From: Eliezer S. Yudkowsky (sentience@pobox.com)
Date: Sat Feb 14 2004 - 17:08:50 MST
As one of the chief early offenders in the "Well, past physics got
revised, so why not current physics?" debate, I'd like to say that today's
physics got a *lot* more impressive once I knew a little more about it.
In the latest physics, the bound formerly known as Beckenstein (now known
as the 't Hooft bound, briefly known as the Susskind bound) is a
holographic bound on the entanglement between a spherical volume and the
rest of the universe, measured in Planck units of the surface area. It's
not something you could defeat by being clever about how you stored your
data - if the holographic theories are correct, the 't Hooft bound is
absolutely fundamental.
It's that "absolutely fundamental" quality to modern physics that might
make the bounds resistant to any amount of intelligence and creativity.
Intelligence isn't actually magic. If it sometimes seems that way, it's
because someone a bit smarter does something that violates your model of
reality, not because intelligence is actually some kind of hole in
reality. According to my own understanding of intelligence, nearly
everything that once puzzled me seems to exist within physics and is
continuous with physics, and I expect that the few remaining mysteries are
not any different. Physics is sovereign over intelligence, not the other
way around. When we tentatively model physics, we are tentatively
modeling something that implements the intelligence with which we
tentatively model it. I particularly regret inventing the phrase
"ontotechnology", which seems to have become an excuse not to learn
physics. This is one of those areas where people really didn't need
another excuse.
Some obvious absolute fundamentals are Special Relativity - no local speed
faster than light, period, because it's woven into the definition of
causality. Even at my worst, I didn't suggest you could go faster than
the local speed of light by pushing real hard. I did suggest that the
various speculations about wormholes and other ways to fake FTL based on
General Relativity showed that the pragmatic "interstellar speed limit"
problem might be solvable. But then again, it might not be solvable. The
prospect no longer fills me with dread. I can have fun in a speed-limited
universe.
You can't tell two electrons apart. Ever. It's not that our current
physics doesn't tell us how to do it, it's that our current physics
*requires* two electrons to be absolutely interchangeable. If you read up
on bose-einstein and fermi-dirac statistics, you'll see what I mean.
Today's hypotheses of physics aren't things you can just change out like a
used air filter. They seem to be telling us something fundamental about
the nature of reality. Conservation of energy doesn't just mean that we
haven't so far found any physical operation that changes the total energy;
it reflects the invariance of the laws of physics under translation in
time, and specifically the invariance of the computation of the
Hamiltonian of a system. Things like that.
The laws of physics we know today are not compartmentalized. They could
be wrong, but you can't just rip out Special Relativity and replace it
with a different hypothesis that allows you to go faster than light.
That'd be... I can't even put it into words. Reality would unravel. The
socks of reality don't have loose strings - if you take out a string, the
whole sock falls apart.
The limits imposed by our models, and sometimes our hypotheses, of
physics, arise from exactly those features of the model that seem to
shout: "Here is something truly fundamental about reality!" Back in the
days when we were tossing theories out the window every half-decade, the
theories we got rid of didn't have that "Here is something absolutely
fundamental!" look about them. That, in particular, I didn't understand
when I made the historical comparison. History doesn't always repeat
itself. Today's physics might be here to stay; or, when it settles down,
all the fundamental limits might still be there. At any rate I am willing
to bet that no one can distinguish two electrons - EVER.
-- Eliezer S. Yudkowsky http://intelligence.org/ Research Fellow, Singularity Institute for Artificial Intelligence
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:45 MDT