IRIE SEOEESIISDIIN IE AN OSNEIERD NASA MEIN EROTEY 6 OPPILE FACMORANDUM © STANFORD UNIVERSITY © GFIRICE MEMOCRANUUS

MOY - . 7B aan Preliminary copies to:
, eats G. Franklin
P. Carpenter

EE. Pierce
Date: May 16, 1973

 

To + Clayton Rich, M.D.
From: gy, Lederberg

Suestct: Agenda Item for the June Board of Trustees Meeting

REFERENCES: 1) Memorandum from Elliott Levinthal to Clayton Rich, M.D.,
Subject: SUMEX Grant, November 14, 1972 (attached).

2) Memorandum from William B, Rowland to Clayton Rich, M.D.,
Subject: Request for Advance of Capital,
December 1, 1972 (attached).

3) Memorandum from Frank G. Riddle to Clayton Rich, M.D.,
Subject: Purchase of PDP-10 Computer for SUMEX Grant
Proposal, December 1, 1972 (attached).

The Stanford University Medical Experimental Computer (SUMEX) proposal,
now pending with the Biotechnology Resources Branch of the National
Institutes of Health, is nearing the final stages of review. The National
Advisory Research Resources Council will consider SUMEX on June 14-15.

As discussed last fall (see referenced memoranda), there is a possibility
that University assistance will be required to finance the SUMEX computer
system. Such assistance was tentatively approved at that time, subject to
Board of Trustees concurrence.

It is proposed that University financing of the computer be considered at

the June 15 Board of Trustees meeting for approval contingent upon NIH

awarding of the grant, and the unavailability of government funds to purchase
the computer system in the first grant year. This financing would entail

an external loan of approximately $870,531, the principal of which would

be paid back in equal installments over the five year term of the grant.

At the end of this term, cumulative interest charges amounting to $152,343 would
remain unreimbursed to the University, assuming a 7% interest rate.

I recognize that the results of the NIH Advisory Council's review cannot be
known in time for the June Board meeting. Also the timing of a grant award,
if made, is uncertain. An award could be made as early as June or delayed
depending upon priorities established at NIH. However, in the event of an
early award, Board action in June would avoid a delay through the summer until
the Board reconvenes in the fall. The disadvantages of such a delay are
presented in subsequent paragraphs.

WHONVAC waW JMO © ALISHIAINN GHOINVIS e WNONVYOWIW 32D1d0 @ ALISYZAINN GXOINVIS © WAONYMOWIW 3DIO © LEICHIAINA GYNnaNnwiC e
May 16, 1973
Page 2

BACKGROUND

The SUMEX proposal has been pending with NIH for approximately one year
(submitted June 1, 1972). In that time the grant has undergone several
rounds of elaboration with the Computer and Biomathematical Sciences Study
Section. The most recent version of the application (March 18, 1973) is
attached.

The SUMEX proposal seeks to establish a powerful national resource for the
development of applications of computer and biochemical analysis techniques

in biomedical research. The resource has two interacting themes: 1) applications
of Artificial Intelligence techniques in Medicine (AIM) and 2) applications
of analytic methodologies such as gas chromatography/mass spectrometry to
problems of biomolecular characterization. Our on-going research effort
(Heuristic DENDRAL), which applies artificial intelligence techniques to
problems of mass spectrometry and body fluid analysis, serves as a core for
this resource. The resource would consist of a PDP-10 computer facility
together with a local group of professionals, experienced in utilizing AI
techniques in a range of problem areas. This group of people is inter-
disciplinary, representing medical science under me as Principal Investigator,
computer science under Professor Feigenbaum, and chemistry under Professor
Djerassi. The resource would be made available to a national community of
NiH-funded users (including other approved projects at Stanford) via a
national communications network, under the review of a National AIM Advisory
Committee established by NIH.

Besides the exploration of advanced applications of computer science and
biochemical analysis techniques in medicine, this proposal addresses the
problems of developing the relatively new concept of nationally shared
resources. Basic communication problems having been solved, facilities such
as SUMEX potentially offer a far more economical means for supporting
computer-related research and encouraging regular intellectual interactions
between remote groups.

