Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!gatech!bloom-beacon!husc6!cmcl2!phri!bc-cis!pluto!dasys1!hoptoad!ptsfa!ames!ucbcad!ucbvax!decvax!decwrl!labrea!glacier!jbn From: jbn@glacier.STANFORD.EDU (John B. Nagle) Newsgroups: sci.philosophy.tech Subject: Re: Definition of science and of scientific method. Message-ID: <17133@glacier.STANFORD.EDU.R> Date: Fri, 17-Jul-87 14:31:10 EDT Article-I.D.: glacier.17133 Posted: Fri Jul 17 14:31:10 1987 Date-Received: Sat, 25-Jul-87 06:35:18 EDT References: <6693@allegra.UUCP> <1664@tekcrl.TEK.COM> <1084@aecom.YU.EDU> <813@klipper.cs.vu.nl> Organization: Stanford University Lines: 19 Summary: Research in planning seems to behave like a science In "Planning for Conjunctive Goals" (Artificial Intelligence, vol 32, 1987, p. 333-377), David Chapman, while discussing the history of conjunctive planning research, writes "The three main points of this section are that in retrospect all domain-independent conjunctive planners work the same way; that the action representation which they depend on is inadequate for real-world planning, and that desirable extension to this action representation make planning exponentially harder. It is much longer than such sections are in typical AI papers because domain-independent conjunction planning is unusual as a subfield of AI in showing a clear line of researchers duplicating and building on each other's work. Science is supposed to be like that, but for the most part AI hasn't been." I agree. John Nagle