AI and cognitive psychology rant (getting more and more OT - tell me if I should shut up)

Andrew Dalke adalke at mindspring.com
Sat Nov 1 18:39:54 CET 2003


Anton Vredegoor" <anton at vredegoor.doge.nl>:
> The more territory modern science covers, the more it becomes clear
> that the known parts of the universe are only a small part of what is
> "out there".

But as Stephen pointed out, the new things we find are new
in terms of arrangement but still following the same laws of physics
we see here on Earth.  (There *may* be some slight changes in
cosmological constants over the last dozen billion years, but I
haven't followed that subject.)   These arrangements may be
beyond our ability to model well, but there's little to suggest
that in principle they couldn't be.  (Eg, QCD could be used to
model the weather on Jupiter, if only we had a currently almost
inconceivably powerful computer.  Running Python.  ;)

> So "objectively" science gains more knowledge, but
> relatively speaking (seeing it as a percentage of that what is
> currently known to be not known, but knowable in principle) science is
> loosing ground fast. Also an even greater area of the universe is
> supposed to exist that we will not even have a chance *ever* to know
> anything about.

That's a strange measure: what we know vs. what we know we
don't know.  Consider physics back in the late 1800s.  They could
write equations for many aspects of electricity and magnetism, but
there were problems, like the 'ultraviolet catastrophe'.  Did they
consider those only minor gaps in knowledge or huge, gaping chasms
best stuck in a corner and ignored?

Is the gap between QCD and general relativity as big?   Hmmmm...

> Trough out the centuries there has been evidence placing humanity
> firmly *not* in the center of the universe. First the earth was proven
> to rotate around the sun and not the other way around, next our sun
> was not in the center of the galaxy and so on.

You use this as an argument for insignificance.  I use it to show
that the idea of "center of" is a meaningless term.  If I want, I can
consider my house as the center of the universe.  I can still make
predictions about the motion of the planets, and they will be
exactly as accurate as using a sun-centered model.  The only
difference is that my equations will be a bit more complicated.

> Maybe now it is time to accept the fact that all the things we know
> taken together are only a small fraction of what can be known, and
> even more that that fraction is not even in a central position in the
> larger universe of the knowable, and that the knowable is just
> disappearingly small compared to the great abyss of the unknowable
> where everything is embedded in.

Why?  I mean, it's true, but it seems that some knowledge is
more useful than others.  It's true because even if there were
a quadrillion people, each one would be different, with a unique
arrangement of thoughts and perceptions and a unique bit of
knowledge unknowable to you.

> How come then that the sciences have been so uncanningly effective
> given that they are such an arbitrary choice within the knowable? The
> answer is of course that there are a lot of other possible sciences,
> completely unrelated to our own that would have been just as effective
> as -or even more effective than- our current sciences, had they been
> pursued with equal persistence during the same amount of time over a
> lot of generations.

I don't follow your argument that this occurs "of course."

It's not for a dearth of ideas.  Humans did try other possible
sciences over the last few millenia.  Despite centuries of effort,
alchemy never became more than a descriptive science, and
despite millenia of attempts, animal sacrifices never improved
crop yields, and reading goat entrails didn't yield any better
weather predictions.

On the other hand, there are different but equivalent ways to
express known physics.  For example, Hamiltonian and Newtonian
mechanics, or matrix vs. wave forms of classical quantum mechanics.
These are alternative ways to express the same physics, and some
are easier to use for a given problem than another.  Just like a
sun-centered system is easier for some calculations than a "my house"
centered one.

On the third hand, there are new theoretical models, like string
theory, which are different than the models we use.  But they are
not "completely unrelated" and yield our standard models given
suitable approximations.

On the fourth hand, Wolfram argues that cellular automata
provide such a new way of doing science as you argue.  But
my intuition (brainw^Wtrained as it is by the current scientific
viewpoint) doesn't agree.

> The effectiveness of the current natural sciences is a perceptual
> artifact caused by our biased history. From a lot of different
> directions messages are coming in now, all saying more or less the
> same: "If asking certain questions, one gets answers to these
> questions in a certain way, but if asking different questions one gets
> different answers, sometimes even contradicting the answers to the
> other questions".
>
> This might seem mystic or strange but one can see these things
> everywhere, if one asks that kind of questions :-)

Or it means that asking those questions is meaningless.  What's
the charge of an electron?  The bare point charge is surrounded
by a swarm of virtual particles, each with its own swarm.  If you
work it out using higher and higher levels of approximation you'll
end up with different, non-converging answers, and if you continue
it onwards you'll get infinite energies.  But given a fixed
approximation you'll find you can make predictions just fine, and
using mathematical tricks like renormalization, the inifinities cancel.

For a simpler case .. what is the center of the universe?  All locations
are equally correct.  Is it mystic then that there can be multiple
different answers or is simply that the question isn't well defined?

> One example would be the role the observer plays in quantum mechanics,
> but something closer to a programmer would be the way static or
> dynamic typing influence the way one thinks about designing a computer
> program.

The latter argument was an analogy that the tools (formalisms) affect
the shape of science.  With that I have no disagreement.  The science
we do now is affected by the existance of computers.  But that's
because no one without computers would work on, say, fluid dynamics
simulations requiring trillions of calculations.  It's not because the
science is fundamentally different.

And I don't see how the reference to QM affects the argument.  Then
again, I've no problems living in a collapsing wave function.

> In a certain sense all science reduces the world to a view of it that
> leaves out more than that it describes, but that doesn't preclude it
> being effective. For a last example, what about the mathematics of
> curved surfaces? Sciences have had most of their successes using
> computations based on straight lines, and only recently the power of
> curves is discovered as being equal or more predictive than linear
> approaches.

Yes, a model is a reduced representation.  The orbit of Mars can be
predicted pretty well without knowing the location of every rock on
it surface.  The argument is that knowing more of the details (and
having the ability to do the additional calculations) only improves the
accuracy.  And much of the training in science is in learning how to
make those approximations and recognize what is interesting in the
morass of details.

As for "straight lines".  I don't follow your meaning.  Orbits have
been treated as curves since, well, since before Ptolomy (who used
circles) or since Kepler (using ellipses), and Newton was using
parabolas for trajectories in the 1600s, and Einstein described
curved space-time a century ago.

> Abstractness doesn't preclude effectiveness, but to try to use
> abstractions to understand the world is foolish since it doesn't work
> the other way around.

You have a strange definition of "effectiveness."  I think a science
is effective when it helps understand the world.

The other solution is to know everything about everything, and, well,
I don't know about you but my brain is finite.  While I can remember
a few abstractions, I am not omnipotent.

                    Andrew
                    dalke at dalkescientific.com






More information about the Python-list mailing list