Creating the Needed Interface
Seite 6: The Change at ARPA
Describing the changed environment, Alan Blue an administrative officer who worked at ARPA/IPTO over a period of time, explained (Interview Alan Blue, Babbage Institute):
From the time I arrived until the time I left, it's objective changed, and the political and budget climate changed.
These changes in IPTO were influenced by pressures from Congress and changes within the Department of Defense during the early 1970s which had led to a change of emphasis in ARPA. Congress introduced the Mansfield Amendment in 1969 and it was incorporated into the Defense Procurement Authorization Act of 1970. The Mansfield Amendment required that the DOD only support basic research with "a direct and apparent relationship to a specific military function or operation." (Barber VIII-19-20)
The problem of this requirement was that it failed to recognize that basic research cannot be directed at any application or it is no longer basic research. Thus even though the language of the Mansfield Amendment was softened and even repealed in subsequent years, it had a harmful effect on the environment at ARPA. The emphasis shifted from basic research to pressure for application oriented research. (See discussion of the introduction of the Mansfield Amendment into 1970 Congressional Legislation in Barber and also changes in Department of Defense organization during this period, Barber VIII-19-27)
Instead of Licklider getting to choose his replacement, as had been the practice, Heilmeier chose David Russell, a Colonel with whom he felt he would be able to work, as the new director of ARPA/IPTO.
In ARPA/IPTO after Licklider left, concern with command and control was substituted back for interactive computing, university funding became less a part of the ARPA budget, and therefore there was less seeding of new leadership in computer science.
During this period in the 1970's, the director of ARPA withdrew the autonomy the director of IPTO had been provided with to develop and support basic research programs. The autonomy was replaced by the pressure to create an application oriented program which would serve the particular needs of the Services. Efforts to start programs that would support basic research were discouraged. As Robert Kahn, head of IPTO from 1979-1985 remembers: "We started a program to develop some centers of excellence in computer science, and that program was phased out because of a decision to not work on centers but to work on specific ideas." (Kahn Interview, Babbage Institute, pg 42)
Such pressures were accompanied by and also in response to the pressure on the DOD and ARPA by Congress and the President during this period in the 1970s. But they began a course of events that led to the ending of the IPTO in 1986, when its functions were merged into other offices at ARPA which were not focused on basic research in computer science.
Blue, describing the change, noted: "It is just a question of doing undirected research at the start and supposedly totally relevant research at the end." [Blue Interview, Babbage Institute]
Describing the change at ARPA/IPTO, Newell proposes that (Newell Interview):
There have been two antagonistic participators....with the ARPA community...with the ARPA office.
Newell explains that one group within ARPA saw the thrust to product oriented research as a tactic to get enough funds to also continue the basic research that was needed to advance computer science. Another set of people, however, he noted, were only interested in getting the good things that scientists could produce. Newell refers to ARPA Director Heilmeier saying that (Newell Interview, pg 102-103):
We have spent $35 million getting this started. Now, what have you done for me lately? Now it's time to deliver.
Newell adds that this is the kind of statement that says, "You have had your vacation; now it's time. The palmy days are over." But he also describes those in the ARPA office (Ibid.):
whose vision is really related to the original vision, although it shifts, because when Lick starts it there is no such computer science thing, and then, in the 1980s, it exists already. The issue is fundamentally one of this same long-term growth of information technology and the application stuff is all tactics. The environment forces you to do this. Like the justification for quoting that thing in the Congressional stuff, you wouldn't put it in there unless you had to, but that's what it takes to get the 250 million, so it's playing the game. And both these kinds of characters have existed throughout ARPA's later history.
Licklider's Contribution to Computing
Describing Licklider's achievement at ARPA, Herzfeld, the ARPA director who followed Sproull in the mid 1960's, noted (Herzfeld Interview, Babbage Institute):
They predicted the future of computing in America remarkably well, number one. I mean, they said "We clearly can do the following. It makes sense and we ought to do it, so let's go do it." And indeed, it happened. Networking, interactive graphics, time-sharing, and all these things that are now commonplace were in the air, and we saw to it that they would happen.
Herzfeld describes lectures Licklider gave at ARPA about the new form of computing that he envisioned. Herzfeld says that Licklider's vision was [Ibid]:
to make computing accessible, to make it more efficient to really use the powers of the computer that were available which he contended -- and I believe was quite right -- were used in a very inefficient manner [before Lick's work-ed]
Continuing his description of Licklider's contribution to revolutionizing the field of computer science and the form of computing available to people, Herzfeld explains:
The idea of networking many computers, the ideal that you would not know on which computer your problem was being solved, nor in which computer's databank did the data reside that you needed for solving your problem, that the whole system became transparent, and you just did what you wanted to do and all the rest was taken care of by the system.
ARPA/IPTO's Contribution to Computing and to an Organizational Form for a Government-Science Interface
ARPA/IPTO continued to function for 11 years after Licklider left in 1975. Some of the most important research continued to be done and supported during this 1975-1986 period in IPTO's history. But the environment at ARPA/IPTO was changed in fundamental ways. In 1986, its functions were merged into other ARPA offices, including the Information Science and Technology Office (ISTO), the Tactical Technology Office (TTO) and Strategic Technology Office (STO).
Research done at IPTO (1962-1986) made possible important new concepts and systems like the Internet. Licklider and other leaders of the computer science community like Ivan Sutherland, Robert Taylor, Larry Roberts and Bob Kahn who worked at ARPA/IPTO have left an important legacy and a foundation to be built on.
