Chapter 3
Measure for Measure
The Technological Prospect

The revival of interest in strategic defense is more a political act of faith in prospective technologies than an effort to exploit what is already known. As we have seen, the decision to create SDI was made, not to take advantage of progress already achieved, but to force relevant technologies to maturity instead of waiting to see whether research already being conducted would succeed. Virtually all the basic ideas receiving consideration in SDI have been put forward in the past. The most significant exceptions are the X-ray laser and the free-electron laser, both of which rely on more recent inventions. Although progress along the frontiers of research has made some of these ideas more plausible than they were in 1972, when the ABM Treaty was signed, only the most ardent believers in SDI claim that the means are already available for an effective defense against a massive nuclear attack. Most technical specialists not only disagree with such claims but are also skeptical even on the question of whether such a defense will ever prove feasible.

For its political sponsors, SDI is an act of faith based on confidence in the United States' achievements and on a determination not to repeat the mistakes of experts who underestimated technological prospects in the past. Examples are not hard to find. In 1932 Albert Einstein declared emphatically: "There is not the slightest indication that [nuclear] energy will ever be obtainable." Adm. William Leahy, chief of staff under President Truman, prophesied with equal confidence: "The [atomic] bomb will never go off, and I speak as an expert in explosives." In 1924 the well-known British chemist J. B. S. Haldane made a doubly wrong

forecast about prospects for atomic bombs and space travel: "We cannot make apparatus small enough to disintegrate or fuse atomic nuclei, any more than we can make it large enough to reach to the moon." Even so distinguished a physicist as Ernest Rutherford wrote in 1933: "To those who look for sources of power in atomic transmutations—such expectations are the merest moonshine." Vannevar Bush, himself a renowned inventor and chairman of the highly successful National Defense Research Committee during World War II, scoffed at the idea that a rocket could propel an atomic bomb for three thousand miles until it "would land exactly on a certain target, such as a city."^[1]    But the undue pessimism of eminent researchers in the past should hardly encourage politicians to ignore the views of the most qualified and experienced specialists. In a project as complex as SDI, even if the optimists prove right with respect to specific technologies, it does not follow that (1) a defensive system in which many new elements must be successfully integrated will be militarily effective, especially against an equally skilled adversary; or that (2) the vast effort to build and maintain it will be justifiable on economic grounds.

Among scientists and engineers engaged in military projects, there are at least some who see SDI as a welcome challenge to their ingenuity. Enthusiasm for such challenges is at the core of their sense of vocation,^[2]    as suggested by the World War II motto of the U.S. Navy engineers, the "Seabees": "The difficult we do today, the impossible tomorrow." Certain of the younger researchers suggest that older scientists cast doubt on SDI because they are being protective of their work on offensive weapons and find it hard to accept the possibility that their achievements will be made obsolete by a new generation. Lowell L. Wood, Jr., a key figure in the work on the X-ray laser, has emphasized the special challenge of SDI: "Frankly," he told a reporter, "the offensive game, in addition to its somewhat dubious intent, is awfully easy. There just isn't much challenge there. Success consists of shrinking off an inch here and a pound there or moving the center of gravity a half an inch forward. It's distinctly an engineering problem. The intriguing thing about defensive weapons is that they have a real, semifundamental challenge to them—to making them work, work effectively, robustly, and to work at very high cost-efficiency, a high cost-exchange ratio, against the offense."^[3]    This vocational interest of defense researchers in continually doing advanced research, however understandable and even patriotic, is one of the factors that makes the arms race hard to control.

The problem of building an effective defense, difficult as it is in its

own right, is compounded by the need to take account of what a determined and capable adversary can do to overcome it. The complexity of individual elements of SDI has been compared to the challenges of the Manhattan and Apollo projects. But these analogies cannot capture the unique, adversarial nature of the technical problem: "The moon and nucleus," as the OTA study on SDI notes, "did not hide, run away, or shoot back,"^[4]    and the Apollo mission, as others have pointed out, could have been defeated by "a platoon of hostile moonmen with axes."^[5]   

The adversarial nature of this technological endeavor cannot be overemphasized. It is hard enough to devise systems that meet unprecedented technological standards on so many different levels. The difficulties are compounded by the presence of an ingenious and equally dedicated adversary who is determined to nullify or evade such systems. The "red team" may work very hard to concoct potential countermeasures, and the designers may succeed in taking account of them, but will the system actually work in battle conditions? Even mature technologies commonly malfunction under unusual conditions. A complex multilayered defense could not be tested under realistic conditions. Although some countermeasures can be anticipated, there would be no opportunity to make modifications or corrections in the light of actual battle experience, as is usually the case for more conventional military technologies. Certain of the relevant technologies, moreover, will be useful sooner for attacking satellites (i.e., either in an offensive or counterdefensive mode) than for intercepting incoming warheads. Ironically, then, an effort to add defensive systems to offense will, in the short run, only make space-based military assets more vulnerable. The United States currently relies on these assets more heavily than does the Soviet Union.

