[extropy-chat] Fragmentation of computations

Lee Corbin lcorbin at rawbw.com
Tue Mar 27 02:49:40 UTC 2007


Mike writes

>> So how can you be so *sure* that in the lookup vs.
>> causal computation process, the lookup process is
>> not akin to simply playing back a perfect 3D movie?
>> You must entertain the possibility that this is so, and
>> that if you are enjoying your life then you will equally
>> enjoy repeat computations, but won't enjoy or
>> experience at all any repeat movies of your life.
> 
> I think being completely sure of anything is impossible without also
> removing the part of the brain that logically measures "sure-ness"

Yes, it was probably unfair of me to Russell to word it like that.

>> to you, then. Suppose that there are infinitely many
>> galaxies (so that numbers don't cramp my style here)
> ...
>> And that between G3 and G4 there is another
>> unmistakable image of Russell, this time 1 billionth
>> of a second past that same time.
> ...
>> So the universe is exhibiting a succession of images
>> of you over that five minute part of your life.  Is there
>> or is there not any experience taking place out there
>> in that sequence of patterns?
> 
> Are you assuming some mundane arbitrarily large number of galaxies
> within a single space-time coordinate system (universe) ?

Yes, I meant to convey an actual aleph-zero collection of galaxies.
(Tegmark postulated such a model, and I gather that a number of
people had before him entertained the idea that the universe is
spatially infinite.)  So, yes, by my lights that includes a coordinate
system that is capable in principle of referencing any galaxy, or
(see below) any real-valued point among them.

> given each spec occurs simultaneously within a given temporal moment,
> then it is a series of pictures you describe and the Casablanca/movie
> is analogous.  I would agree that the computation of the universe at
> this moment has already been done, so the perspective of any spatial
> distance to each spec does not bring new experience (using your
> definition of computation-as-experiential life)

We agree, but I'm not sure how heavily you're leaning on "perspective".
I meant to respond to an earlier post that seemed to be implying
that that which is not observed by someone does not exist. Most of us, 
however, endorse realist models in which our language can *refer*
to objects or structure that is independent of us, and which existed
prior to our perception of it, and even prior to our own existence.

> If there is only one spatial coordinate system, but multiple times are
> used to record the state of each spec at G1, G2, etc. then we
> (observers) must be able to change our perspective outside the time
> dimension as easily as we changed spatial dimension above.  I think
> this is more difficult to imagine since we seem to be bound to moving
> positively through time, but it's really not that much different after
> the initial dis-intuition wears off.

I don't know what you mean by this.  But perhaps what you
write next explains it:

> So imagine we do possess the degree of freedom required to visit any
> space-time in our lives.  The ghost of Christmas past is invoked to
> show us any moment of our life up to the moment it appears.

Arthur C. Clarke mentioned such a history machine in "Childhood's End",
and then he and Baxter wrote "The Light of Other Days" to enlarge on 
the theme.

> Somehow our ghost-enabled self is able to observe the earlier state.
> Thoroughly searching our memory does not provide comparable detail
> because our brain's recording media was too slow and small to record
> everything in high definition - so the "lifelike" resolution proves
> that we are not merely recalling a memory, but observing a fully
> computed earlier moment of existence.

Yes.  This is observation.

> Now observe all the way up to the moment the ghost enabled your
> enhanced observation.  I suggest there is an unexpected limitation due
> to the recursion of ghost-enabled observation of ghost-enabled
> observation (the mirror in a mirror, or camera viewing the output
> monitor feedback) or God Over Djinni infinite stack overflow (from
> Godel, Escher Bach)

I think I see your meaning:  it would be possible with such machinery
to exhibit any moment in the *past*, but moving around information
instantaneously entirely in the present seems very problematic. To
exhibit your current moment to you, the GOD would have to have
already known it before your present moment was computed.

> Every time you solve this problem by allowing an extra dimension of
> freedom in which to change the observation point, you arbitrarily
> create a new context in which the original spec (G1) can be observed
> or related to.  So in that respect, a new computation or experience is
> gained.

Well, yes, a new *observer* experience is created. I could go out
and begin looking at my car from a number of different angles, and
as I do so, I gain a new experience. Although frankly with my car,
once you've seen it from one direction, you've seen it from all.

The model which we were thinking about, however, did not contain
---I think it's proper to say---an observer.  We were discussing
a computation of a subject, that is, of an entity, and that being
need not have any sense experience to confuse the issue. Then
the next step was to load a sequence of such states into storage
so that a "movie" could be re-run, or load them into a look-up
table using for the address a perfect hash function of the state.

> BTW, is your reference universe granular such that there exists a
> smallest unit of space-time (analogous to the set of integers Z) or
> smooth (comparable to the set of Real numbers R)  Are there universes
> modelled on the set of complex numbers?

I think our universe is modeled on complex numbers!  At least
in quantum field theory  :-)    But I was assuming an event at
each point of a real-valued spacetime.  I wasn't thinking of a
cellular or granular universe, although I think that the same
arguments probably apply.

>> > Simulator flips a switch to turn off the Hash Life
>> > optimization and change the program to straight
>> > linear computation.
>>
>> Oh, good.  More genuine runtime for me!
> 
> I get the image of a LeeCorbin worm infecting every processing node
> in the universe and causing such an intense storm of internode
> communication that the entire simulation machine simply gets turned
> off and state is restored from backup.

This is why, I suppose, it's best to try to be a bit moderate in all 
things!  But me being so generous, you yourself can count on being
emulated within whatever computational space I get my hands on.
And since I'm so nice, I'm sure you'll do me the return favor and
run me now and then in yours!  As might be happening to me right
now.............hmm, pretty good, thanks.

Lee



More information about the extropy-chat mailing list