The SUMEX facility would not solicit users from other campus computing
facilities. Thus, no competition for the funding base of campus facilities
is entailed in SUMEX. In fact the National Advisory Committee for AIM
will review all authorized users and control access to SUMEX.

GRANT STATUS

The SUMEX application was reviewed by the Computer and Biomathematical
Sciences Study Section on April 25, 1973. It will be given final review

by the National Advisory Research Resources Council on June 14-15. We do

not know the details of the Study Section deliberations. However, if SUMEX
had been deferred (as has happened previously) or completely disapproved,

we would knew that by now. Therefore available indic cations, while unspecific,

are Chat au action ivr apprevas al some pricricy s&évei Was taken by tac Stucy

Section subject to review by the Advisory Council in June. The effects
on the grant award of Council review and the interaction of the resulting
SUMEX priority score with NIH program goals and appropriations are
May 16, 1973
Page 3

difficult to predict. Our previous contacts with the Biotechnology
Resources Branch indicate a strong interest in establishing nationally
shared resources as the future direction for supporting research activities
requiring large scale facilities. SUMEX appears to fit that model well.

Such speculation, however, does not offset the fact that definitive action
by NIH cannot occur before the last half of June and could be delayed longer.

on
~ .

FINANCLAL CONSIDERATIONS . ~ frvrone7
hen bgrteetion

As summarized in the introduction, previous teerat tone wit NIH--have raised

the possible need for University assistance in g the proposed

SUMEX PDP~10 computer. Purchase is favored over lease in order to meet the
equipment configuration needs of the research program. NIH indicated that

the finance charges contained in the initially proposed lease-purchase plan
(standard with Digital Equipment Corporation (DEC)) could not be all wed and
that a revised budget should cover only the purchase price of the equipment.
At the same time the overall proposal budget was reduced in a way so as to
specifically constrain the machine configuration to be minimal relative to
program needs. An attempt to lease the equipment must increase costs to
cover third party involvement and because of the NIH budget ceiling, these
costs would have to be absorbed as a reduction in the equipment configuration.
Qur subsequent investigations of the experience of existing artificial
intelligence-oriented PDP-10 facilities and discussions with the proposed national
user community have supported the need for an undiminished facility. Hence, last
fall we proposed University assistance to finance the SUMEX machine through
an advance of capital with subsequent reimbursement from the grant. This
proposal received tentative approval as indicated in the attached memoranda.
Our most recent budget is higher than that submitted last fall specifically in
the areas of personnel and supporting supplies. This increase is responsive
to a Study Section criticism of too little support for the external facility
users. The equipment budget remains constrained as above.

 

Since these discussions last fall, the fluctuations in federal budgets have
raised a possible altemative approach. Because of NIH-BRB interest in
pursuing nationally shared resource programs, and the long-term benefit to
their grantee community (including Stanford) of establishing a facility such
as SUMEX, NIH indicated it may be able to advance all of the money needed to
purchase the SUMEX facility in the first year. This prospect is still under
consideration. Recent indications are that a definitive statement on the
availability of such funds could not be made until later as the FY '74 budget
becomes clearer.

Thus two possibilities exist in financing the SUMEX machine: direct

government purchase or Stanford financing. The former is clearly to be
preferred but can be implemented only with NIH impetus and the availability

of adequate federal funds. We will continue to See this alternative, ark

ee cele bead 08 CO, ae wir a lt |
peste poses ae bee eae Saute)
May 16, 1973
Page 4

Should this approach not be feasible, we request Board of Trustee approval
to advance the necessary capital through external borrowing. The following
numbers refine those contained in Dr. Levinthal's memorandum of November 14,
1972 (reference 1). These numbers relate to only the SUMEX portion of the
proposal (as did the November 14 numbers) and do not include the related
DENDRAL research grant. Note that a renewal application for DENDRAL was
submitted along with the SUMEX elaboration on March 18, 1973.

