FYI - NSFNET Controversy - Letter to Conyers (Rep) from Lane (NSF)

[David Farber <farber () central cis upenn edu>]

Honorable John Conyers, Jr.,Chairman
Committee on Government Operations
House of Representatives
Rayburn House Office Building - Room 2157
Washington, DC  20515 6143


Dear Mr. Chairman:


I am writing in response to your April 12, 1994 letter, in which you
refer to numerous allegations questioning the propriety of certain
actions taken by the National Science Foundation (NSF).  The issues in
your letter have been raised before and are not entirely surprising in
an arena of intense competition among telecommunications interests.  Any
major networking development may lead to a potentially lucrative
advantage for those competing in the Nation's emerging information
infrastructure.  In this atmosphere, a number of serious allegations
have been raised for which I welcome the opportunity to provide an
accurate statement of the facts.


Cooperative Agreement vs. Procurement Contract


As you note, NSF plans to support NSFNET activities through cooperative
agreements rather than procurement contracts.  We chose this instrument
because we believe it to be the best and most appropriate vehicle to
stimulate these activities, not -- as some maintain -- as a way to
circumvent Federal acquisition regulations.


Section 6305 of title 31 of the United States Code (originally enacted
and still commonly referred to as part of the Federal Grants and
Cooperative Agreements Act of 1977) states that agencies shall use
cooperative agreements when


"(1) the principal purpose of the relationship is to transfer a thing of
value to the State, local government, or other recipient to carry out a
public purpose of support or stimulation authorized by a law of the
United States instead of acquiring (by purchase, lease, or barter)
property or services for the direct benefit or use of the United States
Government.
(2) substantial involvement is expected between the executive agency and
the State, local government, or other recipient when carrying out the
activity contemplated in the agreement."


NSF used a cooperative agreement because the principal purposes behind
NSF's solicitation No. 93-52 entitled "Network Access Point Manager,
Routing Arbiter, Regional Network Provider and Very High Speed Backbone
Network Services Provider for NSFNET and the NREN(SM) Program" are to
support and stimulate the creation and growth of extensive computer-
based networks for the benefit of the scientific and
engineering research and education community (consisting primarily of
academic researchers and educators).  This activity will require
substantial involvement of various research communities and of NSF staff
in addressing anticipated but unpredictable technical and logistical
problems.  Both as a matter of law and policy, NSF firmly believes that
a cooperative agreement is the appropriate vehicle.


This very issue was directly examined by NSF's Office of Inspector
General (OIG) in a report entitled "Review of NSFNET." OIG (pages 36-37)
independently concluded that "the NSFNET program is a paradigm for the
use of a cooperative agreement," and that "the use of cooperative
agreements for NSFNET's next stage of evolution is appropriate."  A copy
of that 70 page report, dated March 23, 1993, is attached for your
information.


Allegations of Conflicts of Interest


You also cite allegations that "a senior NSF official failed to recuse
himself from involvement in this procurement despite an apparent
conflict of interest caused by his affiliation with the winning vendor's
partially-owned subsidiary and corporate partner in this procurement."


I can assure you that no NSF official is affiliated with MCI, its
subsidiaries, or its subcontractor.  No member of the National Science
Board is on the Board of Directors of MCI, its subsidiaries, or its
subcontractor. Nevertheless, because of the relationship between ANS (a
not-for-profit corporation created by MCI), and the University of
Michigan through Merit, Inc., one of the Members of the National Science
Board with an affiliation with the University of Michigan did recuse
himself from the Board's consideration of the vBNS award to MCI. The
minutes of the National Science Board meeting reflect this action.


