Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!samsung!rex!ames!dftsrv!mimsy!dormouse.cs.umd.edu!hendler From: hendler@dormouse.cs.umd.edu (Jim Hendler) Newsgroups: comp.ai Subject: Re: AI genealogy Keywords: genealogy Message-ID: <31363@mimsy.umd.edu> Date: 13 Mar 91 04:17:38 GMT References: <2571@enuxha.eas.asu.edu> Sender: news@mimsy.umd.edu Reply-To: hendler@dormouse.cs.umd.edu (Jim Hendler) Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742 Lines: 20 I've been watching this debate develop, waiting for someone to point out one other problem with the idea, but as no one has, I'll take a stab. While I think the "parentage" idea is an interesting one, I think there's a time factor that needs to be taken into account in some important way. For example, by your genealogy I'll be a "grandchild" of an attendee of the Dartmouth Conference (MInsky -> Charniak -> me), but Charniak had been on his own for a long time before I came along. It is unclear how much effect, therefore, Minsky still had on him (and thus on me). Further, my more recent students are working on things that are more and more removed from what I did as a grad, so their separation from Charniak (and thence Minsky) will grow (in fact, most of them have never met either person). While I feel Eugene drastically effected how I view AI and etc., it's less clear how much effect that will have as time passes and my own style becomes more and more pronounced. Your advisor has done some work on hybrid systems in information retrieval, maybe you'll have to do some weight space adjusting based on time :-) -Jim H.