Re: Uploading

From: Eugene Leitl (Eugene.Leitl@lrz.uni-muenchen.de)
Date: Sun Mar 10 2002 - 02:37:28 MST


On Sat, 9 Mar 2002, Richard Steven Hack wrote:

> Your purpose of discussion perhaps - my purpose may vary...

In other words, Mr. Hack isn't really here for an argument.

> > [magic universe being impossible to argue about]
>
> Correct - my point, I believe.

Excellent. This means you're just self-destructed your argument base. If
you propose magical physics, no one will argue with you in here. God knows
we have better things to do.

> Doesn't interest me - I was merely responding to assumptions by other
> parties.

He's most definitely not here for an argument.

> Given that I do not know where I will be as a Transhuman in a thousand
> years, and given the likelihood that my brain at that point will be
> quite capable of doing such an analysis in much shorter time than I have
> to deal with this now, I see no reason to worry about physical limits at
> this time, thank you.

Hmm, can't prove there's magical physics right now, so I assume there is
magical physics. Yes, it all makes more sense now.

> Ah, yes, Robert Ettinger's hypothesis, I assume. If you exist at a
> single point in space/time, you can't be immortal; if you are
> distributed, you can be.

Wouldn't call it a hypothesis, just good common sense. If there's
people/natural phenomena trying to nuke you, distribute resources, and
communicate using a remixer network.

> I'm not convinced. Worse, I have a metaphysical problem with the notion
> of "distribution" - that I do not know of any technology to accomplish
> it AND preserve the identity and continuity of the entity in question.

Mr. Hack is not convinced. That's, of course, a killer argument. A
"metaphysical problem", I guess that's the next logical step if one
assumes magical physics.

Remote incremental backups don't require a bidirectional link. The vacuum
is a rather good FIFO, if your signal rate is high enough, and there's
enough of it. So you can ignore relativistic latency, and just fire out
your journaling trail like tracer bullets, as fast as it is being
generated. As soon as the trail stops, because e.g. you've been nuked, the
static image (or images) gets automatically reactivated, and tagged with a
canned response "You've been reactivated at coordinates/time stamp foo,
last good packet came from coordinates/timestamp bar. Better luck next
time".

This not to say that grand scale distributed systems are impractical.
Hierarchical assemblies would have a wide range of temporal operation
signatures. A local node having to deal with a contingency (something
munching through the hull and mucking up the sailpanel; an accident in the
microsingularity factory; galactic center flareup) without having to
exchange messages with a remote node.

> As noted above, the concept of "distribution" seems to be hand-waving
> since we know of no technology (at this point) to do it and as noted
> above preserve the identity and continuity of the entity involved. If

Correction: Mr. Hack doesn't know of any such technology. He's clearly
omniscient, so he doesn't pause to ask whether others know of any such
technology. It's too bad Mr. Hack won't see this post.

> you can do it, of course, then you would be correct and the approach
> would be attractive to me. "Uploading" as it has been described to me
> does not meet my criteria.

That's too bad. I guess we have to abandon uploading now, because it
doesn't meet Mr. Hack's criteria. Excuse me for a moment. (Goes outside,
there's a muffled shot to be heard. Then, silence).

> Ah, I may miss the point. Why would the distributed entities produce
> the least novel info? Mike Lorrey seems to believe the opposite - that
> you distribute yourself to increase your rate of experience. You seem
> to be saying the opposite.

Ecologies have no problems with distribution. If your model has one, it's
most likely your model's fault than that distribution is infeasible.

> But I still see no evidence - other than your notion of distribution to
> avoid space/time catastrophe - that replication would be considered
> desirable by a posthuman entity. The "society" notion advanced by

Hey, this replication thing is just a minor feature of life. No one really
likes replicate. That we're here (and great deal of other critters) is not
really consequential. We *hate* replicating, let me tell you. We'll stop
tomorrow. All of us, including the bees, the flowers, and the
rhabodoviruses. Because, frankly, replication sucks.

> others - that posthumans would replicate because this would increase the
> utility of all - may or may not be true. We don't know what Transhuman
> intelligence will be like - they may desire or need society; they may
> not.

Because I'm less ready to chuck out science and embrace magical physics
and metaphysics, I tend to postulate a *population* of *diverse*
*replicating* beings *stochastically* sampling behaviour space. This
is pretty much business as usual, only more so. If someobody postulates a
departure for this, he must show a mechanism how this will change, and how
it will stay that way.

SysOp scenarios are so far the only proposed mechanism. I don't believe in
SysOp scenarios, but even if I did: the population mechanism would still
apply. You'd have to prove that the entire accessible cosmos (something
like 100 GLyrs across, or whatever runaway expansion would seem to give
us) will get sysoped, every single time.

> On the other hand, it could well be that Darwinian competition continues
> IF in fact there are hard limits and the entities come up against them
> in a reasonably short time (by their lights). If that is the case, then
> the Highlander motto, "There can be only One", may well prove true.

Obviously, Mr. Hack has a very skewed model of darwinian evolution. I am
not surprised.

> Perhaps we haven't been contacted by higher intelligence because HE
> doesn't give a damn... In that case, Doctor Doom starts to look like a
> better role model all the time...

Somebody please pinch me. This isn't happening, is it?



This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:12:54 MST