On August 1, 1999, Robert J. Bradbury wrote:
>
> > Gina Miller <nanogirl@halcyon.com> wrote:
>
> > *Say that there are two "copies" of an original person uploaded. There are
> > two persons with the same uploaded (or downloaded) data, consisting of one
> > originals information, what is the perspective of the consciousness?
>
> There is a very interesting discussion of many interesting aspects of uploading/
> conscious self-editing in "Permutation City" by Greg Egan. I think there were
> problems with the premise on which the book was based, but the exploration of
> the various aspects of this technology was very interesting.
>
Yes, mind copying has been featured in _Permutation City_ and also in Egan's _Diaspora_, now *there's* a fun novel of transhuman adventure, and interstellar travel, complete with strange cosmic dangers and physics breakthroughs -- and all without ever getting into the standard sf "warp drive" scenarios. Actually, mind and/or body copying has long been a fairly common idea in sf, maybe all the more so because hardly anyone else ever touches it, except maybe the occasional philosopher of science. Besides Egan, I think of Poul Anderson, with _Harvest of Stars_, and sequels, uploading/copying are featured there, also at least one or two of John Varley's stories, also the _Tower to the Sky_/_Fourth Intercometary_ series by Philip C. Jennings, etc. Naturally, mind copying is of interest to AI researchers, since the mind of an AI obviously ought to be copyable, same as any other computer-based information.
With the above as "preface", I think what Gina is asking, is, what should we expect the personal experience of copying to be, assuming that copying is possible, and assuming that it were to happen to oneself? To begin with, I'd take it as basic that, if I were copied, the experience would amount to a "fork in the road" of my own personal experience, or history of perceptions. For instance, if one copy goes space traveling, and the other stays on Earth, that amounts to two, basically separate, personal timelines in the end, both of which happen to have shared the same material body and history, up to the event of being copied. Notice that there could potentially be any number of such personal histories coming out of any of our life histories, if there's a chance of being copied later. The trick, it would seem, is to be able to identify with a properly, causally developed "thinking timeline", rather than with any special bit of matter in one's body. For those who've read Egan's _Permutation City_, notice that he seems to cheat on my requirement of a "properly, causally developed" thinking process as a basis for self awareness -- oh well, it's all very controversial, I suppose!
Suppose I were planning in advance to be copied in some high tech procedure or other, one copy to go for a vacation on the Moon, the other to stay on the Earth. Doesn't this sound as though I should expect a 50:50 chance of doing one thing or the other, myself, personally? In other words, if my specified plan comes about, I could end up identifying myself with one of two personal histories, so it's a 50:50 split as to which I'll be, if I think about it in advance. Similarly, if I were planning for three copies, instead of just two, it'd be a 1/3:1/3:1/3 split, as to the way I should look at the chances of which timeline will represent my "self history". This brings up an awkward question: suppose I plan to just make two copies of myself, then have just *one* of those copies *recopied* shortly after? If I've got a 50:50 chance of becoming one or the other in the first copy, then how do I get anything like a 1/3:1/3:1/3 chance of identifying myself as one of three copies in the end? Such a conundrum, some would say that this proves that personal copying must be impossible, I'm sure! As it happens, I have a slightly different idea of how this might be properly regarded. For concreteness, say that of the initial pair of copies, one stays on Earth, while the other goes to the Moon. *Then* the copy that goes to the Moon is to be recopied once he arrives there, there'll be copy 2a and copy 2b, so to speak -- can you guess my actual chances of identifying myself as personally arriving on the Moon, vs staying on Earth?
I've said that I can consider my current history up to now as potentially representing *any* number of parallel histories, so let's put down an *arbitrarily* even number, and say that I could potentially "be" any one of a *dozen* histories that have been identical up to some initial point in time, the beginning of my scenario, here. Out of 12 Potentially Different Timelines, or PDT's, if I were copied right now, the "fair minded" assumption to make would be to think that 6 PDT's would go each way, so I'd expect, as a first guess, that I've got a 50:50 chance of splitting my history either way, that's clear enough. Notice though, that there's no physical process that *guarantees* this particular split, so if I take into account what's to happen later, perhaps this should modify what I assume in this regard? For instance, if the Moon vacationing copy is to be recopied into "me" 2a, and "me" 2b, *that* almost sounds as if two of my assumed 6 earth based PDT's ought to immediately leave, one going to join up with my physical copy 2a, and one going to join up with copy 2b, so that if 2a and 2b would be assumed to have three PDT's initially, they'd then end up with 4 apiece. In this PDT redistribution, my Earth based body is left with 4 PDT's, so I've got a 4:4:4 split, justifying a three way assumption of equal chances of becoming each of my distinct physical "selves" in the end! Oh yes, don't forget, "PDT's", that's "Potentially Different Timelines"; this could also be referred to as "potentially different personal viewpoints". Here, we assume some arbitrary number of these potential viewpoints at the outset and see where it takes us.
OK, if you've followed me so far, here's the kicker! I've said, as a
first assumption, that the recopying of my Moon self seems to imply a
redistribution of my personal outcomes, with 2 of my 12 assumed PDT's,
or personal viewpoints, just *leaving* the Earth to join up with my now
more numerous copies on the Moon. At this point, let's recall though,
that I've also assumed that it's important that a personal timeline, or
viewpoint, be developed in a consistent way. Notice, here, that there is
something really improper about saying that 2 of the historical lines
could just "leave"
my Earth-bound history and join up with the lunar ones! You see, *that*
would
require that some of the Earth based *experiences* be totally erased
from my memory, so that some Moon based experiences (the ones between
the first and second copying event) could be substituted in the case of
those two particular timelines! Since these two assumed PDT's are now
seen to be improper, let's now declare them to be completely invalid,
shall we, so that in the final analysis, we can no longer assume that
they have any relevance to my chances, or to any outcome that I can
identify with? Declare the two "redistributed" PDT's to be invalid and
we are left with 10 of the 12 that I initially assumed, 4 of which stay
on Earth, as potential viewpoints in my copy #1, while copies #2a, and
#2b end up with 3 of the 10 "proper" PDT's each. Now, in this example,
we're down to base 10, very nice, actually a chance of 40%:30%:30% of
going each of the three ways, in the forecast of two future copy events
that I've described.
*Anyhow* the true relevance of possible mind copyings seems bound to be controversial to say the least, since there's a mathematical "history tracking" aspect to it that almost seems to go beyond anything strictly physical! More "Computability of Consciousness" anyone (that was a fairly recent subject header, on somewhat related matters)? Is this preserving of a properly, causally developed mind "track" for personal consideration, as important as I think? Maybe some will say that mind copying just doesn't make sense, or is impossible somehow? Some of you forward thinking people undoubtedly have some ideas about this, meanwhile I'm *serious* about my "recopying by cancelling 2 of 12 Potentially Different Timeline viewpoints". That is, I *think* I'm serious, taking probability recalculations into new realms, this being a "modest proposal" and all that :^) Generally, this method of looking at personal copying prospects would always start off by assuming enough separate "mind timelines" to redistribute among copies as needed, then would cancel any redistributed lines that come out inconsistent in their causal development of memories . ..
David Blenkinsop <blenl@sk.sympatico.ca>