Nevertheless, the advent of the new technologies and the continuing advances in the older ones present intriguing opportunities. In the simplest terms, they seem to promise a way of putting up a "shield" to ward off an enemy's "sword." Given the difficulty of coping with a Soviet attack by other means, these technological options are bound to be attractive, all the more so because they stand, in some cases, at the forefront of research in the physical sciences. Besides, even if advances do not produce a space shield, they could have benefits for offensive weapons and civil applications. The novelty and technical sophistication of much of the endeavor intrigues investigators. And the political excitement thus generated promotes support for laboratory budgets.

This heady mixture of technological frontiersmanship and political glamour is the crux of the problem. It promotes an unwarranted faith in

technology and a disdain for negotiated solutions. The hope for a technical fix stands in the way of diminishing the danger of nuclear war through arms control. Even so, many informed critics agree that some level of research into defensive technologies is warranted, mostly as a hedge against a possible Soviet breakout from the ABM Treaty, but also because of the possibility, at least over the very long run, that an effective defense may one day actually become feasible.

SDI, however, is premised on the belief that the research process can be pushed into early maturity, but that belief is not widely shared. President Reagan's 1983 speech called for a radical acceleration of research already in progress in order to provide for a deployment decision by the early 1990s. As discussed in chapter 1, the task of designing a technical program that might fulfill the president's goals was assigned to the Defensive Technologies Study Committee, chaired by NASA Administrator James C. Fletcher. After that body of well over fifty defense scientists and engineers issued its classified, seven-volume report, other ad hoc technical groups were formed to develop a more detailed "architecture." The result was the design of a "layered defense" that exhibits both the potential advantages of a defensive shield and the problems of achieving it. In 1987 a committee of the American Physical Society composed of leading defense researchers and specialists issued a review of the research to date on directed-energy weapons.^[6]    Other studies have examined the option of an intermediate or first-phase deployment of a layered system relying only on kinetic-kill weapons for boost-phase and terminal interception. These studies have cast considerable doubt on the feasibility of an effective defense, both in interim and long-range perspective.

The Proposed Technologies

The technologies under investigation fall into several categories: sensors to detect the initiation of an attack, to track the missiles and warheads through the flight path, and to direct the targeting of defensive interceptors; destructive devices to damage, deflect, or incapacitate missiles or warheads in the various phases of flight; computer systems to process the data acquired by the sensors, to make "kill assessments" from the data in order to determine which elements of the "threat cloud" have been destroyed and which remain to be dealt with, to "hand off" data from one layer of the defensive screen to the next, and to perform other "battle management" functions—including almost certainly the

decision to initiate interception; techniques for discriminating targets from decoys ; and a variety of methods for defeating countermeasures that might be taken by the offense.

The effort to develop sensors draws both on well-established technology and on more exotic ideas that have not yet been developed into usable devices. Conventional infrared sensors, which are now routinely used for early warning of attack and to monitor test-firings, are thought to be particularly well adapted to boost-phase interception because they can readily identify the hot plume of exhaust gases emitted during a launch. Because the sensors must be able to detect both hot and relatively cold objects (the hotter the object, the shorter the wavelength), the sensors must have a wide range of capabilities. The ability to identify and track the plume, however, does not necessarily make it possible to track and target the missile, the position of which must be calculated from the signal produced by the plume. The plumes are, in general, very much larger than the missiles, and the missiles are not usually located in any convenient "central position" or easy-to-find "hotspot" within the plume. Other sensors are being investigated, including those that use an ultraviolet probe, in order to target the missile itself. The sensing problem becomes more acute in the post-boost phase, when the rocket exhausts are much smaller, and even more difficult in the later phases of flight, when the relatively cold warheads must be tracked in space against a background of starts. For this phase, more exotic sensors, such as those that make use of laser and radar technologies (and are therefore called "ladar" sensors), have been suggested.