BUDGET YEAR 1 2 3 4 5

Estimated Total Direct
Costs (Page P-89 of $765,573 $785,286 $909,109 $877 ,486 $909,167
the SUMEX proposal)

Estimated Equipment
Costs Not Subject to $137,633 $232,633 $292,633 $232,633 $232 ,633
Indirect Recovery

 

Net Total Direct Costs $627,940 $552,653 $616,476 $644,853 $676,534

Estimated Indirect

Recovery @ 47% NTDC $295,132 $259,747 $289,744 $303,081 $317,971

Estimated Total Indirect Recovery = $1,465,675
Assuming initial equipment costs of $1,088,164, including 64 tax, the grant
budget provides for recovery of this principal over five years at $217,633
per year. Thus, assuming recovery starts at the beginning of year 1 and an
interest rate of 7% for the required $870,531 (1,088,164-217,633) loan, the
following interest charges would be paid by the University:

BUDGET YEAR 1 2 3 4 5
Outstanding Principal $870,531 $652,898 $435,265 $217,632 $ 0
Interest @ 7% $ 60,937 $ 45,703 $ 30,469 $ 15,234 $ 0

Total Loan Interest Charges = $152,343
It should be noted that these are estimated budget numbers and must be

negotiated further with NIH if a grant is awarded. In that respect they
represent upper bounds on the numbers.
May 16, 1973
Page 5

We have negotiated a 15% discount with Digital Equipment Corporation for a
portion of the equipment, conditional on a cash payment for the computer
system upon satisfactory installation, on a firm order by July 31, 1973,

and on a minimum order of $700,000. Violation of these conditions makes

this discount subject to renegotiation with DEC. This discount applied to
the DEC portion of the equipment would reduce the purchase price to $946,723,
including tax. This in turn reduces the cumulative loan interest charges to
$132,541. Our ability to realize these savings depends upon grant award
timing and subsequent negotiations.

Because government funding cannot be guaranteed, even with a five year
approved grant, the equipment acts as a guarantee of the loan. The resale
value of computer systems is difficult to predict because they are technology
dependent. The proposed system is the latest large scale machine offered by
DEC (first shipped in the fall of 1972) and represents a significant
investment in design upgrade over the earlier models. The PDP-10 machine is
popular among universities, ARPA contractors, and increasingly in medium scale
terminal-oriented business applications. A very incomplete list of PDP-10
facilities includes Stanford (AI Lab and IMSSS), SRI, Ames Research Center,
Caltech, UCLA, MIT, Harvard University, Carnegie-Mellon Institute, TYMSHARE,
Copley Press, etc. This list of PDP-10 users substantiates the potential
resale market. At the end of each year, the resale value required for the
University to break even on its investment (cumulative interest charges plus
outstanding principal),expressed as a fraction of the initial purchase price,
is summarized below. Note these numbers ignore the possible advantageous effects
of a purchase price discount.

BUDGET YEAR 1 2 3 4 5

Ratio of Loan Costs plus

Remaining Principal to 85.6% 69.8% 52.6% ~ 34.04 14.0%
Purchase Price

At the end of the grant period,athe machine eé _-Loaa-

eostsyer attérnatively a-phitanthropic source could be found to récover these
costs—and—eo-—and the SUMEx TaCr yy permanently to available Universtry

 

 

 

 

 

 
May 16, 1973
Page 6

REASONS FOR BOARD OF TRUSTEES ACTION . ey,
_ UAL

In summary, Board action is necessary to approve the fe
-of- money for possible -University_assistance—in financi& the SUMEX computer.

Board approval is requested contingent—upon awarding-of—the-grant and-tpen
unavailability of government -compiter-purcthase funds. We, of course, will

continue to pursue direct NIH purchase. If this is not possible, we feel it

is in the University's interests to advance the necessary funds to be

reimbursed over the five years of the grant.

 

It is important to obtain this contingent approval as early as possible so
that the various time-critical elements involved in implementing SUMEX not

be impacted. In addition such action confirms for NIH the University's
commitment to support the grant, thereby encouraging its being awarded.

The aspects of SUMEX implementation for which time is important include
internal budget planning, procurement negotiations with DEC, and establishing
a viable resource as soon as possible relative to the national user community.
In some ways the internal problems are most important since delays in
finalizing a SUMEX grant increase the risks we face in personnel and budget
commitments that must be made for the next academic year.

