kew@bigburd.UUCP (Karen Wieckert) (08/29/85)
I am posting this for Dr. Martha Palmer of Systems Development
Corporation, PO Box 517, Paoli, PA 19301:
------------------------------------------------------------------
There are two contradictory criticisms embedded in Hewitt's message.
One criticizes Prolog as a programming language, and the other
criticizes Logic. They cannot both be seen as criticisms of the
same entity. Prolog is not Logic. It is based on it, but the
programming language contains features that take it well outside
the domain of Logic, and make it equivalent to any other programming
language with the power of a Turing machine. There are problems that
are more or less amenable to being programmed in Prolog, just as there
are problems that are more or less amenable to being programmed in
Snobol or Apl or Fortran. It is possible, although it may be very
difficult, to program just about anything in Prolog that can be programmed
in Lisp. It is not in direct competition with Lisp, but is a sister
programming language that is more suitable for certain applications.
As far as the suitability of Logic for particular Artificial Intelligence
systems goes, that is a separate issue. Suppose someone defines some
"extra-logical" system that provides ready-made solutions to the difficult
problems arising in open systems (whoever said unicorns
don't exist?). It is just as possible to use Prolog
to implement that system as Lisp. It is equally possible that the solution
might be found by developing a wonderful representation framework, which
could also be implemented in Prolog. Now if the solution is a brand-new
programming language, with as yet unimagined super-powerful features, then
it probably wouldn't be implemented in Prolog. (Although some people
really like Prolog for writing compilers.)
{allegra,presby,psuvax1}!burdvax!bigburd!palmerchandra@uiucuxc.Uiuc.ARPA (09/04/85)
Yes, that was the best response I have read in reply to Hewitt's argument about Prolog vs. Lisp. Prolog is more like a package rather than a programming language. One really cannot compare the two. It is like comparing OPS-5 (a package) to Lisp... One should not forget that Hewitt is one of the originators of Prolog. The planner program he wrote long ago was the motivation for the modern prolog language we see today. If he thinks logic is not the best way to go, he probably has a very good point.