There are three basic types of destructive devices: (1) conventional ground-based ABMs with nuclear warheads , which are directed by ground-based radars to track warheads in the terminal phase of flight (the system in use when the ABM Treaty was negotiated); (2) kinetic-energy weapons (KEW), which are ground- or space-launched projectiles that use the energy of motion to destroy missiles by colliding with them; and (3) directed-energy weapons (DEW), also known collectively as beam weapons because they make use of powerful beams of electromagnetic radiation produced by lasers (infrared, visible light, or X-rays) or beams of highly accelerated particles—either charged (such as electrons, protons, and ions) or neutral atoms.

The conventional ABM could be outfitted with a kinetic warhead rather than an atomic explosive if tracking and targeting were improved to increase the probability of interception. U.S. and Soviet ABMs developed since the early 1960s have been configured in two modes. One permits high acceleration for rapid interception in the atmosphere (as in the

United States' Sprint), and the other has a relatively slower acceleration for interception above the atmosphere (as in the United States' Spartan). Development is now under way on two kinetic interceptors for endoatmospheric and exoatmospheric interception. Nuclear and nonnuclear systems may both benefit from recent improvements, which have enabled the hardening of radar sites and utilize miniaturized, mobile radar in addition to ladar and infrared systems mounted on aircraft.

Kinetic weapons could also be deployed from a constellation of space-based battle stations. These weapons would be accelerated and directed toward their targets either by chemical rockets or by the as-yet undeveloped electromagnetic launcher, or "railgun." The rocket-propelled projectiles would be fitted with homing devices to correct for any errors in targeting; they might also be equipped with explosives that would detonate near the target, increasing the prospect of a kill. The railgun would use an intense magnetic field to impart speeds greater than 20 km per second—at least several times faster than that achieved with conventional rocket technologies, though still far slower than speeds attained with speed-of-light lasers.

Invented in the 1950s, lasers are essentially devices that produce coherent beams of electromagnetic radiation in the form of light. The radiation may be in the visible, ultraviolet, or X-ray regions of the spectrum. "Lasing" ordinarily occurs when the atoms or molecules in the lasant matter (solid or gaseous) are "excited" by energy—electrical, nuclear, chemical, or optical—pumped in from some external source. The resulting beam can be used to damage missiles and warheads either by thermal means (delivery of intense heat that would burn through the skin of the missile or warhead and alter or destroy sensitive electronic components) or by "impulse kill" (depositing energy in a powerful pulse on the surface of the target, driving a mechanical shock wave through the target). Some lasers are better adapted for one type of kill, some for the other. To hit a target from a great distance requires very accurate targeting, a strong beam, and, for most lasers, large mirrors to focus the beam so as to compensate for its diffusion over long distances.

The lasers usually discussed in connection with strategic defense are the chemical, free-electron, excimer, and X-ray lasers. The process whereby chemicals can be combined to produce the energy needed to pump a laser is well understood, but the required energy levels have yet to be achieved. Furthermore, mirrors larger than any so far produced would have to be developed. Without these mirrors, there is no way to focus the energy to produce the brightness required.

In principle, optical lasers may be based either in space or on the

ground. For space basing, weight and the provision of adequate power are major problems. Distortion and absorption of the beam by the atmosphere are the main problems for ground-based lasers. "Adaptive optics" must be employed to direct the beams through the atmosphere so as to compensate for distortion. The beam would be reflected off space-based relay and "fighting" mirrors to targets in space. The free-electron laser (FEL) is a novel concept. It uses a relativistic electron beam (generated by an accelerator rather than by a conventional lasant) to produce an intense, coherent beam of electrons whose energy is then converted into an extremely high intensity beam of light. Two competing FEL designs are currently under investigation. One, the radio frequency (RF) linac (for "linear accelerator"), is being developed by Los Alamos Laboratory and Boeing. Lawrence Livermore Laboratory and TRW are working on the other, called the induction linac. Once a choice is made between the two designs, a ground-based test version is to be built at the White Sands Missile Range in New Mexico.

The excimer laser normally produces a pulse of sharp laser light, theoretically enabling impulse kill, and would require a smaller mirror than other types of lasers because of its shorter wavelength. Both the FEL and the excimer laser require further development in order to determine whether the requisite power levels can be achieved. The weight required for both, and the power required for the free-electron laser, rule out space basing for the excimer laser and the Induction linac FEL, though not necessarily for the RF linac FEL. Chemical lasers of the hydrogen flouride type, which derive their energy from the chemical reaction of these two substances, may be made light enough for deployment in space basing for the excimer laser and the induction linac FEL, though not necessarily for the RF linac FEL. Chemical lasers of the hydrogen nuclear explosion to pump the lasant, which might be in the form of rods each aimed at a missile. Such a laser would also have to be fired from space, because X-rays are absorbed by the atmosphere, but it could also be "popped up" when needed rather than placed in orbit because it could be more effective at great distances than kinetic weapons and would not need to be housed in a satellite, as chemical lasers are. Research on X-ray lasers still has far to go before application can take place, and its testing in space would violate both the Limited Test Ban Treaty and the Outer Space Treaty. The fission-activated light concept (FALCON), under study at Sandia National Laboratory, also aims to use a nuclear reactor to pump a space-based laser. But it is estimated that the platform for this system would weigh between seven and forty tons, apart from any protective armor.^[7]   

