From: Lee Corbin (lcorbin@tsoft.com)
Date: Sun Jun 09 2002 - 18:30:59 MDT
Wei Dai writes
> > Very interesting. But I choked on the part where you spend
> > a lot of energy moving matter around. Were you talking about
> > interstellar distances? Boy, I hope not.
>
> I was actually talking about intergalactic distances.
Gasp! That's worse! :-)
> If the expansion of the universe is accelerating, everything
> outside our local cluster will eventually disappear behind a
> causal horizon. So if we want matter outside our local cluster
> to contribute to our civilization, we have to go and bring it back.
We have to go lasso that matter only if by "our civilization"
you mean something that does not apply to our colonies. Why
are you not content with our merely sending our programs
(equals our selves) out to the distant matter to colonize it?
It still seems to me that you are wasting absolutely huge
resources. "If the mountain will not come to Mohammed, then
Mohammed must go to the mountain." (The actual Islam phrase
is "If Mohammed cannot go to the mountain, the mountain will
come to Mohammed", but most people on this list will find the
the former version more realistic.)
> It's true that moving matter uses up energy, so there's
> a tradeoff between the total amount of matter/energy that
> contributes to all civilizations versus how big/complex
> each civilization can become. That's a value judgement
> that we may have to make in the future.
Okay. But like I say above, it seems weird you wanna lasso the matter.
> Yes, black holes are entropy sinks, that's why they're good
> for dumping excess entropy.
I've never understood that properly. Could you flesh out a
concrete scenario or example?
> > No, Eliezer and I have gone around this one before. He didn't
> > like total freedom at lower levels because even lower level
> > simulatees might be abused. Don't you agree with him? Above
> > you stated that a simulatee's permission should always be
> > obtained before he was run (which would make most historical
> > simulations very difficult). I, on the other hand, favor the
> > solution closer to private property.
>
> I think I agree with Eliezer here. Assuming a Meta-Golden rule scenario, I
> would be willing to give up the right to abuse simulatees if that meant I
> would not be abused as a simulatee by higher levels. I do consider having
> someone participate involuntarily in a historical simulation to be abusive.
I am concerned about the computational resources required to
adequately monitor what simulatees are doing. Moreover, there
is the extremely tricky notion of what constitutes abuse. Who
exactly is to say whether running Winston Churchill again is
abusive except that person? Moreover, many folks go so far
as to entertain the idea that the people themselves can be
mistaken about things like that.
But I concede that you have an important point. (Also I'm tickled
pink that people are coming to understand that re-running someone's
life is not a no-op. It should matter to you whether or not two
completely identical runs of your life occur, or only one! Yay.)
To be really nice, we should always ask.
> If you really want to run a historical simulation, perhaps you
> should ask for permission both to be in the simulation and to erase the
> memory of being asked for and granting this permission, and reobtain
> the permission every so often as the simulation runs.
Yes, that's hilarious. Suppose I'm being stopped every five
minutes or so and I re-sign all the papers.
But seriously, about simulations: I have always differentiated
between *portrayals* and *emulations*. A concrete example will
explain the difference: suppose I want to run the '60s again,
and try to get everyone's permission. If only half the people
agree, I may not be dead in the water. Supposing I can get the
help of something much smarter than I, I may request that it
simply portray the reluctant to the satisfaction of myself and
the simulatees. In other words, intelligence advanced enough
can merely portray someone without actually experiencing their
emotions.
Lee
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:14:41 MST