COMMENT ⊗   VALID 00003 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	%prejud[f89,jmc]		Prejudices about Stanford economization
C00010 00003	\smallskip\centerline{Copyright \copyright\ 1989\ by John McCarthy}
C00011 ENDMK
C⊗;
%prejud[f89,jmc]		Prejudices about Stanford economization
\input memo.tex[let,jmc]
\title{Thoughts on Economizing at Stanford}

	These opinions (or prejudices) are based only on having
spent 27 years as a faculty member, having founded and directed
the Artificial Intelligence Laboratory, and on having founded and
directed the LOTS low overhead time-sharing system.  They are not
based on a study of Stanford's current administrative costs.
I'm willing to learn more.  However, I share the opinion that
Stanford has allowed administrative costs to grow excessively over
the last 25 years.

1. People in service activities can always think of more services
they could perform with increased resources.  The tendency is to
expand until stopped by external authority.

2. For this reason, when possible, services should be
decentralized to the point where the user of the service can
decide whether to expand his use of this service or put his
resources to some other use.  This often results in greatly
reduced use of services.  Under present institutional conditions,
the saving from decentralization usually outweighs economies of
scale of centralized facilities.

  This is the main advantage of small
computers, and the people who operate small computers typically
spend less on support than the larger centers.

3. Stanford should compare the number of people it employs to
perform each service - purchasing, personnel, computing, registration
and student records, maintenance - both with Stanford's historical
number of people involved in the activity and with the corresponding
activities in other institutions.  The idea is to find which activities
have grown the most at Stanford and how economically the most
efficient other universities do them.

4. Caltech charges 56 percent overhead, Stanford 79 percent and
rising.

5. When stamps were three cents, Stanford put a 10 percent surcharge
on postage to do the weighing and put on the postage.  Now that
stamps are 25 cents, Stanford needs 15 percent.  This suggests
that Stanford has become inefficient 50 percent faster than the
U.S. Post Office.  It's plausible, because there aren't any
continuously active vociferous groups at Stanford demanding economy,
and there isn't any competition.

6. Outside consultants should be employed to survey each activity
that has substantially expanded its personnel since the 1950s.
These consultants should report to a level of administration
committed to reduction of administrative costs---if necessary
to the Board of Trustees.

7. As in most organizations, computers and automation have been
used to make the organization more comfortable internally rather
than to serve its users better.  If a suitable system can be
purchased, Stanford should allow purchase requisitions to be
entered and their progress tracked from computer terminals and
personal computers.  If this were done, the number of people
in the purchasing department might be reduced.  A study might
establish this.

8. Appointment and promotion of research personnel should be
decentralized.  The second guessing by personnel and by the
Provost's office is wasteful and harmful.

9. Committees should be replaced as much as possible by giving
their responsibility to single individuals.  They are often
formed as a sop to group interest, but they are much too expensive.

10. Computers are an activity particularly prone to overgrowth of
personnel.

	a. When LOTS was started 4 FTEs served 3,000 users.
Nevertheless, the personnel budget approximately equalled the
amortization of the computers.  Now AIR has a budget of \$5
million of which only \$750,000 amortizes the computers.

	b. I once investigated how many people it took to operate
a D.E.C. System 10 computer.  The number varied between 1 and 15
depending on institutional traditions and practices.

	c. Stanford has engaged in large scale pump priming in
getting faculty and students to use computers.  This should be
phased out on a definite schedule---say three years.  After that
proposals for help in using computers to develop courses, etc.
should compete with other proposals to use money to develop
courses.

	d. The Computer Science Department examined its computer
facilities.  The conclusion was that the most redundant person
was the Manager, even though he was doing a reasonable job.

	e. Stanford especially needs an outside consultant in
economizing on expenditure for computer personnel.

11. Economy will require the abandonment of many worthwhile activities
that aren't as worthwhile as saving the money.
\smallskip\centerline{Copyright \copyright\ 1989\ by John McCarthy}
\smallskip\noindent{\fiverm This draft of PREJUD[F89,JMC]
\ TEXed on \jmcdate\ at \theTime}
%File originated on 21-Nov-89
\vfill\eject\end