In order to design computers that are sufficient for the needs of a space shield, two things are required. First, suitable high-speed hardware needs to be developed. Second, and far more problematic, software programming with enough capacity, redundancy, and reliability must also be produced. This software must be able to handle the large and diverse sets of data and data-processing suitable for an unpredictable environment: some elements of the system might be rendered inoperative, and the system would have to operate efficiently without previous testing or "debugging." The degree to which the system would have to be centralized or could be disaggregated is much disputed. The further development of Very High Speed Integrated Circuitry (VHSIC) technology is essential for adequate data processing. The relatively new gallium arsenide computer chip may also improve prospects for achieving the high-speed computations needed for this purpose.

Means for discriminating targets from decoys in space are still quite uncertain, but ladar sensors have been suggested, along with various types of "interactive discrimination." One such technique would be to use beams of neutral particles. Projected at targets, these particles can help to determine mass and thereby help to distinguish between warheads and decoys. In principle, a beam of neutral particles, when directed at a target, can ascertain mass because the object releases secondary radiation (in the form of neutrons and gamma rays) in rough proportion to its mass. Problems remain, though: it has not been demonstrated that an operational system can distinguish neutrons thus generated from neutrons naturally present^[8]    and that it can be made small enough to be placed in space.

To achieve battle-satellite survivability against efforts to suppress space-based defenses will require hardening the proposed battle stations against attack from lasers and from the effects of nuclear explosions (including electromagnetic pulse, which has been shown in nuclear-test-explosions to disrupt electrical apparatus). Battle satellites may also have to be equipped with counter-countermeasures, such as decoys, protective satellites, rocket motors for maneuvering, and the ability to shoot back. If the satellites carry radar, they would require high-capacity power sources, which in all probability would have to be small nuclear reactors, either carried aboard the satellites or tethered to them.

Adequate launch capacity is yet another essential technological consideration. Space-based elements must be launched into orbit and then serviced and maintained. The need for this launch capacity has led the SDIO to advocate a variety of new high-capacity launch systems. The National Aerospace Plane, if it should become available, is one such system

and could carry much larger payloads than the space shuttle. NASA's difficulties in achieving a reliable shuttle and in maintaining regular launch schedules suggest that the development of this still more complex and demanding launching system will take time, investment, and a further extension of the existing state of the art. The proposal for the Advanced Launch System claims that the system could put payloads into low earth orbit at one-tenth the current cost of $3,00–$5,000 per pound. A Senate staff report has underlined the difficulties involved in achieving this goal: "While the U.S. launched a total of about 350,000 pounds into earth orbit during 1985, SDIO envisions SDI deployment as requiring as much as 5 million pounds in orbit per year. … The capacity of the U.S. to launch payloads into orbit would have to be expanded enormously while the cost would have to come down by at least a factor of ten." Just how realistic this goal is, as the Senate report points out, may be judged by experience with the space shuttle. In 1972 a White House press release predicted that the space shuttle would reduce launch costs from $600–$700 per pound to $100 per pound in 1972 dollars. Currently, space shuttle launches cost $5,000 per pound.^[9]   

A Layered Defense: The SDI Architecture

The architecture, or basic plan, drawn up by the Fletcher Committee envisions a four-layered missile-defense system (see Figure 5). Each layer would be designed to intercept as much of the attack as possible. The attack goes through four sequential phases: (1) the boost phase , during which the great booster rockets bring the payload up to the speed required for intercontinental ballistic flight; (2) the post-boost phase , during which a "bus," or post-boost vehicle (PBV), sets each of several warheads (MIRVs) onto trajectories leading to specific targets; (3) the midcourse phase , during which the RVs coast along inertial paths far above the earth's sensible atmosphere; and (4) the terminal phase , during which the RVs reenter the earth's atmosphere and ultimately explode at or near their targets. Those who believe there is promise in this scheme base their opinion on two major factors.

