Foreign Relations of the United States, 1981–1988, Volume XLIV, Part 1, National Security Policy, 1985–1988

57. Memorandum for the Record¹

September 23, 1985

SUBJECT

• Issues in the SDI

I. TECHNICAL STATUS

The SDI Organization itself has both felt and exerted strong pressures to plan early tests of systems components and to give priority to concepts which putatively can be deployed earlier. Associated with this attitude has been a public campaign to establish an aura of rapid progress on all fronts in key SDI technologies, and hence a vague but pervasive expectation of the possibility of early deployment. This campaign is entirely explicable given the hostile atmosphere in which the SDI has had to compete for funds, particularly if one accepts the postulate (which has the status of law within the program) that only impressive large-scale technology demonstrations will ensure congressional support.

Unfortunately, the facts do not reinforce this sense of rapid progress toward early deployment. For instance: the biggest single success of the past two years, the Homing Overlay Experiment (HOE), had been planned for several years by the Army; the experiment was fully designed before President Reagan even took office. For instance: the Navy’s MIRACL laser was transferred to the Air Force, moved to White Sands, and recently accomplished there a (well publicized) kill demonstration not significantly different from the one it did at the San Juan Capistrano test range in 1978. It is true that the beam quality of MIRACL was improved markedly in the past two years, but that improvement finally brings it up to the design criteria it was supposed to have had in 1978; during the same time, the Air Force’s Sigma laser (same power as MIRACL, but of the design type planned for extrapolation to very high-power) failed completely to meet beam quality criteria and was quietly abandoned. For instance: the Army’s Airborne Optical Adjunct has continued to make progress in design of multi-channel infrared tracking systems; but it too is behind original expectations, and is a program several years old. More importantly, work with passive IR techniques has now convinced most SDI workers (and, in a formal report, the Army Science Board reviewing the work) that use of such sensors to discriminate decoys from re-entry vehicles is not likely to work.

Further specific instances will be quoted in support below. The general view is that very little in the way of new technology has actually been developed, although a great many paper studies have been performed. Most of these paper studies have revealed the problems associated with continental BMD to be more difficult than had previously been thought. In short, in many areas the “progress” has actually been negative. Boost phase defense provides a good example of this.

1. Boost Phase. The Fletcher panel² emphasized the importance of boost phase for two reasons: the high leverage of killing missiles rather than RVs, and the necessity of multiplying kill probabilities over several levels of defense in order to produce a 99% effective system. The panel clearly favored Directed Energy Weapons (DEW) for this task. More careful studies since have shown that none of the currently available candidates is likely to come close to this mission, a result well understood in the Foster study.³ Thus, DEW priorities have been changed within SDIO to reflect the fact that lasers are now expected to play a role in boost-phase kill only in the distant future, if at all.

Within the DEW part of SDIO, only one technology to address boost-phase kill remains in the main line of priorities; all others have been “re-scoped” and officially declared “back-ups.” The remaining priority project is ground-based, induction LINAC free-electron lasers (FELs). The project was recently advanced from basic research level to second priority in DEW on the basis of an experimental result at Lawrence Livermore National Laboratory (LLNL) last Spring, in which 40% of the electron beam energy was converted to one centimeter microwave radiation at powers of megawatts. In fact, until May of this year, a ground-based FEL boost-phase killer did not exist even as a concept. In this respect, progress has been tremendous: great expectations have been raised by great extrapolation. But the LLNL result, although certainly an impressive piece of physics and a great advance in FEL milestones, remains very far from meeting even the closest parameters of an FEL weapons system. The wavelength demonstrated at LLNL is 10,000 times longer than that proposed for an FEL weapon; the accelerator must be scaled up a factor of six in energy, while increasing a factor of ten in brightness (a longer accelerator—six times as long—increases the difficulties in simply maintaining brightness, far less improving it); the light beam of the laser must be perfectly trapped by the electron beam, a process not yet demonstrated or even studied experimentally; and the mirrors must be protected from destructive harmonics of the laser light. But even if the laser does work, it will not be 40% efficient, or anything close to that: that number leaves out the efficiency involved in creating the electron beam in the first place; and LLNL concedes that one micron radiation from the e-beam will be much less efficient than one centimeter radiation. Los Alamos estimates that their FEL will operate at 2–3%, and considers FELs’ low efficiency a problem to be overcome in design, rather than an asset. Finally, the LLNL design for a boost-phase FEL killer is for an impressive-sounding 100 megawatt laser. But this power can easily be seen (using LLNL’s own parameters) to be too low by a factor of ten to thirty to accomplish boost-phase kill against the SDIO’s official responsive threat (that is, the threat in the timescale on which FEL could be deployed).

