From owner-qed Fri Nov 18 06:24:55 1994 Received: from localhost (listserv@localhost) by antares.mcs.anl.gov (8.6.4/8.6.4) id GAA19346 for qed-out; Fri, 18 Nov 1994 06:24:46 -0600 Received: from uhcl4.cl.uh.edu (UHCL4.CL.UH.EDU [129.7.160.2]) by antares.mcs.anl.gov (8.6.4/8.6.4) with ESMTP id GAA19341 for ; Fri, 18 Nov 1994 06:24:41 -0600 Received: from UHCL4.CL.UH.EDU by UHCL4.CL.UH.EDU (PMDF V4.3-10 #8382) id <01HJM4WW9HAW8WXQL5@UHCL4.CL.UH.EDU>; Fri, 18 Nov 1994 06:26:44 -0600 (CST) Date: Fri, 18 Nov 1994 06:26:44 -0600 (CST) From: "Susan L. Gerhart, Director, UHCL/RICIS" Subject: Re: The Fermat-Wiles Theorem To: LYBRHED@delphi.com Cc: qed@mcs.anl.gov Message-id: <01HJM4WW9HAY8WXQL5@UHCL4.CL.UH.EDU> X-VMS-To: IN%"LYBRHED@delphi.com" X-VMS-Cc: IN%"qed@mcs.anl.gov",GERHART MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII Content-transfer-encoding: 7BIT Sender: owner-qed@mcs.anl.gov Precedence: bulk Lyle says "if QED is just to be an engineering tool, then it's a waste of time (from a scientific perspective)". Why? Would medical scientists say, "if image processing is just to be a tool for medical doctors to treat prostate cancer, then it's a waste of time (from a scientific perspective)"? I don't get the disconnect between science, engineering, and social/economic problems. However, industry will ALWAYS value the innovation of new functionality for business opportunities higher than full assurance but they would appreciate assistance in reducing the COST of achieving assurance. A case that combines many factors that might be considered for QED is the "feature interaction" problem that faces the telephony industry and many others. The main problem is defining interdependencies of features from the underlying primordial logic of the phone service up through the buttons on your voice mail. Bellcore research managers have characterized this problem as one of the major barriers to deploying new services -- they can't predict how features will interact (e.g. 911 and call waiting, call forwarding, 900 blocking) in the presence of different systems from different providers and the ingenuity of cheaters. The problem is one of modeling for predictability. "verification" comes later. There are scientific problems of modeling and algorithm invention to go along with engineering problems of deploying the results. Susan