During the course of IPTO's development, the principles of how to build a community of computer science researchers and have that community provide the continuous and needed input for those directing the IPTO office were identified and applied during a substantial period of time. The successful interface also provided for the support and protection of the IPTO research community so that it could contribute a number of important breakthroughs in computer science, such as interactive computing, time-sharing, packet switching, internetworking, AI, VLSI. (See Uncapher Interview, pg. 19-20) Even more importantly, a new institutional form was created and functioned effectively within government and connecting government with the computer science research community. That institutional form was the Information Processing Techniques Office.
Describing the development of IPTO at ARPA, Newell proposes that it was something special and not necessarily reproduced at other ARPA offices. He attributes the uniqueness to the fact that top computer scientists like J.C.R. Licklider, Ivan Sutherland, Larry Roberts, and Robert Kahn headed the office during various periods in its existence. Newell compares other ARPA offices that he knew about with the IPTO leadership (Newell, pg 58):
I did not see those (other-ed) offices as run by people who were viewed as first class scientists in their field. It's always been part of the IPTO mystic, which we failed on a couple of times, but with much notice among ourselves, that the guys who ran IPTO were ourselves.
Newell explains that the IPTO scientists were "not bureaucrats" and that they were among the best in the field so that they were able to "argue with you at a level where it is absolutely even field."(Ibid.) Kahn adds that while a director while IPTO would not know all areas of computer science, good managers knew who to go to for advice. And that sometimes this would prove a superior way to deal with a problem as one was not limited by fixed ideas in an area of expertise. (Kahn Interview, pg. 25) Though there were other offices at ARPA like the Materials Science Office run by top scientists which also had distinguished achievements, Newell maintains that it is important to distinguish that "ARPA Style is not IPTO style. It's different...its history of working on missiles....So ARPA is not, my view is that ARPA is not characterized by IPTO." (Ibid.)
Changing the Institutional Forms Supporting Basic Research
Newell also describes the tradition of the community choosing who would be the head of IPTO. When leading computer scientists were at the head of IPTO, they were able to provide respected leadership to the computer science community. Newell describes how the ending of IPTO meant the ending of such leadership. He writes (Ibid.) :
So the idea that IPTO in the form of Larry Roberts and Bob Kahn would go and simply create a research project called the ARPANET (or the Internet -ed) and simply do it themselves, meaning not sit around and wait, (made-ed) perfectly good sense. No eyebrows were even raised about that, because they're just guys like us. Now my belief is that none of the other characters floating around in TTO (Tactical Technology Office-ed) and so forth, have that character now. Like, in fact, none of the people in ARPA have much of that character now.
Also certain institutional forms were needed to support basic research, such as office directors getting support from the head of ARPA and the head of ARPA getting support from the Director of Defense Research and Engineering to be strong directors. And they in turn providing support to their program managers to make their decisions based on their best scientific instincts. These institutional forms were harder to continue when Congress was pressuring the DOD to be more product oriented in its research.
The loss of ARPA's ability to continue to support IPTO and the basic research that was at the heart of its work has had consequences. Support for basic research maintains a pipeline of new concepts and principles. The pipeline is broken if the support for basic research is stopped. Also research questions that could be answered by the collaboration of the research community and the IPTO in the past no longer have that important resource. One such important and unanswered research question concerns problems of scaling the Internet. Robert Braden notes how it has been difficult to have this research problem treated seriously. He writes (RFC 1336):
For some years now we have been painfully aware of the scaling problems of the Internet, and since 1982 have lived through a series of mini-disasters as various limits have been exceeded. We have been saying that "getting big" is probably a more urgent (and perhaps more difficult) research problem than "getting fast", but it seems difficult to persuade people of the importance of launching the kind of research program we think is necessary to learn how to deal with Internet growth.
A weakness of the emphasis on product development and the pressure to transition the products to industry is that the products used are more likely to be improved old forms but not new forms or concepts. What is lost is the new that basic research brings forward.
Robert Kahn, a designer of the ARPANET, a co-inventor of TCP/IP and one of founding fathers of the Internet, was ARPA/IPTO director during the 1979-1985 period. Commenting on the problem of focusing on product development and how that affects the basic research pipeline, he explains (Kahn Interview, pg 36) :
It is one thing if you know you are going to buy this micro from Intel and plug it in here, and buy the memory chip and then hook them together. But it is another thing when you are going to create something here, and you are not quite sure what it is or when it is going to be done, or what it is going to look like, or how it interfaces to anything. In fact, the applications got changed by virtue of this process, because it forced them to look at what they could get their hands on now, versus what might show up in the future.
This product-focused orientation leads to emphasis on the product currently conceived and fails to recognize how the new concept will bring forward new products that can't now be imagined.
Vannevar Bush had warned against getting basic research tangled up with industry to avoid confounding basic research with product development. He wrote scientists should be wary of government constraints on their freedom of speech and intellectual pursuits, and that it was also crucial to be wary of the constraints from industry on scientists and scientific exploration. Bush proposed that government refuse to fund scientific work at a university if it had industry funding, especially if the results were to be proprietary. That warning should help to clarify the independence that scientific pursuits must have in order for the scientist to be able to discover the new and create that of which we can't yet conceive. These cautions were directed at the danger of focusing on product development as the orientation, as it is contrary to focusing on and supporting basic research.
Science-Government Partnership
Bush also emphasized that since science creates the new, it is in battle against the old. Thus it brings forth many enemies. Science and scientists need to be protected from those enemies and that is a function for government.
During Licklider's second term at ARPA/IPTO and after he left, there was a hard fight to protect computer science and scientists funded by ARPA. Also during this period important work was done in developing the Internet. This is an important period to study in the stormy partnership of government and computer science.