COMMENT ⊗   VALID 00004 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BASL30\M1BASI30\M2NGR40\M3NGR25\M4NGR20\MFSTA200\
C00006 00003	
C00011 00004	
C00014 ENDMK
C⊗;
\\M0BASL30;\M1BASI30;\M2NGR40;\M3NGR25;\M4NGR20;\MFSTA200;\;
\'3;↓↓\FFS\F2
\'3;↓Q\CSTANFORD UNIVERSITY
\F3\CSTANFORD, CALIFORNIA 94305
\F4



ARTIFICIAL INTELLIGENCE LABORATORY\←L\-R\/'7;\+R\→.\→S   Telephone:
COMPUTER SCIENCE DEPARTMENT\←S\→.415-497-4430
\F0\C24 October 1974









Richard Atkinson
Associate Dean
Humanities and Science
Stanford University
Stanford, California 94305

Dear Dick:

\J	Enclosed is the copy you requested of my 1971 letter to Lyman.
From what you told me today about the potential impact of the budget
cuts on the instructional program, the issues raised in that letter
are much more urgent now than they were at that time.

	The state of my information (or anyone else's) about the costs
of computing is much worse now than it was then.  At that time,
administrative computing was a separate organization and its costs could
be estimated.  The merger of all computing activities into SCIP has made
it very difficult to allocate costs properly.  However, I think that the
potential savings from reorganizing Stanford computing are much larger now
than they were at that time.

	My guesstimate of the potential savings is as follows:

	1. If the 360/67 is considered fully amortized, its rates can
be considerably reduced, and the Provost's fund subsidy to academic
computing can be cut in half with no reduction in the amount of computing
available to students and faculty.  In fact, except for government
accounting considerations, the fund could probably be completely eliminated.
A renewed attempt to get the government accountants to look at the
matter sympathetically would be justified in view of the financial state
of the university.  This cost reduction would require

		a. No longer charging academic computing with the costs
of extending WYLBUR etc. to the 370.  At present these costs are considered
general costs of SCIP and are allocated in proportion to revenues.  At
present the 67 accounts for $2.6 million out of $4.1 million in revenues
and so is charged 70 percent of an expense that leads to little if any
improvement in the facilities available on the 67.  It will also be necessary
to reduce the free courses and the consulting.  \F1A target personnel figure
should be the number of people employed when the IBM 7090 was the main
campus computer\F0.  The estimate of possible cost reduction is based on
the fact the 67 is paid for.\.






\J	2. Judging from what other universities were paying in 1971, I think
the business computing of the university can be done for $500,000 per year.
Serious consideration should be given to contracting all or most of it
to outside organizations, e.g. Bank of America should be asked for a quote
on the payroll.

	Here are some random facts in support of the idea that large savings
are to be made:

	1. SCIP has 300 employees, i.e. is one third the size of the faculty.
The growth of this number with time should be examined.

	2. Some years ago, SCIP quoted the psychiatric clinic a price of
$90,000 per year for patient billing and the clinic got an outside quote
of $15,000.  I believe the matter was somehow compromised and SCIP's
monopoly was preserved, but I suppose this was done by putting more
of the burden on an organization that could not make a comparison of
outside costs.

	3. SCIP quoted $21,000 per semester to run CS206 on an interactive
basis not including the cost of the terminals to be used.  IMSS quoted $4500
for the same service including the use of the terminals.

	4. A (chemistry?) professor surveyed costs of large scale computing
and discovered he could do it for 1/10 the price outside Stanford.

	There are many more examples of this kind, but I'll stop, because
the probability of having any influence is too low to justify more effort
even to the extent of writing this letter more carefully.

	Here is a conjecture on why this growth has occurred, but I don't
know the internal situation well enough to be sure of it.  SCIP has undertaken
many projects, SPIRES, BALLOTS, WYLBUR, ORVYL, OASIS.  The ideas for most of
these projects were generated internally, but most of the costs have at least
nominally come from the outside.  I say nominally, because it is often
arbitrary whether a system cost that makes a project possible should be
charged to the project or considered the cost of a general improvement.
An organization like SCIP that is not subject to overall  cost control
tends to bid low to a potential sponsor of a new activity.  An extreme
example is OASIS which arose through a $200,000 Ford grant many years
ago and whose development cost many times that before it was abandoned.

	The amortization of the 370/158 over ten years is questionable
and mortgages the future of Stanford computing for a very long time.

	A detailed cost analysis would be needed to determine the true
allocation of costs and the very best course to follow.  However, in
this time of extreme budget stringency, a cruder approach should also
be considered.  This is to bring in a new manager without commitments
to present personnel and require of him a specific cut in personnel
an operating costs.  It would probably also be wise to separate academic
computing from administrative and ask the faculty to allocate cuts between
computing and other parts of the academic budget.\.





\J	This letter which you requested is absolutely the last comment
I shall make on the costs of Stanford computing unless more is solicited
by Lyman or Miller.  I have been interested in Stanford computing for
a number of years and have taken part in various low level committees
with negligible result for the work put in.  My conclusion is that advice
is \F1not wanted\F0.  My only reason for acceding to your request this time
is that you may be able to avert some cuts in Humanities and Science by
pointing to one of the administrative areas that can be cut.
I have totally given up trying to get Stanford to make interactive computing
available to the campus generally or to get a good display system for the
Computer Science Department.

	I hope this is of some use.\.

\←L\→S\←R\-L\/'2;\+L\→L
Sincerely yours,



John McCarthy
Director, Artificial Intelligence Laboratory
Computer Science Department
\←S\→L
JMC:pw
atkins.le3[let,jmc]:SU-AI