We urge Board consideration and approval of the proposed financing arrangements
at their June 15 meeting.
November 14, 1972

Clayton Rich

Elliott Levinthal

SUMEX Grant

The following is a summary of the current status of the SUMEX proposal

to NIH for a new PDP-10 computer and a recommendation for arranging the
financing of the machine under a revised grant application. As you know,
the SUMEX proposal was submitted to the Biotechnology Research Resources
Branch of NIH in June 1972 and was site visited on September 11-12. We
have recently received a letter from NIH indicating that consideration of
the proposal was deferred based on the inability of a portion of the
collaborator community to justify the proposed scope of the facility. The
letter further stated that a modification of the proposal reducing its scope and
emphasizing applications of artificial intelligence, would be reconu.idered
by the study section ££ submitted by December 4, 1972. We separately
received a sugpested outline for an overall 45.5% budget reduction.

This budget reduction affects several aspects of the proposal:
1. A significant reduction in personnel support (46.5%)
2. A significant reduction in hardware facilities (28.82%)

3. A disallowance of finance charges included in the original
procurement budget for the computer. We had proposed to purchase
the machine under a lease-purchase plan based on its estimated
long term value as a research tool in the Medical School.

Based on these guidelines we have been modifying the grant application for
the December 4 deadline and find, after examining available alternatives,
that in order to maintain a viable technical solution to the problems posed,
we require University assistance in financing the computer acquisition.

The dilemma is basically that the study section of NIH reduced the computer
system on technical grounds to a truly minimal configuration and at the

same time limited the funding level to cover only 20% of the purchase price
per year over the five year grant. Any attempt to lease the equipment
would increase yearly costs with a corresponding decrease in hardware
complement to a point which would not allow conduct of the research. The
only apparent alternative which preserves the viability of the grant is

for Stanford to advance the capital to purchase the machine and to pay the
necessary interest or the lost revenue on general funds thereby committed
out of indirect funds generated by the grant. The resale value of the computer
equipment adequately guarantees the investment. To this end, we are
attempting to fet hhe Digital Equipment Corporation to guarantee the
repurchase of the machine at a achedule after 3, 4 or 5 yeara corresponding
to Stanford's outstanding investment in the equipment (principal and accrued

intere at). DASS2 tal oT APA VAL kS on viz U,A,ehants rm et tal 3 tiva,

25832 MOTI EL ADL ST Are wseet
Clayton Rich, M.D. Page 2 November 14, 1972

The following are estimates of the incoma and costs involved:

Budget Year:

 

Estimated Total $565.6K 548.7 564.0 581,2 598.3
Direct Cost

Estimated Capital 175.2K 175.2 175.2 175.2 175,2
Equip.Costa*

Estimated Net $390.4K 373.5 388.8 406.0 423.1
Total Direct Costs

Estimated Indirect $179.6K 171.8 178.8 186.8 194.6
@ 46% NTDC

TOTAL INDIRECT = $911.6K

 

 

Estimated Interest $ 49.1K 40.3 30.7 20.6 9.9
or lost Income @ ,

7% on Initial

$876K outlay

TOTAL INTEREST = $150.6K

2 17.2% of initial outlay

 

 

Equivalent Third $ 77.1 77.1 77.1 77.2 772
Party Finance

Charges @ 2.4% per

Month Repeyment

TOTAL THIRD PARTY INTEREST = $385.5K

 

 

*This assumes a purchase price for computer equipment of $876K paid back
with five installments of 20% each from the grant and that a $50K/year maintenance
contract will qualify for indirect recovary.
Clayton Rich, M.D. Page 3 November 14, 1972

Since the government has refused to pay finance charges for a third party
arrangement, we propose that the University advance the $876 purchase price
of the computer to be paid back on a yearly basis at $175.2K/year. Since
the first year payment would be immediately forthcoming this would require
an effective $700.8K advance. At any point in the grant the University's
investment is covered by the resale value of the machine. We are attempting
to have DEC guarantee the repurchase based on the outstanding principal and
accrued interest. It is likely if,we can negotiate such an arrangement, that
conditions will be imposed such as

1. Minimum configuration size

2. DEC maintenance at least at time of resale

3. Adequate insurance coverage

4. No guarantee until years 3, 4, and 5 of the grant

5. Only PDP-10 portions of the system will be covered - approximately
$50K of the original equipment consists of small PDP-11 machines
under separate administrative control at DEC.

