[extropy-chat] Fragmentation of computations

[Lee Corbin]

Rafal writes

> On 3/25/07, Lee Corbin wrote:
>>
>> > Moreover, at the next frame, which is computed, the subject would
>> > suddenly remember perceiving the light and have no recollection that
>> > anything unusual had happened.
>>
>> That's right.  It takes us back to the by-now old observation that God
>> could have created the universe 1 second ago, and we'd be none the
>> wiser.
> 
> ### I would disagree here. From what I know about physics of our world
> you cannot predict the future of a physical system at an arbitrary
> point in its trajectory without precisely tracing the whole
> trajectory, or rather, the sheaf of possible trajectories of the
> system over time.

Yes, that's reasonable. But you are thereby outlawing an omnipotent
deity, whereas I was just assuming that if Such existed, It would be
able to create our present universe in its current state from scratch.

Very well. 

> If all you know is the state S1 at time T1, you have
> to recapitulate the whole multiverse in the light cone starting at T1
> with state S1, to describe the probability distribution of all states
> at any time T2 after T1.

Could a semi-omnipotent God go around to each point in the quantum
fields which make up reality, and juggle the local group at random
thereby putting into the universe the information?  My point being
that if such is conceivable, then by incredibly remote chance---or He
is incredibly lucky---He could have come up come up with this
moment's arrangement without doing any other calculation.

> The same applies to the consciousness simulated on a GLUT - lookup of
> the state of a physical system *is* a calculation of that state. It is
> computationally impossible to  "compute" only select "frames", you
> must run the whole film.

Well, we can argue about the GLUT lookup method later.  Since I
sense that several of us here may not be quite really understanding
each other, you would allow, I assume, that simply *loading* each
frame out of memory into some predesignated register---so that the
appearance of computation arises---still no actual computation
comes about, and we could arrange any order we wished for the
sequences of frames or states to display?

> The same pertains to a non-periodic CA - you really have to run
> the whole calculation from state #1 to state #1000 to actually see
> state #1000.

Yes, unless you loaded frames from an earlier run, or you got 
incredibly lucky in the specification of an initial state.  (The present
state of our universe is, of course, incredibly non-random,
containing as it does immeasurable redundancy and structure.)

Lee

> So, no time-sovereign god and no zombies are possible.