TO

FROM :

SUBJECT:

Ln. AE OERG ERC

January 11, 1974

SUMEX Executive Committee File

T. Rindfleisch LCR

Action Items and Notes From the Afternoon Session of the
Second SUMEX Executive Committee

Meeting - January 9, 1974

Attendees: (See attached agenda A)
Morning : J. Lederberg, S. Amarel, W. Baker and C. Brewer
Afternoon: As above plus T. Rindfleisch and E. Levinthal

We discussed the machine configuration and current delivery
schedule (see attachments B and C). A change in the configuration
was proposed as in diagram "B" to ensure that the machine be made
operational as early as possible. This consists of hooking the
system to ARPANET through the standard BB&N TIP/PDP-10 inter-

face initially and interposing the PDP-1l at a later time. The
long term attractiveness of the PDP-1l remains in the ability to
off load the TELNET protocol from the PDP-10 and perhaps to provide
alternative file system access when the SUMEX machine is down. No
objections to this approach were raised.

The hardware delivery schedule was presented identifying the two
current problem areas: 1) The proposed slip in line printer delivery
until March 31 by Data Products and 2) The as yet tentative TIP
delivery schedule. Dr. Baker offered to assist in accelerating
delivery. This was deferred pending the outcome of on-going vendor
negotiations to solve the problems.

Dr. Amarel described the PDP-10 facility they plan to implement at
Rutgers saying they were having problems locating the necessary
expertise to mount the TENEX software system. We have the same
problem and are trying to sélve it through collaboration with local
TENEX groups (IMSSS, SRI, AI Lab., etc.). We offered to assist
Rutgers as best we can (e.g. having one of their people come to
Stanford to learn the system) to develop the necessary capabilities.
Memo to SUMEX Executive Committee File
January 11, 1974

Page 2

We discussed the general question of file security provisions in
TENEX (see pp 30 ~ 45 of the "TENEX EXECUTIVE MANUAL"). There
was some concern over the conflicting needs of confidentiality
for some administrative and patient medical data as opposed to
the general desire for an open environment condusive to sharing
both internally and externally. We agreed to make a presentation
at the next meeting of the available file security system to
further evaluate the need for selective encrypting of data.

We discussed the preliminary announcement of the SUMEX resource
and approved the news release in attachment D. There was general
agreement on a two~phased announcement process; an initial general
release (press, ARPANET News, etc.) and a more detailed solicita-
tion of participants after the system is established and the
Advisory Committee set up.

The immediate need for a more detailed write-up of SUMEX objectives
and capabilities was emphasized to be available to BRB, Stanford,
Rutgers, etc. to support informal discussions with interested
people about the project. We said we would generate such a docu-
ment in several weeks.

Dr. Baker requested a "Mission Statement" for the Stanford half

of the project (SUMEX) describing its research objectives, organi-
zation, and capabilities. This is needed by BRB for administrative
reasons. Dr. Lederberg pointed to some general quidelines he was
formulating to screen potential SUMS users such as inability to
obtain necessary computing support from other Stanford sources
(SCIP), and relevance of the research to Al. These will evolve as
potential users become more aware of the facility and its goals.
Dr. Lederberg agreed to formulate an initial SUMEX mission state-
ment.

The problem of terminology was discussed from the viewpoints of
overall organization and operational access through the ARPANET.
There was general agreement that Dr. Lederberg's descriptors,
while long, were appropriate:

Overall Facility = National Shared Computer Resource for
Health Research at Stanford (NS CRHR-SU)

Half of the facility is allocated to "SUMEX" (Stanford
Medical School users)

Half of the facility is allocated to the "National Program
for Artificial Intelligence in Medicine" (NPAIM)

The ARPANET designation for the NSCRHR<SU will be something like
"SU-MED" or "SU-MEDAI" or "SU-AIM".
Memo to SUMEX Executive Committee File
January 11, 1974

Page 3

10.

The status of interactions with ARPA relative to NIH use of the
network was discussed. In general, no problem is apparent - all
contacts have indicated strong interest on ARPA's part to have
NIH as a user. Details are not well documented however in terms
policies or verbal agreements. We pointed to two areas where
problems may arise, particularly within NIH's relatively rigid
grant budget constraints: 1) ARPANET usage costs and 2) ARPA
cooperation in helping NIH users gain access to the network
through ARPA-sponsored nodes (perhaps with NIH-borne costs for
additional modems, etc.). Dr. Baker said the latest word he

had was that ARPA's divestment of the network to BB&N management
would be delayed up to three years by FCC approvals. Meetings
between BRB and ARPA are planned for the near future during
which these concerns will be discussed.

The general organizational problems of shared resources within
NIH was discussed revolving about the issues of whether advisory
committees should be formally set up by NIH or not and how to
coordinate the charters of the Study Sections and operational
resource advisory committees (the Computer and Biomathematical
Sciences Study Section declined to accept the SUMEX advisory role
at the site visit in March of 1973). Carl Brewer said these
matters were sensitive and would evolve with time. The first
step would be to informally set up an advisory committee for

AIM under the AIM Executive Committee. As the general concept

of resource sharing is worked out at NIH, this advisory committee
May move to become a part of a more formal "Shared Resource
Advisory Group" (SRAG) encompassing all types of shared resources
and set up by NIH. In either case, the AIM Executive Committee
is responsible for receiving and implementing within its policies
the advice from the AIM advisory group.