Basis for Awards


Apparently, the allegations made against NSF have caused your Committee
to question the fairness of NSF's selection process for the vBNS service
provider and, specifically, whether the awardee was selected on the
basis of familiarity and favoritism rather than merit.  This was not the
case. Our solicitation sought proposals from any organization that
wished to submit one. It was open to competition from all
U.S.telecommunications companies. A merit review panel of technical
experts from academic and industrial research communities and
representatives of affected communities was assembled in accordance with
customary NSF practices.  Panel members were asked to evaluate proposals
on the basis of the criteria listed in the solicitation, and to reach a
consensus recommendation for award.  The panel did so, finding the MCI
proposal technically superior by a substantial margin. Before and after
the release of the solicitation, steps were taken to encourage broad,
fair participation from the community.  A draft solicitation was widely
distributed to give companies and other interested parties an
opportunity to help us shape the final solicitation.  Also, once the
final solicitation was put forward, questions regarding it were
collected, analyzed and answered in a manner that ensured equal access
to information and fair treatment of potential proposers. I believe
there was nothing improper in the way the Foundation handled the
solicitations and awards for the NSFNET and NREN programs.


I hope this letter, and the attached responses to the nine points you
raised will assure you that NSF acted responsibly and appropriately in
helping form the nation's largest and fastest network for research,
education and technology transfer.  Should you need any additional
information, please do not hesitate to contact me.


Again, I appreciate being given this opportunity to address your
concerns and thank you for your continued interest in the National
Science Foundation.


Sincerely,
                                           Neal Lane
                                           Director
Enclosures


--------------
Detailed NSF Response to Questions Posed by Chairman Conyers


1.  Why has the NSF chosen to acquire these services through a
cooperative agreement
rather than a competitively awarded contract?


As described in NSF solicitation 93 52, the services of the vBNS are
intended to provide the scientific and engineering research and
education community with access to an extremely high speed computer
network that will operate seamlessly with the existing Internet.
Although the vBNS will be provided by the awardee, NSF staff will
continue to be involved in addressing anticipated but unpredictable
technical and logistical problems as they arise, and will participate
with the awardee in their resolution.  Under these circumstances,
cooperative agreements are clearly the appropriate award mechanism.


Section 6305 of title 31 of the United States Code (originally enacted
and still commonly referred to as part of the Federal Grants and
Cooperative Agreements Act of 1977) states that agencies shall use
cooperative agreements when


"(1) the principal purpose of the relationship is to transfer a thing of
value to the State, local government, or other recipient to carry out a
public purpose of support or stimulation authorized by a law of the
United States instead of acquiring (by purchase, lease, or barter)
property or services for the direct benefit or use of the United States
Government.
(2) substantial involvement is expected between the executive agency and
the State, local government, or other recipient when carrying out the
activity contemplated in the agreement."


Clearly, the choice of award instrument (cooperative agreement vs.
contract) must depend on "the principal purpose" of the award, which, in
this case, is to stimulate the development of an extremely high speed
computer network for the benefit of the scientific and engineering
research and education community.  Certainly, Federal agencies will
benefit from the growth of the NSFNET along with other members of the
scientific and engineering research and education community, but this
does not compromise the "principal purpose" of the vBNS: to support the
nation's research and education community.  Indeed, recent analysis of
traffic on the NSFNET Backbone Service shows that less than 2% of the
traffic has U.S. Government entities as both source and destination of
the traffic, and less than 6% involves any U.S. Government organization
at all; nearly 95% of the traffic is among non-U.S.Government sites.


Because the principal purpose of the vBNS award is to provide networking
services in support of a public purpose, a cooperative agreement is an
appropriate award instrument.


2. What evaluation decision structure was used to select the winning
vendor?


After receipt of proposals, a merit review panel of technical experts
and representatives of affected communities was selected in accordance
with customary NSF practice; the panel was chosen to avoid both the fact
and the appearance of conflict of interest.  Panel members were given
copies of vBNS proposals from which pricing information had been
tripped, and asked to evaluate them according solely to the criteria
listed in the solicitation.


Each panel member individually scored the proposals.  After this
technical evaluation, pricing information was revealed, and the panel
was asked to reacha consensus recommendation to the NSF.


3.  Was the proposal submitted by the winning vendor in this procurement
determined to be technically superior to the proposals ubmitted by
competing vendors?  If so, on what basis?


