IBM system dying
of financial anemia
at Medical Center

By TIM COBURN

The inter-office memo contained phrases like “had
been studied and found to be unsound financially,” “the
sum of resources is not adequate,” and finally
“termination of ACME service and recommendation that
users move their computing services.” Ron Jamtgaard
doing his job. Business as usual with everything
proceeding in an orderly fashion. Even in death there is a
form to follow. In academia the words are important,
and the subject is treated objectively.

ACME, the Advanced Computer for Medical Research
is dying. That’s what the words really mean. The staff
already knew and have been hustling new jobs for a
month or more. Many of the users had been warned not
to make a major investment in the present system. The
aroma of terminal illness, not entirely foreign to the
hospital, is clearly discernible—only this time it is hitting
one of their own.

Computers, or more’ properly, computer systems
don’t often die. They are replaced with newer, more
expensive, more versatile generations of computers. Each
new one is to be the ultimate computer, the computer
that will do everybody’s job. Once | thought ACME
would be like that. Back in 1966 anything seemed
possible.

The problem is that ACME can’t seem to pay for
itself. It has received some three million dollars in direct
support from NIH and thousands more from various
grants and contracts. Now the support is dwindling and
by the end of June will be shut off. .

The glass cage, lit up 24 hours a day for the last six
years, will be dark. The flashing console lights will no
longer blink at me as | wander by the operators desk.
The WAITING FOR YOU signal on the terminal,
designed in a rare moment of consensus by a tall blond
German and a continuously jabbering Frenchman, will
lie peacetully jn some storage room with loose wires
dripping like neurons from their necks. .

ACME was born midst all the excitement of the
middle 1960’s. ! remember the first meeting of
Prospective users. It could have been the first anti-war

“ protest’ of the medical school community. The large
lecture hall was filled with faculty members, technicians,
an assortment of students, and of course the salesmen
from IBM with their shiny suits and briefcases. Nobody
else carries a briefcase around the medical school. It’s
not part of the form.

Academia likes questionnaires and sure enough there
was to be a questionnaire, and a committee to examine
ail the possible hardware and software systems, and
hiring a director, submitting grant proposals, and
timesharing, on-line data collection, an interactive
system with hundreds of fittle fingers stretcning into
every cranny of themedical school. And then there was
PL-1, a language anybody could use and understand.
Smail computers were out, irrelevant. Walking across
the field to throw a deck of cards into the 7090 hopper
was inefficient. Stanford would have ACME, and !BM
could do it all with their new equipment and extensive
experience. The technological millenium had come.

Behind the scenes the birth was not so easy. | only
heard the rumors. Lederberg wanted his own system.
The comp center, Fisher, Miller et. al. wanted to have
one giant system for the whole campus, or at worst to
contro! ACME, and IBM was worried that ACME might
commit the heresy of buying a brand-x computer. So
they promised the moon.

The computer freaks were skeptical, but they had
only a few years of experience under their belts. Their
reservations were mere impediments to this magical new
system. This was the big time with big money. You go
with the pfoven winners. ACME went with IBM and Ed
Feigenbaum. The former was wealthy and the latter had
written a book on computer theory. The computation
center, SLAC, the hospital administration, and the
university business office also went with !BM.
Coincidentally, IBM gave Stanford a million dollar grant
for communications research involving automated
retrieval and storage of information.

The development phase for ACME was a prolonged
series of disasters. The first disaster was the “pie file,” a
magnetic storage device designed to handle millions of
bytes of data. Lederberg doubted that there was that
much useful information in the whole medical school.
Nobody else had such a device. It was the prime reason
for getting an IBM system. ,

'Certainiy the pie file was one of the mechanical
marvels of the decade. | used to sit and watch the strips
of tape get “picked” off their hangers, transferred to the
reading rollers, and instantaneously speed’ around the
head assembly. The. rhythmic clinkety-clank of an
automatic bottling machine added to the stark beauty of
this marvelous machine. Little children would stand
awed by the scene.

But the IBM service men were less than awed. They
were completely frustrated by their daily attempts to
keep the device running. Recently, | ran across one of
those gentlemen in the mountains back of Santa Cruz.
Although bald on top he had very long hair and a beard.
He was sitting yoga style facing the ocean, slowly
shaking his head back and forth. | was tempted to sneak
up beside him and whisper “‘clinkety-clack”’ in his ear,
but it seemed unkind. He hac suffered enough. -

The pie file was just the beginning. Core problems,
balky interfaces, unworkable or incomprehensible IBM
software followed one after another with monotonous
regularity. The summer of ‘68 had 75% down time.

“°F ‘particulatly Yémember the software problem. The
timesharing software system was the second major
reason for investing in a business machine in a scientific
environment. It wasn’t ready when the equipment was
delivered of course. Well, that was understandable. Then
it was delayed again after six months, and again, and
again. Finally, it wasn't going to make the scene at all.
So the staff, Gio Wiederhold, Gary Breitbard, Jerry
Miller and many others, wrote their own. It wasn’t bad-
Gio had brought his own version of a timesharing PL-1.
“system from Berkeley where he had worked with
Feigenbaum. Whether its implementation should have
been charged to a medical center facility is another
question. One of the more serious problems in writing
your own system has just recently surfaced, however.
PL-ACME is not compatible with other PL-1 systems.
The conversion costs are high. But then compatibility
seems mostly imaginary between any system.

ACME was not a total failure. Even f, one of its more
persistent critics would point to its educational value as
a major accomplishment. Both users and staff developed
a considerable understanding of the problems in medical
computation. They learned what things could be done
easily and which problems were expensive. And they
began to understand how difficult it is to define
rigorously the type of problems that are most
interesting. Perhaps this knowledge will be valuable in
future systems. Currently the plans fora joint, facility
between the hospita! administrative office and some
limited PL-ACME service are in development. | suspect
‘tthe users will be much more realistic in their
expectations and requirements.

Who then will mourn ACME? A noble experiment? A
step in the direction of future systems? A compromise.
between the possible and practicable? | guess | will be
sad to see it go. It had a certain childlike appeal to it. It
wasn't businesstike. !t wasn’t cold and impersonal and
invaribly correct. Its traumas were very down to earth.
And certainly it wasn’t ACME’s fault that it was never
allowed to be.an Advanced Computer but only a
converted accounting machine.

Built and owned by IBM, it will be returned to IBM.
Its transistorized heart will be stopped—in spite of
Norman Shumway—and its faceless cabinets will be
packed up and carted off to some never-never land for
technological remnants. Perhaps, fll actually see it go,
without even a whimper. “Here lies ACME, born
1966—died 1973. Economic and political’ requirements
were not compatible with life. There are no survivors.”

page three