First, each defensive layer compounds the effectiveness of the others. As the proponents point out, if each layer intercepts 90 percent of the warheads reaching it, altogether they would intercept 99.99 percent of those launched. Thus, if ten thousand warheads were launched, nine thousand would be destroyed in Phase I, nine hundred of the

Fig. 5. Major SDI Sensors and Weapons

remaining thousand in Phase II, ninety of the remaining hundred in Phase III, and nine of the remaining ten in the final phase. Unlike the single-layer missile-defense systems previously considered, if the weapons systems constituting any single layer failed or were overcome by successful enemy countermeasures, the other three could, in principle, largely compensate.

Second, defensive systems have not undergone a thorough review since the early 1970s. And, of course, many of the pertinent technologies (especially lasers, infrared detectors, and computers) have evolved very substantially since that time; important new ones (notably the free-electron and X-ray lasers) have also emerged. SDI proponents claim that these new technical advances open new technological possibilities for intercepting offensive RVs during each of the four flight phases and thus lead to the very high levels of "compounding" that make a real defense against missiles truly feasible. To appreciate both the possibilities and the remaining obstacles, it is essential to examine the technical issues affecting each phase.

Phase I: Boost-Phase Intercept

The boost phase is the first stage of a missile's flight. It includes that part of the missile's trajectory during which the great booster engines of the rocket are under power—that is, from takeoff until the last engine is shut off, typically at an altitude of about 250 km, some three minutes after launch. SDI research has focused most attention on interception during this phase. There are three good reasons for doing so.

First, offensive missiles are in their most vulnerable, or "softest," state during the boost phase. Boosters consist mainly of fuel tanks that must be as light as possible. It is therefore far easier to puncture these tanks or otherwise damage the missiles while they are still in powered flight. Otherwise, either the bus or the RVs themselves must be destroyed during later phases. Second, typical modern strategic missiles carry many nuclear warheads—up to fourteen in one case. Successful interception in the boost phase stops them all and thus accomplishes much more than would the interception of a partially empty bus or a single RV in one of the later phases. Third, the offensive systems are especially easy to identify and track during this phase. Both the United States and the Soviet Union have long had satellites deployed in geosynchronous orbits that are capable of detecting and, in principle at least, tracking the brilliant rocket plumes so characteristic of the boost

phase. Current systems can provide early warning that an attack is being launched, but they cannot provide the kind of tracking information needed for an actual intercept. SDI planners are therefore seeking to develop the Boost Surveillance and Tracking Satellite (BSTS), which would have this added capability. Essentially, the aim is to achieve a very substantial upgrading of the systems now in place in order to make them adequate for this new, more demanding purpose.

After the enemy strategic missile has been detected and tracked, it must somehow be destroyed. There are two quite different approaches to this problem. One is based on kinetic-energy kill weapons (KKW) devices, and the other on "directed-energy weapons" (DEW). Each is further subdivided into a number of specific technical possibilities.

According to SDI advocates, the most promising near-term boost-phase interceptor is one based on the use of Space based kinetic kill verhicles (SBKKVs), designated in 1987 as space-based interceptors (SBIs). As currently conceived, these vehicles would be relatively small (approximately 100 kg) missiles capable of accelerating a small but destructive payload (a projectile) up to velocities of several kilometers per second. They would be equipped with homing devices, presumably relying on infrared sensing, capable of causing the projectile to collide directly with and thus destroy the offensive missile while it is still in its boost phase. Individual SBKKVs would be mounted in groups (perhaps of ten) on a "mother" satellite—sometimes called a garage, sometimes a battle station—that would release them at the proper time. The mother satellite itself would also play a role in overall command and control and perhaps in acquiring and tracking the target as well.

Given the finite velocity of the SBKKV (perhaps 6 km per second)^[10]    and the short duration of the boost phase (currently some hundreds of seconds), the battle stations must be within roughly 1,000 km of their targets at the moment they are needed. This in turn means that they must be in low earth orbits and hence moving at very high speeds, comparable to those of their ICBM targets and the SBKKVs themselves. In order to be ready to intercept an attack at any arbitrary time, the battle stations must be scattered at locations above most of the earth's surface. Only a relatively small fraction of them will be in range of any particular missile launch at the moment it takes place. Thus, many more SBKKVs must be in orbit than would be needed to destroy the total inventory of enemy ICBMs. The interceptors that cannot reach their targets in time are called "absentees."

