00100	STATEMENT IN SUPPORT OF TENURE FOR JEROME FELDMAN
00200	
00300		I  would  like  to  strongly  support  the  Computer  Science
00400	Department's  decision  to ask for tenure for Feldman.  I should have
00500	done this long ago. Part of my hesitation was simple negligence,  but
00600	the  main  reason  why  it  is  important to keep him has become much
00700	clearer recently.  At present, Stanford is the leading place  in  the
00800	world  in  making computers carry out visual processes, and it is now
00900	clear to me that Feldman rather than merely the  environment  of  the
01000	Artificial  Intelligence  Laboratory  is mainly responsible for this.
01100	Feldman works mainly through students and does not usually  co-author
01200	the  papers  based  on  their  theses, but it is clear that the basic
01300	ideas are his.  This is not only true of the theses produced over the
01400	last  few  years,  but  also of the work that newer students are just
01500	beginning.  All this work is part of a coherent  plan  for  advancing
01600	knowledge of computer vision and applying it to robotics and applying
01700	the robotics to industrial automation.
01800	
01900		Here are some details:
02000	
02100		1. First of all, there  is  the  general  "world-model"  into
02200	which all the separate vision and manipulation packages fit.  This is
02300	almost entirely Feldman's design although it has been implemented  by
02400	various programmers.
02500	
02600		2.  Feldman  has specified and is further developing the SAIL
02700	language in which most of these programs are written.
02800	
02900		3. In my opinion the theses of Sobel, Falk, Tenenbaum,  Gill,
03000	Agin, and Paul represent substantial contributions to computer vision
03100	and manipulation and Feldman played a leading role in  all  of  them.
03200	The recent work of Agin and Paul are especially important.
03300	
03400		4. The current work of Grape and Yakimovsky and Nevatia whose
03500	theses are ready for defense will also be important.
03600	
03700		Besides this Feldman has a thorough  knowledge  of  all  work
03800	being done in the area of artificial intelligence in the country.  In
03900	my opinion, his understanding of what is being done everywhere is the
04000	best in our department.
04100	
04200		If  we  lose him, the Artificial Intelligence Laboratory will
04300	be severely handicapped for a long time.  There is  no-one  available
04400	here  or elsewhere who could replace him.  In particular, he has done
04500	much better than Raj Reddy who left us for  Carnegie-Mellon,  and  he
04600	has  done  better  than Tom Binford who works here in the same field.
04700	Feldman is simply unmatchable for getting students to do good  theses
04800	in  the  vision  area.   Only  Minsky  produces  better  students  in
04900	artificial intelligence generally.
05000	
05100	
05200					John McCarthy
05250	
05300					Professor of Computer Science
05350	
05400					Director,   Artificial   Intelligence
05500					Laboratory