Thus, for boost phase the only DEW remaining in SDIO’s main line is a promising technology with far to go to demonstrate fundamental physics, and much farther yet to reach weapon-scale engineering.

This leaves small rockets (either with HE charges or hit-to-kill) in satellites as the only contender remaining active for boost-phase. The systems architecture studies recently completed typically plan to deploy over 60,000 of these rockets on over 12,000 satellites, about 5% of which would actually be engaged in the boost-phase battle. This notion has been carefully critiqued before (since 1962) and always found wanting; it has great problems of cost, command-and control, complexity, effectiveness and countermeasures, but the key difficulty has always been survivability. Edward Teller and his Livermore staff analyzed this system during the High Frontier studies (of which Teller was originally part; he resigned over this issue) and demonstrated that it was unworkable because the killer satellites can be effectively attacked. This analysis, in fact, resulted in the “Teller Dictum,” that large, expensive space platforms are always cheaper to shoot down than to protect.

The Teller Dictum, of course, applies to all space battle stations, not just KEW ones. Consider only the most likely option: Soviet attack of the defensive system with direct-ascent rockets carrying nuclear weapons (or, cheaper, some of which carry nuclear weapons). Within three years the Soviets will already nave more than 3500 such missiles without expending any additional funds at all. Suppose that we now spend $100 billion to put up a 90% effective (SDI optimistic number) boost-phase layer. This layer will neutralize 90% of the Soviets’ total investment in ICBMs, which is greater than $200 billion. Hence, on a cost-effectiveness basis, the Soviets can afford to expend about $300 billion to buy additional interceptor missiles, and they only have to succeed in knocking down between 5% and 10% of our satellites. Their cost effectiveness leverage is gigantic. No system proposed by or studied by SDIO can stand up to such an attack. Note that this argument does not apply to the critical sensor satellites of a generic SDI, because these satellites are cheap enough to proliferate at many altitudes and small enough to be effectively hidden in space.

Summary: Boost-phase kill is still interesting, but no current proposal seems credible in addressing its critical difficulties. Major breakthroughs in several areas will be necessary.

2. Mid-course ICBMs spend about 3 minutes in boost-phase, and warning and decision time must be subtracted from that. But they spend over twenty minutes in mid-course, with warning and decisions accomplished. The difficulty is that we must discriminate RVs from decoys in the midcourse: this has been the central problem of BMD since the late 50s. If one can identify the RVs uniquely, one might kill them by a variety of means. The HOE demonstrated this, and SDI studies have come up with other promising KEW concepts.

Discrimination remains an unsolved problem, but the most promising thing about “new technology” is that it offers new ways of discriminating. Indeed, many new and interesting ideas have been proposed and superficially studied in the last year or two. Most of them, and all the best ones, involve some form of “active” discrimination: one actually illuminates or perturbs the decoy cloud rather than passively watching it. Many such techniques involve DEWs. Unfortunately, the SDIO has selected neutral-particle beams in space as its central discrimination technology. Such beam generators would be large and expensive, hence subject to the Teller Dictum. And the neutral beam demonstration program is expanding so rapidly it will soon absorb the funds for more ingenious prospects, which are funded at very low levels as it is.

A further part of the problem is that mid-course battle management is the most stressing part of the overwhelming software problems accurately summarized by Professor John Parnas in his letter of resignation as an SDIO consultant. This doesn’t mean that the problem is intractable: discrimination techniques which allow direct identification and homing by interceptors (without keeping track of all the decoys separately) have been proposed. The intractability arises from the form in which the problem is currently posed.

In summary: Mid-course discrimination is the key problem. Solving it will certainly make it possible to breakup structured attacks and defend most military targets; this is also a necessary step in constructing a true continental defense. Many new solutions may be possible using novel technology or even nuclear weapons. The current SDIO is not vigorously encouraging and exploiting these solutions: its current mainline approaches are, at best, unpromising.

3. Terminals. We can now do hard-point terminal defense using interceptors with small nuclear weapons. This is the only BMD technology which has ever demonstrated any military effectiveness; it is by far the least technologically stressful problem in SDI. It is also the one area in which we have totally given up R&D, apparently for political reasons.