The "absentee ratio" (the ratio of the total number of interceptors to

the number in position) was much debated during the first years of SDI. Some asserted one particular ratio, and others denounced the use of differing ratios as politically motivated. In fact, the actual absentee ratio is critically dependent on the detailed parameters of the systems involved, including especially the maximum speed of the SBKKVs and the duration of the boost phase of the offensive missile. There is thus no such thing as the absentee ratio but, rather, a very wide range of possibilities from as low as 7:1 for the most optimistic assumptions (implicit in the Marshall Institute report, for example)^[11]    up to values many times that figure.

In view of the obvious importance of the weight and the speed of the KKWs, the current R&D program focuses much attention on both parameters. Most analysts agree that the current weight (approximately 300 kg if the KKWs are outfitted with the current F-15 ASAT payload) must be reduced by ten- to thirtyfold in order for the system to become economically practicable, and that speeds closer to 10 km per second would be advantageous. No physical laws stand in the way of either goal, but the engineering problems are formidable. For instance, for the rockets now available, the final weight of a rocket warhead accelerated to 6 km per second is less than one-twentieth of its initial weight, and the final weight must include the homing system as well as its destructive component.

Other KKW systems under consideration are based on alternative means of producing very high velocity projectiles, such as electromagnetic railguns, but the acceleration and guidance systems for such projectiles are still at a highly speculative stage. For chemical rockets, the final velocity of the payload is limited by practical considerations to values a few times that of the rocket-exhaust gases, which in turn have a practical limit of 3 km per second. In the proposed electromagnetic railgun, no such limits apply, at least in theory, because the projectile is propelled to extremely high velocities by a combination of fields of electrical currents. But adequate means for controlling and steering such projectiles have not yet been invented.

SDI advocates also believe that a number of directed-energy weapons hold promise for use as boost phase interceptors in the long run. These include chemically and electrically driven optical lasers, the free-electron laser, and the new and potentially much more powerful X-ray laser. In principle, the properties of lasers are particularly well suited for such applications. These devices produce extremely intense beams of so-called coherent light or other forms of electromagnetic radiation that

can be focused down into an extremely tight beam—one that spreads out very much more slowly than any other known type of light ray—as it travels for great distances. Within the atmosphere, various scattering processes can cause the beams to disperse, but in empty space beams can travel indefinitely over great distances without much diffusion.

In some cases (the chemical laser, for example) the laser would be mounted on a battle station, which would aim the laser at a particular target and cause it to operate at just the right moment. In other cases (such as the FEL, discussed below), the laser would be mounted on the ground, and the beam produced would be passed to a series of relay mirrors in space, eventually focusing it on the target. Because the velocity of light is for these purposes practically infinite, these systems are not constrained by time-of-flight considerations, as is true of the SBKKV, which must be within perhaps 1,000 km of the target in order to reach it in time.

But even a laser beam spreads out after it travels very large distances. So huge mirrors must be used to focus the laser beam down to an adequately small spot, perhaps a meter or less. The larger the source-to-target distance, the larger the mirror has to be. For typical situations (involving infrared lasers operating in space at a power level even as high as 25 mw), the mirrors would have to be 10 meters or more in diameter. For a configuration in which the lasers are mounted on the ground and their beams routed to the target by relay mirrors, the primary mirror must be even larger. Mirrors up to a few meters in diameter have been used both in space and on the ground, but the giant mirrors apparently required for the purposes of SDI are still only theoretical possibilities. A currently favored approach is to break the single huge mirror up into a large group of smaller ones, making a so-called segmented mirror. Such mirrors must be aligned with each other to an unprecedented degree of precision, however, in order to get them to act in the sort of unison required to produce tightly focused beams. In technical parlance, their output beams must all be "coherent" in order to achieve the focusing that a single large mirror produces. In principle, this is possible, but it has never been accomplished on the scale or in the circumstances required for this application.

The absentee problem also applies to DEWs. Each available laser must therefore destroy many targets sequentially and quickly. To do so, the "dwell time" on any one target, combined with the time needed to "slew," or move, to another one, must be kept very small—seconds or less. This in turn means that the power of the laser—the energy it can

deliver in unit time—must be correspondingly large when used in a BMD mode.

