00100	       DRAFT REPORT OF CSD COMMITTEE ON FUTURE COMPUTING NEEDS
00200	
00300		Three major classes of computing needs were identified.
00400	
00500		1. On line computation including interaction of the user with
00600	his own program.  This is needed because many of the programs we want
00700	to  write  are interactive, debugging of all programs proceeds faster
00800	interactively, any kind of  computer  teaching  requires  interactive
00900	use, and we need to teach our students to write interactive programs.
01000	
01100		2. We need access in the department to all comuters on campus
01200	that our faculty and students have occasion to  use.   This  includes
01300	the   Computation   Center   Facility,  the  Artificial  Intelligence
01400	Projects's computer, SLAC's  big  computer,  and  the  IC9000  if  it
01500	materializes.   Other  systems  may  arise  on campus that CSD people
01600	should use.  Besides this, there are computers in other  places  such
01700	as  ILLIAC  4 and the proposed super-list-processing-computer that we
01800	should be able to use.
01900	
02000		3. We should go to a documentation system  in  which  reports
02100	are kept in the computer and are accessible from consoles.
02200	
02300		4.  The  systems and hardware courses require access to small
02400	computers that can be used on a hands-on basis.
02500	
02600	
02700		In  order  to fulfill these needs, the following measures are
02800	proposed:
02900	
03000		1.  The  Computation  Center  be  urged  to  go  to  a  fully
03100	interactive  time-sharing system.  It should be further urged to take
03200	measures to reduce the cost of interactive use and to support a  wide
03300	variety  of  terminals  of  both  display  and  hard-copy types.  The
03400	present system permits interaction with only a  few  special  purpose
03500	languages,  and programs in these languages are limited to small size
03600	because  they  are  supported  by  partitions  of  the  core  memory.
03700	Moreover,  the  cost  of interactive computation is excessive because
03800	the IBM 2741 is very  expensive,  unnecessarily  expensive  telephone
03900	connections are used, and the IBM 2703 concentrator is too expensive.
04000	
04100		We have investigated possible solutions to these problems and
04200	have concluded that  adoption  of  the  University  of  Michigan  MTS
04300	time-sharing  systems  would  solve  many of these problems and would
04400	make possible the solution of the others.  As far  as  we  have  been
04500	able  to  determine,  MTS  has  all the virtues of our present system
04600	except that the editor is not as good, plus the  advantages  of  user
04700	interaction  with his programs and full use of paging so that program
04800	size is not limited by partitioning.  Our editor WYLBUR  can  be  put
04900	into  MTS.   The only major problem is that of making the transition.
05000	We believe it should be faced.
05100	
05200		It has been suggested that the users should  be  surveyed  to
05300	determine  what  they  want.   The  average opinion of the users will
05400	surely favor small improvements in what they have  already.  This  is
05500	natural  because  they  are  not  computer  scientists  and  are  not
05600	inventive in this area.  Fortran and  Algol  did  not  arise  out  of
05700	surveys of the user populations that were then programmming in octal.
05800	
05900		2.  The Computer Science Department should consider trying to
06000	acquire a terminal support computer system  and  a  large  number  of
06100	terminals.   This  system  should  provide  CSD  users  with  display
06200	terminals in their offices and the ability to use  through  them  any
06300	computer on or off campus that will permit the use.
06400	
06500		The  investigation of the possibilities for a terminal system
06600	is not complete enough to make  a  concrete  recommendation,  but  it
06700	looks  like  a suitable system could be obtained for between $50K and
06800	$100K.