic transhumanism on
    the basis of Occam's Razor? I don't see why.

    > Why stop there? Why not postulate that every religion is true?

    Look, if you claimed that each of us would go on to become Gods of our own
    religions, under which it would be possible for people to become Gods of
    their own sub-religions, and so on, I think you'd be out of your gourd if
    you claimed that you yourself are *not* under a God. If you claimed as
    all this, saying "Sure, according to my theory, most people would be under
    some God, but I'M not", then arguments like Bostrom's would license us to
    conclude that your prediction/assertion is wrong.

    > That all fictional stories are real? That imaginary creatures are
    > everywhere when we're not looking? Etc.?

    Simulated people are, in their own way, very elaborate fictional/imaginary
    creatures. In an important sense, these sorts of things are exactly
    what's under discussion, with a non-trivial group of people claiming that
    one day we will have imaginary friends who will be substrate-independent,
    but that we are not ourselves anyone's imaginary friends.

    > - Theories are supposed to predict observed phenomenon. This theory is
    > untestable. It predicts nothing, explains nothing, and provides no
    > further information about how the universe works. Unless some theorized
    > difference between a real universe and a simulated universe can be
    > predicted, this theory will never be testable.

    Bostrom's simulation argument is a REFUTATION of a specific claim about
    the future: that *we* will go on to simulate people. Being as it is a
    refutation of a prediction, it is therefore a prediction.

    > - The very calculation of "more likely to be in a simulation" seems
    > suspect to me. To do such a simple calculation we would need real
    > measurements, and more importantly, some sort of mechanism for counting
    > universes and simulations. If many-worlds interpretation says there are
    > an infinite number of real universes, how can there be more simulations
    > than infinity?

    Everett's interpretation, often called the "Many-Worlds" Interpretation
    (MWI) doesn't claim that there are an *infinite* number of worlds,
    thankfully. So, assuming that we only have the huge-but-finite set of
    Everett worlds, counting them (and, more to the point, dividing over them,
    which tends to normalize out the alternative worlds) is in no way
    troublesome.

    Specifically, Bostrom's variable f[P] (fraction of all human-level
    technological civilizations that survive to reach a posthuman stage) gets
    no larger simply by virtue of the fact that there are more
    civilizations... we can still make sense of the FRACTION of such
    civilizations even when there are vastly more such civilizations than we
    might have expected. (Though, sure, perhaps not if there are an infinite
    number of such civilizations, though even there we could do a great deal
    with a simple measure theory.)

    > If every possible universe exists, wouldn't that include many (or most)
    > universes without simulations?

    Yup. So what? They get into the denumerator of f[P].

    > - As an absurd argument, why not count dreams? They are simulations of
    > the universe that seem real to us at the time. There are many more
    > dreams than universes (either simulated or natural). Wouldn't this
    > imply that we are in a dream instead of a simulation or reality?

    You've got no substrate-independence in dreams, (dream people being in no
    way on a noetic par with real people,) so Bostrom's argument wouldn't
    apply to dream-people.

    [Or, at least, not to OUR dream-people. You might say that a posthuman
    civilization is itself a highly sophisticated dream, in which case, yes,
    we could all be in some posthuman's dream.]

    Sleep well,
    Dan

          -unless you love someone-
        -nothing else makes any sense-
               e.e. cummings



    This archive was generated by hypermail 2.1.5 : Mon Jun 02 2003 - 21:52:14 MDT