Non-nuclear terminal kill, on the other hand, is one of the most difficult of SDIs challenges. It is extremely unlikely that a solution to the problem will be available within the next decade, and quite possible that no solution will be found. The essence of the problem is that incoming RVs are capable of executing extremely sharp turns (ours have done 250 “G” turns and better), much faster than any interceptor can follow. Our engineering has never been able to reduce to zero the miss distance between interceptors and non-maneuvering RVs in clear skies; the possibility that we will solve this problem in the realm of the real responsive threat—maneuvering RVs and nuclear environments—seems remote. We can, on the other hand, probably track and kill RV’s just outside the atmosphere, as the Army’s ERIS program is designed to do, but only if we can first solve the discrimination problem, since the decoys will be present in full effectiveness until the RVs pass down through about 100km altitude.

4. Conclusion. The SDI must be thought of as precisely what the President specified, an R&D program. We do not yet know with any confidence how to accomplish military BMD missions that were not already essentially in hand three years ago. We must protect and nurture the innovative parts of the SDI program to give us the highest chance of finding these options. There are many promising things to work on, particularly in mid-course discrimination and associated kill mechanisms.

II. SOVIET RESPONSES AND DOMESTIC POLITICS

I. Terminal Defense The Soviets now have the only program in state-of-art R&D in terminal defense using nuclear weapons. Their SA–12, which begins deployment soon [1½ lines not declassified]. Since they are fully-tooled with essentially open production lines, they hold a lead-time-to-deployment advantage over the U.S. of about 6–8 years (CIA/Army estimate) in proliferating such defenses outside the Treaty. To the extent that we convince them that SDI will produce an effective national defense, we provide them with a strong incentive to break the ABM Treaty and cash their defensive advantage now.

The most critical part of this problem is that we have no prototype state-of-art terminal system against which to test our penetration aids; so we can have no confidence of defeating Soviet defenses if they proliferate. We should have an operational prototype at Kwajalein.

2. Domestic A strong R&D program has been publicly endorsed even by SDI critics; so have hard-point terminal and even the President’s concept of total damage denial. In fact, three of the SDI’s most effective scientific critics (Richard Garwin, Sidney Drell and W.K.H. Panofsky) signed a document supporting these things. The problems arise from the perceptions of a rush to deploy an imperfect system, or to demonstrate publicly the elements of an imperfect system.

The solution is to emphasize the R&D nature of the program, and to avoid demonstration projects. This does not mean canceling large-scale experiments, but it suggests that they be restricted to actual useful experiments, focused on the most serious technical problems. This focusing will have a far more salutary effect in Congress than any big demonstrations, which actually detract from the SDI’s credibility.

3. Geneva There are several ways in which we can bargain seriously and to advantage in Geneva. We cannot hurt the essential R&D of SDI by reaffirming the ABM and Outer Space Weapons Treaties. We can certainly bargain over (and afford to give up) several different kinds of space-based demos and even kinds of space-based battle stations, for instance space chemical lasers above certain powers. Whether this would be verifiable is not a matter of concern to us: we don’t think such things are workable anyhow. But the Soviets are still worried about them, and this would allow us to place the verification shoe on the Bear’s foot for a change. We can demonstrate good faith and gain much public leverage by bargaining for a range of space-based technologies individually. We should also ask the Soviets to specify what their frequent references to space-based “strike weapons” actually refer to. It’s likely there is something here we could use to advantage too.

III. Management

The SDIO is sorely lacking in staff, both in size and facilities. It also lacks the power to keep the Services from using its funds to satisfy their own prerogatives; this problem is actually getting worse.

Most critically, the SDI lacks a unified doctrine on political, military and arms-control issues, and the SDIO has demonstrated itself incapable of producing one. The OSD level board currently advising LGEN Abrahamson has been totally ineffective in resolving any of these issues. A Presidential-level Advisory Board is clearly required.

Thomas H. Johnson
Director, Science Research Laboratory

1. Source: Reagan Library, Jack Matlock Files, USSR Subject File, 1981–1986, Johnson, Tom Papers [Strategic Defense Initiative]. No classification marking. Drafted by Tom Johnson.
2. Documentation of the Defensive Technology Study Team led by James Fletcher is scheduled for publication in Foreign Relations, 1981–1988, vol. XLIII, National Security Policy, 1981–1984.
3. “National Security Implications of BMD,” (Secret), June 1983. A year-long study chaired by John S. Foster for OSD. The study employed more people than the Fletcher Panel and lasted 4 times as long. [Footnote is in the original.]