Yes, the MCI proposal was judged by the merit review panel to be
technically superior by a substantial margin to the other proposals
received.  As stated above, the criteria were those listed in NSF
solicitation 93 52, namely, the seven criteria for all components of the
solicitation listed on page 11 and the five criteria specific to the
vBNS component listed on page 12 of NSF solicitation 93 52.


4.  Does the winning vendor currently offer in the commercial
marketplace the Asynchronous Transfer Mode (ATM) services which NSF is
purchasing in this procurement?


MCI does not currently have a commercial ATM service offering.  It will
announce a beta ATM testbed and trials with customers in May 1994, and
will be conducting ATM interoperability trials with local exchange
carriers (LECs) starting in the third quarter of 1994.


NSF solicitation 93-52 requests Internet Protocol (IP) and
Connectionless Networking Protocol (CLNP) service at 155 megabits per
second, a service not available over ATM -or in any other form - as a
commercial product from any vendor or carrier.


Under the cooperative agreement to be concluded with NSF, MCI will
participate in joint studies leading to the development of IP/CLNP at
data rates of 155 mb/s and higher.


According to the solicitation (page 6), the vBNS "...will provide for:
high speed interconnection of NSF Supercomputing Centers (SCCs); the
development of a national high performance computing environment (the
metacenter); applications involving distributed high performance
computing and isochronous visualization; and connection to the NSF
specified NAPs.  The vBNSconnections to the NAPs will, for example,
facilitate connecting the SCCs to research institutions that have
meritorious high bandwidth network applications."


Thus the solicitation focuses on the desired outcome to the research and
education community (i.e., the "public purpose" per 31 USC 6305) of
establishing the vBNS, not on the specific technology (e.g., ATM) to be
used. This choice was made deliberately to allow for innovative
proposals.  At the same time, NSF staff were aware of the state of
technology and also of the potential dangers of idiosyncratic,
proprietary technological solutions; accordingly, one of the evaluation
criteria in the solicitation was "...(the)relationship of proposed
services to emerging international standards such as SONET, ATM, SMDS,
and FDDI."


All telecommunication carriers intend ultimately to use ATM in tandem
with SONET, a transmission technology especially designed for use in
fiber-optic high-speed networks.  MCI is an industry leader in the
deployment of SONET in its network.


During the review of proposals, panelists (though unaware of MCI's ATM
roll-out schedule) believed that the state of ATM technology is such
that the success of an extremely small-scale deployment in the vBNS
would not be jeopardized by MCI's lack of a large scale commercial
offering.


5.  If the winning vendor is unable to meet the delivery goal, or
exceeds the original funding amount, what recourse will the NSF have?


The award for the vBNS has not been made and the award amount and other
details have not been finalized.  Based on consideration of this project
and the winning proposal, the National Science Board (NSB) authorized UP
TO (emphasis added) $50m over 5 years (as requested in MCI's proposal)
for the project.  The actual award amount could be lower or higher; NSB
approval authority allows NSF award amounts to exceed the amount
authorized by up to 10% before having to return for additionalNSB review
and approval.


The cooperative agreement is expected to be on a cost- reimbursement
basis with a ceiling amount and will call for the awardee's "best
efforts" in undertaking and accomplishing the project activities.
However, NSF intends to include appropriatemilestones, performance
standards, timelines and other appropriate developmental goals as part
of the award requirements.  Progress reports and other required
interactions between the awardee and NSF will provide NSF with
information to help evaluate progress on a continuing basis.  Among
other provisions, NSF willinclude appropriate audit and records,
suspension, and termination clauses in the agreement. Using these
mechanisms, changes or adjustments in the award can be made if it
appears that the awardee is or will be unable to perform in a reasonable
time, or if the quality of the performance does not meet the
requirements in the cooperative agreement.


6.  Did any NSF officials involved in this procurement hold discussions
with the winning vendor at any time during the evaluation process which
preceded this award?  If so, which other vendors were given the same
opportunity?


