Around 2014, as Stephen (Steve) J. Lukasik proceeded well into his 80s, he began to consider ways to capture the enormous sweep of activities and history in which he was a key figure. Indeed, that sweep was so broad and often compartmentalized, and his output so prolific, that even his closest associates only knew of slices of his accomplishments. So he began sorting through his career and produced this autobiographical essay on his accomplishments at ARPA that is being made posthumously available now. The essay also reflects the methodical precision with which Steve approached everything in which he was engaged—even his own work. He also completed an exhaustive book on ARPA that he recently completed and remains to be published. —Anthony M. Rutkowski

* * *

There are two criteria used to construct the following shortlist. First, it has to be something to which I made a major personal contribution. Second, it has to be of more than average importance, something one wishes to be remembered. One consequence of these criteria is that one would have had a continued an interest in the area well beyond ARPA, thus supporting its personal significance to me. That said, here is the list.

Nuclear weapons and their control

By age 14, I had already gravitated to physics, via chemistry and biology, as the area most likely to provide me with answers to questions that interested me. The newspaper on 8 August 1945, dealing with the Hiroshima weapon, fixed my academic direction for life. I avidly studied the report by Henry D. Smythe, “Atomic Energy for Military Purposes,” the official history of the nuclear weapon program. Reading “Hiroshima” by John Hersey added a human perspective to the physics. Another book in my library, “The Oppenheimer Case,” by Charles P. Curtis published in 1955 brought me up to date on the “Super,” the decision to move from fission weapons to thermonuclear weapons, a hundred times more powerful. My eventual choice of a job with Westinghouse in their naval reactor program was my first explicit move into what has become a lifetime preoccupation.

Earlier, I explained the accidents that led me into the nuclear test detection field, but in ARPA I was at the interface of the technical issues and the political issues relating to the their importance for near-term national security while at the same time finding ways to control their potentially limitless growth that constituted a “race to oblivion” (the phrase is the title of Herb York’s 1970 book on nuclear arms.)

Steve Lukasik in 1971 during his tenure at DARPA. (Photograph from Lukasik’s archive)When I went to ARPA, I knew that limiting nuclear weapons was not a first priority in DoD, and I did not disagree with that posture. But my personal inclinations on national security issues never impacted my work of limiting/eliminating nuclear testing. I have never had trouble keeping more than one idea active in my brain at any one time, nor have I had difficulty in working on both sides of an issue. In fact, this need to look deeply into both sides of an issue has helped me immeasurably in forming judgments that fairly reflect the at least two-sided nature of any issue.

So, what did I do? The program in NTDO was well and sensibly established, so my role was to build on that base to steer the program toward a conclusion, conclusion in the sense that it would answer the question of how far we could go in verifying whatever test ban treaty we might enter into. I also understood that aside from assessing the limits on detection, I would look at the other side and see how easy, or hard, it would be to successfully defeat the detection technologies and systems we were researching, developing, and thus implicitly recommending.

Oppenheimer Case,” by Charles P. Curtis published in 1955, brought me up to date on the “Super,” the decision to move from fission weapons to thermonuclear weapons, a hundred times more powerful. My eventual choice of a job with Westinghouse in their naval reactor program was my first explicit move into what has become a lifetime preoccupation.

Earlier, I explained the accidents that led me into the nuclear test detection field, but in ARPA I was at the interface of the technical issues and the political issues relating to the their importance for near-term national security while at the same time finding ways to control their potentially limitless growth that constituted a “race to oblivion” (the phrase is the title of Herb York’s 1970 book on nuclear arms.)

When I went to ARPA, I knew that limiting nuclear weapons was not a first priority in DoD, and I did not disagree with that posture. But my personal inclinations on national security issues never impacted my work of limiting/eliminating nuclear testing. I have never had trouble keeping more than one idea active in my brain at any one time, nor have I had difficulty in working on both sides of an issue. In fact, this need to look deeply into both sides of an issue has helped me immeasurably in forming judgments that fairly reflect the at least two-sided nature of any issue.

