From: Daniel E. Azzopardi (desva@slac.stanford.edu)
Date: Mon Nov 27 2000 - 11:24:45 MST
On Mon, 27 Nov 2000, Eliezer S. Yudkowsky wrote:
> "Eliezer S. Yudkowsky" wrote:
> >
> > GUTs are epicycles. GUTs are another pet peeve of mine. Who moved these
...
> > questions are.
>
> Let me clarify the phrase "reductionistic paradigm". I don't mean that
> reductionism in general will stop working, of course; we appear to live in
> a reductionist Universe. I mean that the specific method of looking for
> new particles with higher energies has played itself out, and so has the
...
> changed our picture of the character of physical law.
>
Hi Eliezer,
I've chosen to reply to this instead of your previous reply
(because I've got a thesis to write ;-) ). Your previous reply contained
many good points many of which I agree with, and I'll try to work what I'd
like to say into here.
Firstly, in direct response to the above, CERN has recently
announced they have a Higgs candidate (they're ont going to be able to
confirm it however...) - this is the holy grail at the moment, and
understanding this and CP violation (hopefully measured at SLAC, but I'm
biased) could well open up those GUTs to direct experimental assault.
There are billions of people alive in the world today, millions of very
smart people - why shouldn't a vanishingly small fraction of them spend
their time inventing competing theories which can't yet be put to the
test? Even if they are barking up the wrong tree, creatively mapping out
possibilities for future investigation is better than doing reactionary
science.
> Or, like relativity, the next major discovery may involve no changes
> whatsoever in our description of the particle families.
>
You talk about state vector reduction and the non-local nature of QM, and
I think you are suggesting that our next big breakthrough will be a
paradigm shift due to some deeper level of understanding gained there.
Well, it's a difficult problem, and most physicists are content to leave
it in the realm of philosophy rather than science (in much the same way as
the many worlds interpretation). Bell's theorem "shows" QM is non-local,
and there where a whole series of experiments in the late 80s/early 90s
(and probably more since then, but I haven't had time to keep up with
developments in this sub-field since I first studied it) that attempted to
prove this experimentally. There was also a recent macroscopic observation
of the collapse of the wavefunction [Physics World, about two issues ago].
Experiments such as these are already happening, and may give us further
insight, but then again we (~90% of the Physicists that have ever lived)
have been worrying about EPR for 7 decades and still haven't gotten
anywhere... it may be that there is nothing else to know and we've reached
the limit there. That hasn't stopped QM becoming relativistic QM, becoming
calculable quantum field theories, leading to the Standard Model.
Hopefully we still have some surprises beyond the SM... we need those
surprises if we're going to choose among theories and develop our models
further.
Given the number of Physicists around and the lack of
understanding in the fundamental areas of QM and GR, we probably need
better (ie: machine) intelligence to give us the breakthrough you predict.
> > GUTs throw hugely increasing amounts of complexity in
> > exchange for vanishingly small returns of prediction, and this should be
> > enough to tell us that GUTs are on the wrong track.
Well, we don't know until we put it to experiment! Is there substructure
below the three generations of fermions? Why the heirarchy of masses? More
importantly, do any of these questions matter to readers of this list? I
think they do. The study of symmetries and groups is connected with
particle physics (would the pioneers of QM have guessed this?), who knows
what advanced, and probably highly abstract, mathematics could be useful
to someone developing a complex system (such as an AI). On the other hand,
I make a lot of use of statistics (which I absolutely hate, and don't
really consider to be advanced math) - and this seems to appear
everywhere. Anything involving evolution, complex systems, many body
problems, seems to require statistical analysis at some point.
At the end of the day, to get from SL(n-1) to SLn we're going to
need people/entities that can take apart a complex problem and code up a
solution atomicly. I think approaching coding as an art, implementing ad
hoc functionality as required, will get one nowhere. I think most sensible
companies realise this is the case and don't hire artists for programmers.
(That said, I know companies which claim the opposite, perhaps just
because they don't want to alienate all the artists who form the majority
of the graduate pool).
For our vision I'd guess we would absolutely need mathematically
literate programmers who (like yourself) have an eye on physics, and
probably biology and psychology also. Right now I'm firmly in the physics
side (but I do lots .. too much ... coding ;-) ) - my goal would be to
gain Knuth like abilities in my approach to coding whilst maintaining an
interest in the sciences.
I agree with other posters who state we will absolutely need
mathematical models of mind. Hopefully, we'll develop them in tandem, as
has been the case (almost) for the history of modern physics.
Cheers,
Dan
/----------------------------------------------------------------------------\
| Daniel E. Azzopardi | http://www.desva.co.uk | QMW: +20 7882 |
| Back in the UK ! | mail:desva@slac.stanford.edu | 3767 |
\----------------------------------------------------------------------------/
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:35 MDT