The study produced for the American Physical Society noted that chemical-laser technology is not yet adequate for the demands of boost-phase interception: "We estimate that chemical laser output powers at acceptable beam quality must be increased by at least two orders of magnitude for HF/DF [hydrogen fluoride/deuterium fluoride] lasers for use as an effective kill weapon in the boost phase."^[12]   

The free-electron laser (FEL) is a very powerful device that, in its current versions, is large and heavy and requires great amounts of electric power. In the first place, it involves the construction and operation of an extremely powerful electron-beam accelerator. For typical SDI applications, the energy of the individual electrons must be at least 100,000,000 electron volts, the currents must be from 10,000 to 100,000 amperes (for a total peak power of 1 to 10 trillion watts), and the quality of the beam must be very finely controlled. The beam is passed through a magnetic "wiggler" that produces certain crosswise oscillations in the beam's motion. Under the right circumstances, this "wiggling" beam can be used to generate (or amplify) light to form an extremely powerful and highly focused laser beam. By varying either the beam energy or the spacing of the wiggler, it is possible to vary the wavelength of the light emitted. Thus, in principle, the FEL has considerably more flexibility than most other lasers, including the chemical type described above. The idea is to deploy FELs on the ground and to deliver the laser energy to the targets by means of a series (at least two) of huge relay mirrors orbiting in space. Problems include the construction, maintenance, and operation of these mirrors (including fast slewing). In addition, because the laser is ground-based, the output optics must cope with distortion caused by the atmosphere—the very distortion that causes stars to "twinkle." For the application envisioned here, such distortions, if uncorrected, would cause the beam to spread out to such a degree that it would become totally ineffective.

In brief, the distortion of FEL beams could be corrected by, first, determining the nature and cause of the instantaneous distortions by means of a "reference" beam. The laser's main beam-forming mirror would then have to be distorted so as to compensate for them. Corrections are feasible, in principle—but, again, they have never been accomplished on the scale and in the circumstances involved in SDI applications. In addition to all these other problems, ground-based lasers cannot penetrate cloud layers. Therefore, enough FELs must be deployed in many meteorologically independent locations, each with good enough

average weather so that there are virtually no periods when all might be blocked by clouds. The APS report concluded that for strategic-defense applications, FELs "require validation of several physical concepts"^[13]    thus far only theoretically developed.

The nuclear-pumped X-ray laser is a novel concept dating from the late 1970s. The enormous energy of a nuclear-weapon explosion is used to energize the lasing medium, which creates a laser beam consisting of relatively soft X-rays. The X-rays produced by such devices are readily absorbed by air, however, and are therefore useful only when operated in outer space and fired against objects that are themselves located at extremely high altitudes, generally above 80 km. Research on these unusual devices is said to be making progress, but nearly all authorities believe they are still a very long way from any military application. The total energy that could be produced by such a system is huge and can be divided, in theory, among many independently directed beams. Precisely because the X-ray laser is potentially so powerful and destructive, it would constitute an especially rich and attractive target for any counterdefensive action that might precede a large-scale offensive strike. For that reason, its advocates usually propose that it not be placed permanently in orbit, but be employed in a "pop-up" mode instead. That is, it would be deployed on the surface or in submarines and be lofted into space only when needed. But the pop-up mode requires the system to be forward-based, very near to Soviet territory; the rockets would have to be ready to be launched on seconds' notice.

Because such constraints and requirements greatly limit the utility of such a device in a strictly defensive mode, the X-ray laser is now usually thought of as a potential counterdefensive system—that is, a weapon for destroying an enemy's space-based defenses, including its own counterpart. Thus, the nuclear-pumped X-ray laser might be used in the first phase of an attack to destroy an enemy's space-based strategic defense. Such a strike would assure successful penetration of the follow-on offensive forces. In this role, the X-ray laser would be used essentially as an ASAT, a role that is always less demanding than BMD. The APS study reserved judgment on the military utility of the X-ray laser pending further demonstration of the technical potential, concluding that these devices "require validation of many of the physical concepts before their application to strategic defense can be evaluated."^[14]   

Countermeasures and Counter-Countermeasures Countermeasures designed to thwart all of the boost-phase systems described above have been proposed and were reviewed at the outset by one of the panels of

the Fletcher Committee. Decoys and certain tactical ploys can be used to confuse the boost-phase acquisition and tracking system. In the case of the interceptors themselves, the easiest countermeasure to institute and the hardest for the defense to overcome is the so-called fastburn booster. Indeed, existing U.S. ICBMs and the generation of Soviet ICBMs (in particular the SS-24 and SS-25) now being developed have much shorter burn times than the aging Soviet SS-18, against which the first-generation SBKKVs are oriented. Moreover, by pushing currently available solid-rocket technology to its limits, the offense can, at only a modest penalty in performance, cause the rocket boosters to complete the task of accelerating payloads in less than 60 seconds while still at an altitude of only about 80 km—greatly aggravating the "absentee" problem. In the view of most analysts, the number of battle stations that would then be required for SBKKVs is so high as to become impossible in practical terms. The lower altitude of booster burnout also makes the X-ray laser much less promising: X-rays can penetrate the atmosphere down to only about 80 km. And to counter the type of directed-energy weapons that can easily penetrate the atmosphere, the booster may also be spun about its long axis so as to spread out the incoming laser energy; lightweight protective coatings may also be added for further protection.