So, what did I do? The program in NTDO was well and sensibly established so my role was to build on that base to steer the program toward a conclusion, conclusion in the sense that it would answer the question of how far we could go in verifying whatever test ban treaty we might enter into. I also understood that aside from assessing the limits on detection, I would look at the other side and see how easy, or hard, it would be to successfully defeat the detection technologies and systems we were researching, developing, and thus implicitly recommending.

I objected on the ground that we had not evaluated the performance of the array, LASA was an experiment, not a production prototype, and was too close to known U.S. test sites in the U.S. to use our nuclear tests to calibrate its performance at teleseismic distances. Rather than simply argue with them, I announced what ARPA would do next was deploy a next-generation array, NORSAR, capable of seeing both U.S. and USSR underground explosions. This put a halt to talk of operational deployment of a system no one had any idea how well it would perform against seismic events, real and artificial, worldwide.

By the time NORSAR was finished, the ARPANET was up and running, so NORSAR used it to send its data, along with LASA’s to a central recording location in Alexandria VA thus serving to examine optimum signal processing approaches to discriminating earthquakes from explosions. ARPA also jointly funded, with AFTAC, an Alaskan Long Period Array (ALPA) since surface waves, by that time, had emerged as an important discriminant.

Thus, when I left ARPA we had in place much needed new digital seismometers and central processing facilities to enable us to approach signing a CTB on a firmer basis. The growth of worldwide seismic facilities nucleated both a threshold test ban in 1974 and eventually, a Comprehensive Test Ban Organization by 1997. While the U.S. Senate has refused on several occasions to ratify it, the U.S. nevertheless observes its terms. Bottom line for me is that what I went to ARPA to do in 1966, successfully led to a CTB thirty years later.

Information technology, worldwide connectivity, and the man-machine interface

The published paper, “Why the ARPANET Was Built,” tells the particular story of the ARPA work I pushed in networking that led to the Internet. The paper is quite nuanced to reflect that no single person is responsible for the ARPANET, or the Internet, or anything else that represents a major scientific and engineering enterprise. I feel my contribution was helping to wrench this R&D project away from its creators, who would have liked to research it forever. Instead, I put it into the hands of people who could, and would, use it. In part, this was sensible management to avoid holding on the projects for the sake of research and, by moving projects out of ARPA, make room for the new opportunities for military innovation.

We had added packet radio nets for mobile, i.e., military, users and satellite links to provide for global links. The military concept of net-centric warfare was the payoff for DoD, but the entire defense establishment profited from the commercial Internet, which served business, government, and academic communities.

Photograph of DARPA’s directors during 1971 soon after Lukasik had taken over. From left: Gen. Austin W. Betts, Dr. Jack P. Ruina, Dr. Robert L. Sproull, Dr. Charles M. Herzfeld, Dr. Eberhardt “Ed” Rechtin and Dr. Stephen J. Lukasik. (Photograph from Lukasik’s archive)

ARPA pushed, during my tenure, to widen the range of ways people could interact with computers. The RAND tablet for graphic input has been noted. In 1970, on a trip with Larry Roberts to MIT’s Lincoln Laboratory, I challenged them to develop a way for computers to understand speech. My incentive at the time was to assist NSA who had to employ vast numbers of transcribers and translators to make sense of the multitude of communication channels they monitored. We also worked on using artificial intelligence to transcribe manual Morse code, still used on many military circuits. My focus was never on the machines and communication networks, it was on the man-machine interface and use of graphics to present information in understandable way to decision-makers. The need was to match slow intuitive people to the fast idiots we built with microelectronics.

ARPA’s work was greatly enhanced by Xerox PARC, where a number of IPTO Principal Investigators went when Bob Taylor left ARPA and recruited them. I went to Xerox when I left ARPA at Bob’s urging, but left when I found that the XEROX “paperless office” amounted to using digital technology to create more documents to be printed on analog xerographic engines. Most importantly, the Ethernet that forms the basis for LANs was invented there, and this created an enormous number of private nets that then had to be coupled to the ARPANET just as radio and satellite nets had been

Shortly after I left Xerox, I found myself Chief Scientist of the FCC, charged with introducing new technology into the communication industry it regulated. The approach was not to force innovation by regulation, no longer possible in the deregulatory environment that developed, but to encourage it by relaxing regulations and allow new things to happen. Under my tenure there, we authorized the first cellular telephone systems, authorized six direct TV broadcast satellite, introduced the idea of HDTV, and authorized spread-spectrum transmissions that enabled WiFi, Bluetooth, and the panoply of mobile digital services we enjoy (or not as the case may be) of smart phones, iPads, iPods, the Cloud, etc.

