perm filename SCIP[CUR,JMC] blob
sn#150743 filedate 1975-03-19 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 Comments on "A Preliminary analysis of present computing
C00007 ENDMK
Cā;
Comments on "A Preliminary analysis of present computing
alternatives for Stanford University" by Dickens, Franklin, Roberts,
and Williams
by John McCarthy
p.1. "Nowhere was the growth greater than in the area of administrative
computing". It is not indicated whether this is regarded as good, or as
pathological as I regard it.
BALLOTS, SPIRES, and medical services are areas in which optimism about
what others are willing to pay to use Stanford facilities has led to mistakes
and is apparently continuing to do so.
"The trend of this policy is toward an open market ...". You should advocate
this trend or deplore it since you are the decision makers here.
A computer for instructional purposes not operated by SCIP might avoid some
of the "most favored customer" problems.
You should consider PDP-10 alternatives other than the KL-10. A KL-10 would
service many more than 32 terminals, and a second-hand KA-10 system would
service 32 just fine and might be obtainable at a much lower cost.
If the proposed communications network can meet both the technical specifications
and the cost goals (25 cents per connect hour) mentioned, then I think it
is a good enough idea, although there may be some inadequacy I haven't
noticed. It now seems to me that the AI Lab would want to be connected
to it.
The statement that the 158 is necessary for the administrative work alone
of the University and the Hospital is dubious. In my opinion it reflects
the inefficient and overblown way this work is done. When I made my
survey several years ago, Stanford had the largest adminstrative computing
costs per student or dollar of operating budget of any school, and that was
with just the 145. AT LEAST SOME SUBSTANTIATION BY COMPARISON WITH OTHER
SCHOOLS SHOULD BE MADE. It would be better to do the administrative work
outside and keep the 158 for academic work. Banks and computer service
bureaus do administrative computing at competitive prices. Try the payroll.
Of the five alternatives offered for "Low overhead time sharing computing",
the multi-mini and the dedicated APL are unacceptable to the Computer Science
Department. The multi-mini doesn't allow for large programs and would require
too much language development, and the Department doesn't especially like
APL and would not want to be tied to any single language.
If I understand the numbers quoted for costs, the manpower costs are
overstated in SCIP's inimitable way and the hardware costs of the KL-10
alternative are, alas, understated.