The necessary percentages of initial value to cover Stanford's investment,
assuming 7% interest, are:

Budget Year 1 2 3 4 5

Outstanding Principal

and Accrued Interest 85.6% 70.2 53.7 36.1 17.2
as % of Original Cost

at the End of Each Year

The impact of this arrangement after 5 years is that Stanford would own a
computer of significant value as a research tool and could choose:

1. To sell the equipment to regain invested capital and accrued
interest for a net zero cost.

2. To regain invested capital through fee for service use of
the machine from subsequent grants or other sources.

3. To accept the investment of 17.2% of the original price as a
valid addition to University research facilities.

We understand that this use of University funds must, of course, compete
with other priorities but we feel that the investment is justified both in
enabling the acquisition of a $500-600K per year research Brant as well as
in providing a valuable long-term computing asset to the school. We
recommend approval of this method of financing the SUMEX facility.

cc: W. ‘Rowland
To 5

From:

Suasect:

William B. Rowland

my bw A Aj

Date; 1 December 1372

Clayton Rich, M.D.
lS

Drs. Lederberg and Levinthal request clearance to commit
between $800,000 and $900,000 of University general funds

as a cash advance for the purchase of a new computer (PDP-10,
SUMEX Grant Proposal). This program was approved by you
earlier this year when the first grant application was sub-
mitted to NIH. At that time no capital advance was required,
but there was a potential obligation of $75,000 in general
funds for alteration expense. This earlier grant proposal was
reviewed by the Computer and Biomathematical Sciences Study
Section and returned with certain revisions recommended. The
revised proposal (attached) is ready for submission.

I recommend that you approve this proposal for forwarding to
NIH. The highlights are as follows: ,

Program - approved by you earlier when no medical school
general funds were involved.

Financial Committment -

~-$170,000 in meaical school general funds for interest
expense (Stanford would borrow the money and recover
the principal by charging the grant over a five year
period).

--Contingent liability to pay off the loan out of general
funds in the event that the government should terminate
funding before the five year grant period was endec.

~-It will be necessary to obtain Board of Trustees approval
to borrow the necessary funds on University credit.

 

Means of Recovery or Escape - The grant application does
not commit the University to any particular means of
financing (open to negotiation so that the stucy section
is not involved in the financing mechanism). The University
is not committed to proceed until the grant has been offered
ana the University has accepted.

 

The Department (Joshua Lederberg) will seek gifts to cover
the general fund requirement for interest expense
($170,000).

The potential liability may be reduced by negotiating the
reco. ' fox che ecuivment to be pucgeted in three years

instead of five years.

MA ra reese

ALISHTAINT GUYOINVIS *

WonsrerarvyTW WDBp4si98 e

PPI TAISEN moey Estee

oN cree wae PP aS a

woke

teers
Page 2
Memo to Clayton Rich, M.D.
l December 1972 ,

The liability may be reduced by negotiating a fifteen percent
discount over the present price arrangement.

The Reason for Approving

--The research program has been approved in principal by you
last May.

--The University will acquire the research equipment at a
relatively minor cost.

--This substantial research project will provide a broader
base for spreading University and Medical School overhead.

-~-The computer may, under certain conditions, be available
to other projects. Tat is, the application contains the
phrase "projects (would) be entitled to cooperate with SUMEX
by furnishing poolable resources in exchange for prorata
shares of time on the augmented resource".

