Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83 (MC840302); site ttds.UUCP Path: utzoo!watmath!clyde!burl!ulysses!unc!mcnc!decvax!ucbvax!ucdavis!lll-crg!seismo!mcvax!enea!ttds!alf From: alf@ttds.UUCP (Alf Thomas Sj|land) Newsgroups: net.ai Subject: Comments to FOOLOG and Hewitt vs. Logic. Message-ID: <1027@ttds.UUCP> Date: Fri, 13-Sep-85 20:05:24 EDT Article-I.D.: ttds.1027 Posted: Fri Sep 13 20:05:24 1985 Date-Received: Wed, 18-Sep-85 02:47:27 EDT Reply-To: alf@ttds.UUCP (Alf Thomas Sj|land) Organization: The Royal Inst. of Techn., Stockholm Lines: 54 About FOOLOG: It is correct that Martin Nilsson made FOOLOG, "the world's smallest prolog", in the way described in an earlier message. The authors of LM-Prolog are Mats Carlsson and Ken Kahn. Since all three of them were at UPMAIL in Uppsala, Sweden, during the period of development the confusion is understandable. Martin used to have the FOOLOG-system printed with the smallest fonts of their laserprinter on the back side of his card ! About Hewitt's anti-prolog (anti-logic ?) opinions: It seems to me that two points of view are mixed here. One is the philosophical concerning the relative appropriateness of different AI-programming paradigms particularly in an open systems environment. The other is the issue of what is the best programming language for future AI-programming environments: Lisp or Prolog. Some questions occur: 1) How can you at all "make inferences" without having implemented some logic in your system in order to reason ? Can an actor system based expert system in Common Lisp be said to "draw conclusions" from its "knowledge" if it does not implement an inference engine ? 2) Would you (Hewitt) say that the programming language C is more powerful than Lisp just because nobody would get the idea of implementing a C-interpreter in Lisp ? (exchange C for Lisp, Lisp for Prolog). To me they are just formalisms (languages) on different levels of abstraction. 3) Perhaps the critical magic words in Hewitt's criticism are "foundation of AI". Is this phrase properly defined ? 4) Isn't it enough to keep the database static while you perform a deduction in order to be able to use logic in an AI-system ? Would you mean (like some mysticists) that no deductions are possible or do you just mean that logic is an inappropriate tool for AI-systems ? In either case, what is your alternative approach to the notion of a "reasoning system" ? My view is that of communicating inference engines, where the communication aspect is separated from the deductive. In the communication aspect I tend to include altering the database, thereby avoiding non-monotonic logic. Thomas Sjoeland, i.e. alf@ttds.UUCP -- Brought to you by Super Global Mega Corp .com