X-Message-Number: 11624
Date: Tue, 27 Apr 1999 01:54:36 -0700
From: Mike Perry <>
Subject: Again, to Thomas and Bob

Thomas Donaldson, in #11607, notes some difficulties with the idea of
"attempts to devise a computing machine able to act in the world" (with
related comments in #11613). And I have no fundamental disagreement with his
conclusion that a Turing machine would make an impractical device for this.
(Did I ever say otherwise?) A Turing machine, however,*is* useful for
understanding computation at a theoretical level. It has something useful to
say about what processes are likely to be doable at all, and it also has
something to say about efficiency considerations. All our computers up to
now are not only emulable by a Turing machine, but even emulable within a
polynomial time bound by a Turing machine, that is "efficiently" (in an
abstract sense). This is a point I'll come back to.

Now, to Bob Ettinger, #11612

>Late comers who may want more detail and background, please see recent 
>several days of Cryonet. Here a couple of very brief responses to Mike 
>Perry's latest:
>
>>Off the top of my head, [we need many-worlds] because we have to account for 
>the >interference of single particles, i.e. for what happens when only one 
>photon at a >time is involved. Self-interference is straightforwardly 
>explained by many-worlds,
>>which I don't think is matched by any other theory.
> 
>All you need to explain self-interference is to have a wave associated with 
>every photon. Call it a "pilot wave" or whatever you like, or think of it as 
>wave motion in some underlying medium of lower-level particles. Again, think 
>of phonons. We KNOW that these "phenomenological" quanta or "emergent" quanta 
>result from wave motion in underlying media that can be regarded as 
>classical. Interference of a phonon with itself is not mysterious and does 
>not require many-worlds. If there is a valid analogy with fundamental quanta, 
>such as photons, then photon self-interference does not require many-worlds 
>either, and in fact the hidden-variables explanation (not hidden at all in 
>the phonon case!) is not only a potentially adequate explanation, but far and 
>away the most economical one.....Of course, the analogy is only conjecture so 
>far, and few physicists currently seem to pay any attention to it, but it is 
>in many ways much more elegant than many-worlds.
>
To me wave-and-particle or particle-with-wave explanations of the photon are
more contrived than just a wave-only explanation, which, interestingly
enough, is what you have with many-worlds. That is, particles emerge as
virtual effects of the splitting of worlds--all you really ever have is
waves alone. In one sense, then, you can invoke Ockham's razor in favor of
many worlds (it's "long on universes but short on postulates"). The ideas
Bob has about phonons are interesting, and no doubt bear investigation. But
I will say this, if phonons behave like particles at the quantum level, then
maybe a quantum computer can be emulated with them, and then maybe we could
implement Shor's factorization algorithm. If, on the other hand, phonons are
purely classical effects not requiring many-worlds, it would seem then to
open the possibility of solving the prime factorization problem in a
classical device in polynomial time. Whoever can do this has scored a major
mathematical triumph. So far I haven't seen any discussion of this
possibility, and I would certainly like to hear more about it if anyone
knows anything.

Otherwise, it appears that a quantum computer, unlike any other computer we
have, may be able to handle some computations *more* efficiently, in a
fundamental sense, than a Turing machine. That is, no Turing machine, hence
no classical computer, could emulate a quantum computer within a polynomial
time bound (though such emulation would still be possible in an exponential
time bound). If this is true it is fantastic.

Returning to the factorization problem, there is no known way of doing it in
time a polynomial of the number of digits of the number being factored, i.e.
in polynomial time, if you are limited to any kind of classical computer.
But a quantum computer using Shor's algorithm *would* do it in polynomial
(actually linear) time. This can be turned into a good argument for
many-worlds--see Deutsch, *The Fabric of Reality*, pp. 216-17.

>On emulation of a person by a Turing computer:
>
>>This is where I don't follow you. The tape stores a description of the
>>mental state of the person. (That is the paradigm I have in mind, at
>>least.). 
>
>Again, this appears to me to be equivalent to saying that a description of a 
>mental state by any means, including numbers in a book, would be equivalent 
>to the actual existence of a mental state in some living entity. 
>

I think you can relate actual events in the hypothetical Turing machine
environment to events in the "real" person. In neither system are you
dealing with something like unchanging marks on paper. The changes clearly
are happening.
As you quoted me:

>>This description is being constantly refreshed or updated, albeit
>>very slowly and awkwardly. The process of refreshing is being done by the
>>machine's writing the tape. So there are indeed "changes in the recorded
>>sets of numbers or data stores that correspond  to the mental state of the
>>emulation." 
>

Then you say,

>But almost all the changes are mere "scratch paper" intermediate 
>calculations, not elements of the final description of a successor mental 
>state.

But every element of the "final description" is being updated to reflect all
that is happening in the mind of the emulated person. This is not hindered
by any extra work necessary at the "scratch paper" level.

>No doubt you could attach labels to the latter, or have them assembled 
>on separate pages of a book; but please look again at the alleged 
>equivalencies:
>
>a) In the "real" world we have physical objects or systems--obeying laws not 
>yet fully understood--that stand in certain necessary relationships, and 
>changing in certain necessary ways, to produce a living entity.
>
>b) In the world of (imperfect) isomorphism,

who said it *must* be imperfect? 

> the Turing tape, we just have 
>successive sets of marks on a tape, equivalent to successive sets of numbers 
>in a book. In principle, we can envision this as a very big book with a whole 
>lot of  very big pages; each page is just a set of numbers and corresponds to 
>a particular mental state of the emulated person. Can you seriously believe 
>that this book not merely describes but CONSTITUTES a living individual?

Yes, because it is no ordinary book, because it is constantly being updated
or rewritten. (Books on my shelves certainly aren't like this! My brain is!)
Moreover, as I envision it, there are ways in principle that I can
communicate meaningfully with this "book" and have it produce meaningful
responses. Not exactly your conventional book.
 
>
>I don't mean to appear to hammer on Mike (not that he needs protection), but 
>I keep trying to ascertain where my attempted expositions fail to be clear or 
>persuasive--or where I might (Is it possible?) be wrong. I'm sure Mike does 
>the same, in polishing up his book.
>

Thanks for raising these points, and good luck with your book too.

By the way, to answer Bob's request, it's okay with me if anybody wants to
reprint or publish anything I post here, unless I indicate otherwise, except
that material I quote from other peoples' postings will, of course, have to
be cleared with them too. (I think though that these quotes will usually
fall within the "fair use" limits.)

Mike Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=11624