COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	THE CASE FOR A STANFORD PDP-10 TIME-SHARING SYSTEM
C00010 ENDMK
C⊗;
THE CASE FOR A STANFORD PDP-10 TIME-SHARING SYSTEM


	The subcommittee on research and  instruction of the Stanford
Computer  Facilities Committee proposes that  the Stanford Center for
Information Processing acquire a  PDP-10 computer system and  operate
it  to provide  time-sharing services  to the  Stanford community  in
conjunction with the IBM 370 computer system it is operating.

	It  is proposed that  each computer system  be operated using
the most  advanced system software  available from its  manufacturer.
The two  computers should  share an IBM  3330 disk system  and should
communicate through this system.

	The proposal is based on the following considerations:

	1. The main direction in which Stanford needs to improve  its
computer facilities  is in  making possible  more interactive use  of
the computer.   This is hindered at present by (a) poor and expensive
terminals, (b)  very  high  priced  disk storage,  (c)  an  operating
system that  provides few  and awkward  interactive facilities.   The
main  technical flaws of the present system  arise from the fact that
it provides interactive facilities in a basically batch environment.

	2. An  economical interactive system  will make possible  the
following  new  services: (a)  interactive  use  of  the computer  by
students in  computing classes  so that  they will  learn more  in  a
given time. (b) interactive  use by faculty and graduate  students in
research  which   will  permit  more  sophisticated  programs  to  be
debugged with  given  effort  (c)  direct use  of  the  computer  for
teaching non-computer  subjects (d)  the use of  interactive services
such  as editors and other aids  to document preparation, the keeping
of reports and working  papers in computer files for  more convenient
use  and for  greatly  increased possibilities  of  collaboration (f)
extension of such collaboration via  the ARPA net and its  commercial
successors to  collaboration with  other universities and  scientific
institutions.

	3.  Stanford  has a  heavy  commitment to  IBM  software, and
besides that,  IBM users  are the  largest group,  and Stanford  will
want to use new programs  and languages for IBM computers originating
both from IBM and other users.

	4.  IBM is not the  leader in interactive  software.  This is
because they have  only recently had  memory relocation, and  because
they provide  such interactive facilities  as they do  in the context
of a system  that is basically  batch.  They  have some  time-sharing
systems such  as CPCMS, but  they are inferior  to D.E.C.'s  and have
only  limited  compatibility  with the  rest  of  IBM  software.   In
particular, it would be difficult  to convert the services that  SCIP
is already committed to supply  the Medical School and Administrative
Computing  to such a system. Moreover, IBM  has no development in the
works  that  will  provide  high  quality   interactive  services  at
reasonable cost.  The appendix contains an IBM  article describing an
interactive system and some comments  aimed at showing that  Stanford
can do better.

	The importance of  interactive computing and the  benefits of
the  above mentioned  services can be  verified by  consulting people
who have had the opportunity to  use them and compare them with  what
is now  available.   They will  not be  found by  a market  survey of
present  users any more  than the  need for steam  shovels would have
been found by a market survey of ditch diggers.

	5. The largest  community of interactive users  is associated
with the  D.E.C. PDP-10.   It comprises  more than 200  systems using
two substantially  compatible operating  systems, TOPS-10  and  TENEX
which will probably be merged in the next few years.

	6. Stanford  presently operates two  PDP-10 systems -  at the
Artificial   Intelligence  Laboratory   and  at  the   Institute  for
Mathematics in  the Social Sciences.   A  third system  SOMEX in  the
Medical  School will  probably  be  funded by  N.I.H.   Both  systems
provide  interactive services superior  to those provided  by the IBM
67 as many users will testify.

	7. A  survey  reported in  the  August Datamation  rates  DEC
highest in cost-effectiveness and in "most ethical".

	
	The  attached  budget  gives  an  estimate  of  the  cost  of
acquiring and using  a PDP-10 system.   It is  an incremental  budget
based on adding a PDP-10 in the same room  as an existing IBM 370 and
under  the  present SCIP  administration.   If  additional  costs are
found, the prices  can be  raised somewhat.   In our  opinion $1  per
hour is  a critical level  for connect time  in the sense  that going
above  it will seriously limit the systems  utility for students.  In
the other  direction, not  much would be  gained by  going below  the
$.40 per hour proposed for off-peak use.