Re: JP Barlow, Ph.D, Social Engineering

Dr. Rich Artym (rartym@galacta.demon.co.uk)
Sat, 21 Sep 1996 13:30:02 +0100


In message <Pine.SOL.3.91.960919151057.11096F-100000@sun2>,
Eugene Leitl writes:

> Computer science as a whole is not a science. It's mostly engineering.
>
> Only a tiny part of it, e.g. people involved with algorithm proving,
> complexity theory, etc. can be considered "true" scientists. This
> computer science is indistiguishable from maths.

Eugene, there are a few parts of computer science that *are* real science,
ie. they employ the scientific method properly. It's those parts of the
discipline that create complex emergent systems that are *way* beyond full
human comprehension, and that then measure them much like one would measure
the world outside in order to substantiate theories by not disproving their
hypotheses.

Operating systems clearly fall into this category, as do most large
networks and control systems. Here there is usually an element of
interaction with the external world, which tends to increase the overall
complexity of systems because of non-determinacy and real-time constraints.

There are also systems studied by computer science that are entirely self-
contained yet also result in emergent complexity that needs to be studied
in the same way as would a physical system. In this type, there is usually
an evolutionary element, such as in genetic algorithms, artificial life,
neural networks (when learning from a database), and so on. Sometimes the
systems are not analytic *in practice* because the space of discourse is
too big, and sometimes the reason is that evolution is non-determistic
because of imprecision in computations or by design through concurrency.

Either way, the only way you have of knowing what's going on is observation.
I did my research in massive parallelism, a decade ago, and it became mind-
blowingly obvious (I was quite shaken to discover) that systems of thousands
of independent but interacting concurrent processes have a mind of their own
that is quite unpredictable ... In this situation, you have to employ the
usual procedures of science if you are to get anywhere with any degree of
certainty that you are on the right track and not just hand-waving. That
kind of computer science cannot be done as if it were purely mathematics,
even though in many other parts of the discipline you are entirely right.

Rich.

-- 
###########  Dr. Rich Artym  ================  PGP public key available
# galacta #  Email   : rich@galacta.demon.co.uk         158.152.156.137
# ->demon #  Web     : http://www.galacta.demon.co.uk   194.222.245.150
# ->ampr  #  AMPR    : rich@g7exm[.uk].ampr.org 44.131.164.1 BBS:GB7MSW
# ->NTS   #  Fun     : Unix, X, TCP/IP, kernel, O-O, C++, SoftEng, Nano
###########  More fun: Regional IP Coordinator Hertfordshire + N.London