I've been reading all of these posts on the "tautology" thread, and I've
been meaning for a long time to put my thoughts down, but it has seemed
like too large a task. Well, rather than keep putting it off, I guess
I'll just do an incomplete job of it.
Jacques' ASSA idea has considerable merit, IMO, but I still am quite
reluctant to adopt it. It has made more and more sense to me as I've
gone along, gleaning dingy pearls out of Jacques' ranting and raving.
I guess he thinks he's being cute, but it gets old (it got old a long
time ago.)
I still think, however, that he claims more for this concept than it
has in it. Let me try to justify myself -- I'll put my current
understanding of the ASSA in my own words. Let's see if there's a
match.
Very simply, the ASSA states that for each of us, the observer moment
we are experiencing right now is chosen at random from the entire set
of observer moments, based on their measure. Calculating the measure
is problematic, but it is related to the number of (physical?)
implementations of that state within all possible universes. Anyway,
the measure should, in principle, be calculable.
With "strict ASSA", it is meaningless to talk about what I'll observe
tomorrow, or at any future time. All I can do to make predictions is
to calculate probabilities such as "given that an observer sees the
time as t, and given that this observer has memories a, b, c, ...,
he can expect to observe M with probability P." Thus, with what I'm
calling "strict ASSA", I can't necessarily identify that observer with
myself, if the time t is not the current time.
This resonates with me -- it reminds me of a cliche aphorisms like "all
we really have is now", and "yesterday is but a memory, tomorrow is but
a dream, ...". One could certainly make the philosophical case that
it is only meaningful to talk about one instant of time.
On the other hand, this definitely does defy common sense. That's
not necessarily an argument against it, but I think it shifts the
burden of proof. I asked my wife (who is convenient for this sort of
thing), "when you wake up tomorrow, do you expect that it will really
be *you* who wakes up?" I think I phrased it better than that, but
you get the idea. She, of course, rolled her eyes, and said yes. I
remember reading about a court case a few years back where the
defendent tried to make the case, based on his philosophy, that the
person who committed the crime was really a different person than
the one going to court now. Though the body had evolved gradually,
he tried to argue that he was sufficiently different that he should
be considered legally a separate person. Needless to say, that
argument didn't work, and I remember everyone I talked to about that
article had the same incredulous reaction.
I have a strong sense that it is me who continues into the future.
This, of course, leads me to care about what happens tomorrow, etc.
Now, Jacques himself has admitted that the "strict ASSA" can be
supplemented with a definition of "you", in order that you *can*
make predictions about what you'll see tomorrow:
> one must first define "you". There are three reasonable
> possibilities in the ASSA:
> 1. One particular observer-moment. You have no past and no future.
> 2. A set of observer moments linked by computation. With this
> definition the problem is that "you" may be two (or more) people
> at the same time! The advantage with this definition is that one
> can predict effective probabilities of what "you" will see at other
> times similar to what you want to do with the RSSA. Thing is, if
> there is nonconservation of measure, the predictions start to differ
> from the RSSA about things like how old you should expect to be.
> Remember, testable prediction do NOT depend on definitions, so it is
> often better to use def. #1 to prevent such confusion.
> 3. A particular implementation of an extended computation. Similar to
> 2; allows death, when that implementation ends. I prefer this or #1.
I'd submit (and I'm going to guess that this is anticipated) that the
ASSA with definition 2 (which I'll denote ASSA-2) is the same as what
we've been calling the RSSA.
It's not ambiguous: if you define yourself to be a set of observer
moments linked by computation, and you make a prediction of what you'll
observe tomorrow, then you are, by virtue of that definition and premise,
excluding the possibility of death. I disagree that the "predictions
start to differ from the RSSA ...", because this sort of prediction from
ASSA-2 would be of the form:
"If I were to survive 1000 years, then I'd have a 50% chance of
being purple".
And this says nothing about the absolute (the A in ASSA) chance of
surviving 1000 years. But as Jacques vociferously points out, the
ASSA itself predicts that probability to be low.
The other main thing I wanted to comment on is Jacques insistence that
the ASSA does not involve any randomness:
> > > > [Jacques wrote:]
> > > > NO. In my view everything is deterministic. There is NO
> > > > randomness. Just a lot of observers with different observations.
> > > > *Effective* probability is proportional to the number/measure of those.
> > >
> > > [Russell:]
> > > What about what your observers actually observe? That is random.
> >
> > What we have here is a failure to communicate. I don't know how
> > to convey an idea you don't seem to understand. I hope I won't have to
> > keep trying. The idea is very very simple after all.
> > Suppose there are 1000 observers. For simplicity, assume each can
> > be labeled by a brain state which we can number from 1-1000, and that each
> > sees a different observation, which we can label by the observer number N.
> > Consider number 463. He sees an observation with various
> > characteristics, which we have labeled #463.. There is NOTHING random
> > about that, not in any way, shape or form, at all. Period. I can't
> > emphasize that enough.
It's interesting that Jacques picked a pseudo-random number (463) to
illustrate his point. I keep thinking that maybe I'm missing something
here, but I'm not. Of course what each of us observes, at this particular
moment (our little observer moment) is randomly SAMPLED from the set of
possibilities, hence the ASSA. Any way you slice it, from a subjective
point of view, there is randomness. Where there are probabilities, there
is randomness. In the above example, observer 463 would say, "my, 463
seems like a pretty random number, I don't think I could have predicted
that."
A couple more comments below.
"Jacques M. Mallah" wrote:
>
>
> > > The ASSA, together with our theories of physics, very obviously
> > > implies that the effective probability of observing a large age is quite
> > > low. This is effectively a prediction, and most observers will find it to
> > > be correct. Using simple Bayesian reasoning, effective predictions are
> > > just as good as any other predictions. In particular, you can plug the
> > > effective conditional probabilities into the Baysian formula to update
> > > priors about proposed models of reality. In this case to endorse ASSA and
> > > reject RSSA.
If I read you right, you're saying that you reject the RSSA because you're
not old. But I'm more inclined to reject the ASSA because I'm not dead.
>
> Re: Fabric of Reality
> >Incidently, it appears that it is
> >possible to derive Occam's Razor, or something like it from the AUH
> >(as you call it - others call it the principle of plenitude). I am
> >currently writing this up as a paper, and will post this to LANL
> >eprints when ready, but it largely draws upon arguments discussed in
> >this email list.
>
> Sounds like circular reasoning, as stated above, because the AUH
> is itself justified only because of Occam's razor.
I disagree. I think the AUH could be argued for because it is a "zero-
information theory", independent of Occam's razor. Granted, it *is* the
simplest possible explanation, but that's not, in my opinion, the
fundamental justification for it.
> A more interesting
> question is whether the experimentally observed apparent simplicity of
> physical laws (which leads to sucessful use of Occam's razor) is predicted
> by the AUH; maybe that's what you meant. In any case I hope that in your
> paper you will give full credit where it is due to those who made the
> arguments.
>
> - - - - - - -
> Jacques Mallah (jqm1584.domain.name.hidden)
> Graduate Student / Many Worlder / Devil's Advocate
> "I know what no one else knows" - 'Runaway Train', Soul Asylum
> My URL: http://pages.nyu.edu/~jqm1584/
--
Chris Maloney
http://www.chrismaloney.com
"Donuts are so sweet and tasty."
-- Homer Simpson
Received on Fri Oct 15 1999 - 21:35:23 PDT