ARPA’s robotic programs went into eclipse as Eb dismissed it as “eye-and-hand engineering” but we did a lot in the early days and I am glad that ARPA-encouraged industry, through its Grand Challenge program, to move into replacing delicate and expensive people with cheap and easily retrainable robots.

In addition to networking, my other enthusiasm was to encourage artificial intelligence. In a talk to an IPTO contractors, I drew the usual ARPA expectation of programs that started, peaked, and ended in five years. But I drew a special “curve” for AI. It was a straight line without end. I felt then, and still do that the major utility of computing is not simply faster and more flexible repetition of what humans can do, but to look at problems where human cognition can be enhanced by having a “thought amplifier” on a computer. Unfortunately, this has been hyped into smart cars, smart refrigerators, smart toilets, and, over-the-top, “brilliant” weapons. Hype aside, it is the direction to go and to keep going. Supercomputers and their big data applications are important too but seem to me to be a less exciting, albeit useful, direction.

One last bit of irony. Under nuclear weapons and their control, I recorded the positive things ARPA did to address an important problem created before it was in existence. But realize this. All the useful things mentioned under Information technology, worldwide connectivity, and the man-machine interface resulted in a problem as potentially serious as nuclear weapons: cyber conflict.

Technology is not good or bad. It is the users of technology who direct it to good or bad purposes. The technology of good guys today becomes the technology of the bad guys of tomorrow. People skilled in writing software can either write software for driverless cars (wonderful!), or they can write malware that, when introduced into cars, will cause them to collide (ykes!)

Non-nuclear warfare, aka conventional warfare, or tactical warfare e.g., WWII

ARPA, when created in 1958, was told to solve a space problem. It did.

Then it was handed three problems circa 1960: ballistic missile defense (an unassigned mission from its space program); nuclear test detection so we could sign or not sign nuclear test ban treaties; and counter-insurgency. If you look at these problems, you can say space, ballistic missile defense, and nuclear test treaties are very high-level. Counter-insurgency is, on that scale, a more limited: limited fighting low tech adversaries in someone else’s civil war.

Harold Brown, DDR&E in 1950, assigned ARPA a task in Command and Control also. He had nuclear command and control in mind, but as interpreted by Jack Ruina, it became a project in war gaming and man-machine interaction. It was assigned to computer scientists who then turned it into a subject of computer science and, incidentally, feathering their own nests. When I entered the front office, they revealed none of this to me, but in fairness, they told me what they thought was important, which was to push computer science.

With some digging, by 1967, I found the original command and control charter, and then, without knowing until recently Harold Brown’s rationale, I made the connection between store-and-forward packet-switched networks and survivable nuclear command and control to combat NCA decapitation attacks.

When the very large ballistic missile defense program was moved from ARPA to the Army, and no more Presidential issues were assigned. ARPA was on its own. What was obvious to me was we largely had strategic nuclear programs that had little connection to the wars the U.S. was fighting and was likely to fight in the future. The MAD strategy, in effect, took nuclear war off the table.

But we had pieces that related to current and future wars. Counter-insurgency was conventional, though limited war; Advanced Sensors Office had turned into a parallel counter-insurgency project with more of a hardware approach and less of a systems approach; and the new Advanced Engineering Office was entirely non-nuclear. Also, we had no connection to the Navy, and for the Army, our only connection was in ballistic missile defense. All three offices were subcritical in size and lacked a coherent process for priorities among the numerous mission areas encompassed. So, I rolled them into one, called the Tactical Technology Office, turning three potential liabilities into an integrated office devoted to land, sea, and air warfare.

There are two aspects of tactical technology that closely abut strategic nuclear issues. When two nuclear states are engaged in a non-nuclear conflict, the potential for escalation to nuclear war overshadows every move the states make. And second, continuing nuclear proliferation was creating new nuclear states, thus changing our relationship with them from non-nuclear to nuclear.