 

Additional space will not be required. General funds will not be
required for additional personnel or alterations. The project

has been cleared with Gene F. Franklin for University wide coordin-
ation of computer facilities. The Controller's office has reviewed
and approved the financial proposal. *

*Financial report from Frank Riddle

WBR:mk

Attachments
Grant Proposal

ec: Richard L. Balch
Frank G. Riddle
Elwood C. Pierce
To t

From

Suauect:

Date: December 1, 1972

Clayton Rich, M.D.
Frank G. Riddle

Purchase of PDP~10 Computer for SUMEX Grant Proposal

After analyzing the request for capital to purchase a PDP-10
computer for the SUMEX grant proposal, I'am recommending your seeking
outside borrowing for the required $893,000 (based upon a $1,116
purchase price). External borrowing will require Board of Trustee
approval which you will have to prepare. Interest on the $893,000
investment would be covered from internal resources of the Medical
School. The recommendation is based upon the facts described in William
Rowland's November 17, 1972 letter to you, Elliott Levinthal's letter
of November 14, 1972, also to you, and supplemental information supplied
by Elliott Levinthal on November 30. It is my understanding NIH is.
willing to fund only the original cost of the computer in a three to
five year period, with the University committed to pay interest or finance
charges. It is also my understanding that Digital Equipment Corporation
has refused to guarantee repurchase of the computer.

The following rationale was developed in making the recommendation
to borrow externally:

(1) The computer will be used solely for the SUMEX project and will

“not compete with other central computer facilities.

(2) The current priority needs on the University's plant fund
reserves preclude funding the PDP-10 from these reserves.

(3) Third party financing at 2.4% per month is too costly.

(45 The use of Medical School Ford funds would also be a costly
investment. The attached schedule shows a loss of $210.1K in growth

potential and $112.9K loss in income for a total investment loss of
$323.0K. ,

Borrowing externally at the following interest rates will cost
$133.8 - $178.7K in interest payments for the five years which must be
covered from Medical School resources.
‘Clayton Rich, M.D.
November 30, 1972

 

 

 

 

Page 2
(000's)

. Beginning Interest Rates

Year Balance 64 74 8%
1 $893 $53.4 $62.5 ‘$71.4
2 670 40.2 46.9 53.6
3 447 26.8 31.3 35.8
4 224 13.4 15.6 17.9
5 0 - - ~
Total Interest $133.8 $156.3 $178.7

A partial offset to the interest payments will be the indirect cost
recovered on the project and allocated to the Medical School which per-
tains to the use of equipment. This amount is approximately 3.8% of
the indirect cost recovered. Using Elliott Levinthal's indirect cost
calculation, the amount pertaining to equipment usage is $25,200 for
the five year period.

Another added factor is the potential of tax exempt borrowing under
the proposed "California Educational Facilities Authority." If approved
by the California Legislature, the act will give the University the ability
to borrow with tax exempt status and reduce interest rates by 24 to 2.5%.

You should discuss obtaining Board of Trustee approval and borrowing
with Rod Adams as soon as possible. I have copied him in on this memo
so that he will be informed on the subject.

, .
Attachment Ae /
, +

ec: R. Augsburger
K. Creighton
G. Franklin
R. Adams
W. Rowland
Attachment

A.

B.

D.

Investment Loss When Financing PDP-10 From Medical School Ford Funds

Assumptions

1. Required Investment $893 repaid in four equal annual payments of
$223.

2. Yield and Gain Pool Rate of Return.
Yield -— 4%
Gain — 82%

Potential Yield and Gain if $893,090 Remains in Y and G Pool

 

 

Period
1 2 3 4 5
Growth $893.0 $964.4  $1041.6  $1124.9 $1214.9
Yield 35-7 38.6 41.7 45.0 48.6

Total Yield = $209.6

Potential Yield and Gain if $893 is removed from pool and repaid
in 4 equal payments

 

 

 

 

 

 

 

 

 

 

 

 

Period
Payments i 2 3 4 5
7 $0 $ 0 $ 0 $ 0 $ 0
1 223.0 240.8 260.1 280.9
2 223.0 240.8 260.1
3 223.0 240.8
4 223.0
Total 0 223.0 463.8 723.9 1,004.8
Growth -
Yield 0 8.9 18.6 29.0 40.2
Total Yield = $96.7
INVESTMENT LOSS
Without With Investment
Investment Investment Loss
Principal at $1214.9 $1004.8 $210.1
Year 5
Yield for five 209.6 96.7 112.9
Yours

Investment Loss $323.0