Jesse Mazer writes:
> It's a bit hard for me to come up with a satisfactory answer to this
> problem, because I don't start from the assumption of a physical universe at
> all--like Bruno, I'm trying to start from a measure on observer-moments and
> hope that somehow the appearance of a physical universe can be recovered
> from the subjective probabilities experienced by observers...
I have a similar perspective. However I think it will turn out that the
simplest mathematical description of an observer-moment will involve a Big
Bang. That is, describe a universe, describe natural laws, and let the
OM evolve. This is the foundation for saying that the universe is real.
> But my speculation would be this: multiple copies only increase first-person
> measure to the extent that, from a third-person perspective, they are likely
> to create more *distinct* observers with a memory of the split going one way
> than of the split going the other. So if we look at some very large region
> where there are 90 copies of our local region of the universe where the
> quantum coinflip went one way and only 10 copy of our local region where the
> coinflip went the other way, then even if those 90 copies run in lockstep
> for a while, chances are good that eventually random events in each region
> will cause them to diverge. The end result would be that, after sufficient
> time, in this very large region you'll have 100 distinct regions which all
> share the same history up until that coinflip, with 90 having records of the
> flip going one way and 10 having records of it going the other.
I meant to imply that this kind of differentiation could not occur in
my thought experiment, because the subject involved was a simulation in
a computer. He is not interacting with the environment, he is running
a deterministic program. So I can give him a good or bad experience
and he will not split.
Do you think you would say, in that case, that when I flip my 90/10
quantum coin, there is no reason to give him the good experience with 90%
probability and the bad experience with 10% probability? It would be just
as good to reverse the odds?
Remember, I'm not running copies; I'm just running one program, and
flipping a biased coin to decide what to do.
> And given my
> views on splits having an "anticipatory" quality (so if you're split into
> two and then one copy is scheduled to be split 999 times later, that will be
> reflected in your subjective probabilities in the initial split), I
> naturally tend to think that this means you'd have a 9 times greater
> probability of experiencing the coinflip outcome that will later lead to 90
> divergent copies, even if the 90 copies run in lockstep initially.
This is an interesting idea; one thing that occurs to me though is the
problem of amnesia. (And let's face it, 99% of what we experience,
we don't remember.) If you're split into two and experience something
in the one copy that is forgotten before it gets split 999 times, does
that to-be-forgotten experience still gain the benefit of the 999-fold
In other words, is it by virtue of later memory of the event that you
expect the subjective probability to be magnified, or is it a more
> But like I said, it's difficult to make any definite claims about why we
> experience quantum probabilities the way we do, especially since no one has
> come up with a way to derive a simple frequentist notion of probability from
> the universal wavefunction in the MWI.
That is indeed the case. I hope to write an essay soon about what I
think such a solution would look like (not to offer a solution, merely
to try to clarify what exactly I think it needs to explain).