There were no substantive discussions germane to the solicitation
between NSF officials and any proposer during the vBNS evaluation
process. MCI is a joint study partner with ANS, a sub awardee to Merit
under its current award for NSFNET Backbone Services, and other vBNS
proposers currently have awards from the NSFNET program.  Routine
contacts between NSF program officers and the proposers has at no time
been proscribed, but contact has been limited to that necessary for the
management and prudent oversight of existing awards and was unrelated to
the vBNS or any other element of NSF solicitation 93-52.


7.  Was the proposal submitted by the winning vendor priced lower than
the proposals submitted by competing vendors?


When prices were revealed to the review panelists after they had ranked
the MCI proposal highest on technical merit alone, one proposal which
had been judged "Technically Unacceptable" by the panel was seen to be
substantially lower in price.  In the written instructions given to the
merit review panelists when proposals were first distributed,
"Technically Unacceptable" was defined as "a proposal which is so
seriously flawed or of such low overall quality that even if it were the
only proposal received it would not be acceptable for award."


Even if the NSF had used a procurement process for the vBNS, we believe
this lower priced proposal would have been determined to be technically
unacceptable or non-responsive and thus excluded from any competitive
range; further negotiations with the proposer would not have occurred.


8.  Are any members of the National Science Board also on the Board of
Directors for the winning vendor, its subsidiaries, or its
subcontractors? Are any National Science Foundation officials involved
in this procurement affiliated with the wining vendor, its subsidiaries,
or subcontractors?


No NSF official is affiliated with MCI, its subsidiaries, or its
subcontractors. No member of the National Science Board is on the Board
of Directors of MCI, its subsidiaries, or its subcontractors; had that
not been the case, the National Science Board conflicts of interest
procedures and process (copy attached), which have been reviewed and
approved by the Office of Government Ethics, would have precluded that
member's participation in any discussion, deliberation, or decision
regarding MCI.


Perhaps the "senior NSF official" mentioned in your letter is a
reference to James J. Duderstadt, the President of the University of
Michigan and Chairman of the National Science Board.


Merit, Inc., a consortium of Michigan universities including the
University of Michigan, is entitled to membership and is represented on
the Board of Directors of Advanced Network and Services (ANS), a not-
for-profit corporation created by MCI Telecommunications, IBM, and
MERIT, a consortium of Michigan universities.  Dr.
Douglas E. vanHouweling, a vice-president of the University of Michigan,
occupies Merit's seat on the Board of Directors of ANS. MCI is a joint
study partner with ANS in its subaward from Merit for the NSFNET
Backbone Services.


Dr. Duderstadt is not now, and has never been, a member of the ANS Board
of
Directors. Even so, because of the relationship between ANS and the
University of Michigan, Dr. Duderstadt did recuse himself from the
portion of the National Science Board meeting at which the motion to
make the vBNS award to MCI was discussed and voted upon.  National
Science Board minutes from that meeting (attached to this response)
document the recusal, and confirm that Dr. Duderstadt literally removed
himself from the meeting in an abundance of caution and concern over
even the appearance of any conflict.


These allegations of conflicts of interest lack any factual or legal
basis. Further, by failing to identify specific names or relationships,
they have the potential to unjustly impugn the character of any who were
involved in this important evaluation and selection process.


9.   What legal arrangement (i.e., grant, cooperative agreement, or
contract) was used, or is expected to be used, to acquire the related
services of a Network Access Point Manager and a Routing Arbiter
organization?


Substantial NSF involvement in the operation of the NAPs resulting from
solicitation 93-52 will be needed to ensure that the facilities are
adequate, that commercial traffic does not impede that of the research
and education community, and that fairness is exercised in all
activities.  NSF staff will also be involved in the operations of the RA
awardees, primarily to coordinate the routing policy research required
in the solicitation to be carried out among the RA, NAP, and vBNS
awardees.  Thus, as described in the response to question no. 1 above,
NSF believes that the use of the cooperative agreement award instrument
is clearly appropriate and proper.  Consistent with the wording of NSF
solicitation 93 52 and the Federal Grant and Cooperative Agreement Act
of 1977, NSF intends to negotiate cooperative agreements for all of
these awards.