A list of names was discussed for potential members of the AIM
Advisory Committee. The establishment of the advisory committee
will be pursued by Dr. Lederberg and BRB.

No date was set for the next Executive Committee meeting pending
administrative review by BRB of overall committee structure plans
and setting up the NSCRHR facility.
Memo to SUMEX Executive Committee File
January 11, 1974
ATTACHMENT A

(Prepared by Carl Brewer of HIH)

SUMEX-AIM Executive Committee

STANFORD UNIVERSITY MEDICAL CENTER
Stanford, California 94305

Room S047

JANUARY 9, 1974
9:00 A.M,

AGENDA

STATUS OF PROGRAM
BRB Activities
Participants, funding, ARPA, etc.

SUMEX Activities

Hardware acquisition and acceptance

Information Publication with dates and plans for
responses to inquiries

POLICIES ON ADDITIONAL AIM PARTICIPANTS

ADDITIONAL MEMBERS OF SUMEX~AIM EXECUTIVE COMMITTEE
A roposed con hasaractin change
Memo to SUMEX Ta \ggatomma sree File C2L 179

January 11, 1974
ATTACHMENT B |_ne®

py ue MEMORY

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(72 “ words ) T Swartiney
| | STORAGE
| WALA =r mes | Le 94 woes)
l Menoaky { rere Ob ve.
Corp,
ZNICKLACE |
l | Cenrran dise Srraae
| PROCESSOR
| Pop-nfro ! ker-10 C40 woess )}
| TELNET
P |
| ROrocod -—-—-7
—— —— dn POP+10 p=
~~” |
-- [ATER AE |
——w am enw cee ond
32 Lae. ~ Znousrry
reennay ——| 7 /P CIRGNENC TALE
Pears — betn Crock
——— C2 Ones)
fp \ BEC Tare
AR PANETT CO buat deve )
Line PeinreR
(309 4A]
lam feooucrs

 

 

 

Aevised Ounex Comeurer
CONFIGURATION

Freure |
1974

January

Feb ruary

SUMEX SYSTEM SCHEDULE

March

April

May

June

July

August

September

 

Board of Trustees
Approval & Firm
Orders (PDP-10, Data
Prod. Printer, DDC
Disk, TIP

Drop Ship. Data
Prod. Printer to DEC

— oe amb -onp aap |)

 

Ship PDP-10 w/o DDC
Disk to Stanford

2 INSNHOVLLV
yL6T “TT Avenuer

 

 

Ship TIP

 

Install, Checkout &
Accept PDP-10

 

 

Install, Checkout &
Accept TIP

OTHA [9937 TMUOD |ATInpoex| XAWNS 02 oWaW

 

Preliminary System
Available with RP-03
Paging

E+

-?

 

Drop Ship DDC Disk to
DEC (Santa Ana)

 

Final System Available

Ship DDC disk and
Controller to
Stanford

Install & Checkout
DDC Disk

 

Incorporate Disk into
System

with Full Equipment

 

 

 

 

 

 

 

 

 

 

VY |

 
Draft 1/11/74

Memo to SUMEX Executive Committee File
January 11, 1974
ATTACHMENT D

DRAFT - ANNOUNCEMENT OF SUMEX GRANT

NATIONAL COMPUTER RESOURCE AT STANFORD UNIVERSITY

Stanford University Medical. School has been awarded a five year grant
of $2.75 million from the Division of Research Resources of the National
Institutes of Health to establish a "national shared computer resource."

Under the direction of Professor Joshua Lederberg of the Genetics Department,
the SUMEX (Stanford University Medical EXperimental Computer) facility

will be dedicated to applications of Artificial Intelligence in Medicine (AIM).
Its main objectives are: 2) further exploration of computer science
applications in medicine and 2) the development in partnership with the
Biotechnology Resources Branch of NIH, of the relatively new concept of
national shared resources.

The SUMEX resource will initially comprise a powerful PDP-10 computer
facility to be staffed by a group of scientists experienced in artificial
intelligence techniques, The computer is expected to be delivered and
installed this coming April, and will be available to approvedusers of the
national community through a computer-communication network. An AIM national
coordination group is being organized to advise on the management of the
facility and to review applications for eligibility to use this resource.
| Core research efforts at Stanford will include on-going interdisciplinary
projects involving the collaboration of Professors Lederberg, Edward Feigenbaum
(Computer Science) and Carl Djerassi (Chemistry). The heuristic DENDRAL
project seeks to apply artificial intelligence techniques to the interpretation of
mass spectra. A new Project, in collaboration with Professor Stephen Freer
U. C. San Diego, will seek to infer protein structure from xXw~ray crystallographic
data. Other projects that have been identified as initial users of the resource
include an NIH-funded program of AI research in biomedicine at Rutgers University

under Professor Saul Amarel. Two additional projects at Stanford are
Page 2

"Computer-based Consultations in Clinical Therapeutics" under Professors
Stanley Cohen and Stanton Axline of the Medical School and "Higher Mental
Functions Modelling" under Dr. Kenneth Colby of the Computer Science

Department.