00100	Computer Science Department
00200	
00300	February 20, 1971
00400	
00500	
00600	To: Computer Facilities Committee
00700	
00800	From: John McCarthy, Professor of Computer Science
00900	
01000	Subject:  The  report "Computing at Stanford, Report of the Long Term
01100	Planning Project"
01200	
01300	
01400		In  my  opinion, the report is seriously deficient in that it
01410	does  not  make  a  plan  for  the  conversion  of  Stanford's  batch
01420	processing use of computers to time-sharing.  The simple expansion of
01430	the numbers of terminals attached to existing systems, in  particular
01440	the  Campus  Facility,  will  not  make  much  improve-  ment, if the
01450	facility continues to be  operated  in  the  same  way.  M.I.T.,  the
01460	University  of  Michigan,  and Dartmouth College already have general
01470	campus facilities in operation in advance of anything even  envisaged
01480	for  the future in the report.  If the general facilities of Stanford
01490	evolve along the lines of  the  report,  then  the  Computer  Science
01500	Department  should  plan  to ask for general funds to make a separate
01510	system for teaching its undergraduate and graduate  courses  and  for
01520	the  research needs of its students and faculty.   Otherwise teaching
01530	at Stanford will fall behind that of other schools.
01540	
01550		In my opinion, the reason the report turned put as it did, is
01560	its  reliance on the survey method of estimating needs. It is just as
01570	unreasonable to expect an average computer user  to  propose  systems
01580	outside  his  immediate  experience  as  to expect the need for steam
01590	shovels to become apparent from opinion polls of ditch diggers.
01600	
01610		I would particularly like to take issue with Bruce Lusignan's
01620	prediction  that  a  system  of mini-7090s will come to dominate.  He
01630	underestimates, no  doubt  from  experience  with  Stanford  computer
01640	facilities, the services that modern on-line systems can provide.  He
01650	is also  mistaken  about  the  cost  of  terminal  and  communication
01660	facilities;  they  are also coming down since they too are electronic
01670	and bare wires rent from the telephone company  at  about  $3.00  per
01680	mile  per  month.  Although  he  mentions  the use of data bases as a
01690	reason for central systems he  again  under-estimates  the  size  and
01700	immediacy  of  the  issue.   For  example, ARPA is installing at Ames
01710	Research Center a trillion bit file which will be available  to  many
01720	Stanford  users.   The  capacity of the file is equivalent to 250,000
01730	books at a storage cost per book of $4.00 which is less than that  of
01740	shelf space in the projected new library.
01750	
01760		Another issue raised by the report is the $1,000,000 per year
01770	cost  of  administrative  data  processing  coming  entirely  out  of
01780	Stanford   general  fund  and  the  expectation  of  increasing  this
01790	expenditure at 6% to 10% per year.  It is difficult to  believe  that
01800	Stanford  gets  its  money's worth out of this expense although it is
01810	treated as a sacred cow in the report.  I could not be  sure  without
01820	detailed investigation, but it seems that an empire has arisen within
01830	the  Stanford  administration  that  would  not  stand  the  detailed
01840	scrutiny  that a proposal to spend $1,000,000 per year out of general
01850	funds for an academic purpose would get.
01860	
01870		If the Computer Facilities Committee is interested I would be
01880	glad  to explain my reasons for believing that Stanford should plan a
01890	transition to  time-sharing  since  I  have  found  talking  to  Dave
01900	Phillips and Bill Miller about this like shouting down a well.
01910	
01920		Frankly,  I  do not expect to be able to change the projected
01930	course of action, but only, through raising doubts, to cause Stanford
01940	to  waste  less  money on it.  Eventually, IBM will stop falling over
01950	its feet, make a proper time sharing system, and  then  the  salesmen
01960	will convince Stanford to do the right thing.