Thus, the conduct of non-nuclear war must increasingly relate to escalation control. This puts a somewhat more pointed end to even non-nuclear conflict. So tactical warfare must take on a greater awareness that there is more than just “winning.” One has to do it at the lowest possible level of violence. Thus, limited war, while exhausting, is preferable to bigger wars. Military aid to allies: training military forces of allies to defend themselves; dealing with rapid changes in national leaderships; finding ways to use soft power in lieu of hard power; dealing with sectarian violence that follows different channels from those that figure in major-state political-military calculations; leveraging intelligence to enable precise targeting for allies; use of unmanned systems to limit military and collateral casualties and avoiding inherently escalatory crises over captives and hostages; and the sub-state violence of terrorism now all become missions that are part of tactical warfare.

ARPA work on networking, precision weapons, unmanned air, land, and maritime systems, stealth, theater missile defense, and other foundational technologies conceived in the late 60s and the 70s worked their way into the military services. Thus, when the Cold War ended and was replaced by the need to meet tactical regional contingencies, the U.S. was prepared with next-generation capabilities.

Completing ARPA’s management processes: program initiation and termination

When ARPA was established in 1958, it did not have to figure out what research to initiate. It was told. It examined technically viable approaches to solving problems that were obvious to the naked eye. Neither was program termination a problem because nothing had reached a point of completion through either success or failure.

As told by Mitch Waldrop in his history of ARPA’s information technology, “The Dream Machine,” the early enthusiasm over ARPA had faded, and the free-wheeling styles of Ruina, Sproull, and Herzfeld were no longer appreciated by Secretary of Defense McNamara. ARPA was a fun place with brilliant people, but nothing was coming out of it. By early 1967, the Secretary had lost confidence in ARPA. Deputy Secretary Vance was in favor of shutting ARPA down. DDR&E John Foster, who replaced Harold Brown in 1965, was of the view ARPA was needed but only after a thorough overhaul. Herzfeld left in May 1967 and the sequence of events described earlier ensued. Waldrop’s description of these events is, uncharitably titled, “Waiting for the Grown-Ups.” Eb (Eberhardt Rechtin) and I turned out to be the “grown-ups.” Luckily, I knew none of this at the time, or I might have been distracted. Eb and I simply did what we thought was needed to fix ARPA’s revealed deficiencies. More accurately, Eb was the designated grown-up, and I was the needed insider to connect him with the existing organization. I did well getting him up to speed as he prepared to take up his duties. But the way it worked out Eb only served as Director for two years, 68-69, while I was in his position for five years, 70-74. Despite occasional differences of opinion on details, I thought what he was doing made sense, and it agreed with my views of ARPA’s strengths and weaknesses.

What Eb started, and I finished, was always warmly supported by Johnny Foster. It amounted to adding two critical processes to the ARPA repertoire. The first was the toughest: what should we work on for SecDef or the President if they did not help us set priorities? The second that followed from the first was when and how to stop doing something absent being threatened periodically with being abolished. Budgetary pressures cleared the mind marvelously. In the rush to build ARPA, the input and output stages were omitted from the design. Eb and I had to strengthen the foundation and top off the building without interrupting the work being done on the floors in the between.

What we did has already been described, so I will restrict myself to some high-level observations. One starts an agency to save the country, and no one thinks it is important to establish what “saving the country” means over the long term. Nor was there any suggestion initially of how to assess ARPA’s perceived value to provide continuous and adequate, feedback.

There is an alternative interpretation of the events. Perhaps the period 1958-1968 was a pass-fail exam? Either we were smart enough to survive on our own or we weren’t. Fail, and we are better gone and allow the DoD to try some new approach to keeping ahead of the balance of technology. But while the ARPA of 58-68+ may have been good enough for Cold War, by 1990, the Agency came up for serious review.

What Eb and I put in place in that period was apparently good enough to pass. ARPA worked well through the Cold War. But what about the period since 1990? It is 24 years later, and the Agency is still around, but my insight into its operation is insufficient to know if it continues to meet the military-technical challenges the country faces.

The positive conclusion is that whatever ARPA is doing is still passing the bureaucratic survival test. The proliferation of ARPAs in Homeland Security, Intelligence, and Energy suggests the ARPA brand is still valued.