It is possible, in addition to deceiving and foiling the defenses, to attack them directly as well. All the space-based components are vulnerable to attack by similar systems deployed by the other side or by so-called space mines; they can be attacked by ground-based ASAT systems as well. As already noted, the pop-up version of the X-ray laser is one example of such a counterdefensive system.

But the story does not end there. All these countermeasures have, conceptually at least, counter-countermeasures. The deployment of more, lighter, and especially faster SBKKVs might mitigate the problems created by the fast-burn booster. Lasers with still more power might come along to defeat the boosters' defenses against earlier DEWs. And the space-based defensive assets could deploy a variety of self-defenses—including proliferation, maneuvering, decoys, and deception—to foil or confound a counterdefensive attack.^[15]   

The basic point is that this process is endless. In a "man-againstnature" contest, such as getting a man to the moon, once the problem is solved it stays solved. In a "man-against-man" contest, once a particular problem is solved, once we have learned to deal with a particular countermeasure, another arises to take its place—and so it goes, on and on,

generally without ever coming to an end. There is no "last move." It is misleading to suggest there might be. Strategic defenses are not something that can be developed and installed at a certain time; they must be viewed as a process that, once begun, never reaches a permanent end.

The forty-year contest between the strategic arm of the U.S. Air Force and the Soviet National Air Defense Force (PVO Strany), reviewed in chapter 2, is a case in point. Throughout that long period, the Soviets have been creating and modernizing a defense in depth against an airborne attack. They have tried to build, in effect, a multilayered air-defense system analogous to that promoted by Reagan and the SDIO for defense against missiles. Over the same forty years, however, the U.S. Air Force has been constantly improving and updating its procedures and technologies for assuring bomber penetration. These have included electronic countermeasures, greater speed, the ability to fly very close to the ground, and a variety of standoff weapons, from the Hound Dog of the 1950s to the SRAM and ALCM of today. At no time during this long period has the total multilayer Soviet defense ever even approached the 90 percent kill-capability that SDI advocates project for each individual layer of the proposed missile-defense system. The U.S. Air Force remains confident of its ability to penetrate Soviet defenses.

There is at present no reason to believe that the final result of an effort to build strategic defenses will be any different in the case of SDI than in the case of the Soviet air Soviet air-defense system. The problem for both offenses and defenses will be far more complex, but the results will not necessarily favor the defense. Michael M. May has hinted at the possibilities in a way that makes the use of the Star Wars label especially apt:

If space-based weapons were to be developed and deployed … all space assets, whether needed for defense or offense, for warning or other purpose, would have to be operated in a hostile environment. They would have to be hardened beyond anything now contemplated, at commensurate cost, or alternatively, be mobile, defended, proliferated or hidden. Space hardware would replicate the characteristics of earthbound military hardware. We would have the space analog of tank corps, carrier battle groups and stealth bombers.^[16]   

Phase II: Post-Boost-Phase Intercept

After the boosters burn out, a post vehicle, commonly called a "bus," launches each of its RVs onto trajectories leading to particular targets. This procedure takes several minutes, requiring the bus to fire small, on-board rocket engines more or less continuously as its course

is adjusted from one trajectory to the next. It is therefore possible, in theory, to detect and track the bus by means similar to those used for the main booster. But the engines of the post-boost vehicle produce much less thrust and energy, so the detection systems must be much more sensitive. Similarly, once acquired and tracked, the bus may be attacked by systems similar to those used for the boost-phase intercept. But, again, the problem of attacking the bus is harder because it is smaller, less visible, and generally tougher than the booster. As the bus deploys its RVs and decoys, interception gets progressively harder, and the third phase begins. Moreover, a bus is not an absolute requirement if independent guidance systems were to be incorporated into each warhead. All of the defensive weapons described above as interceptors for Phase I can be used for interception in Phase II, including some of those SBKKVs on platforms that were out of attack range during Phase I.

Phase III: Midcourse Intercept

After phases I and II are completed, the midcourse phase, lasting some twenty minutes, begins. During this phase the individual RVs (perhaps originally amounting to ten thousand, but perhaps greatly reduced by the interceptions effected during the previous phases) ineluctably follow ballistic trajectories toward the targets. They still emit infrared radiation, but at longer wavelengths and at far lower power levels than during the previous phases. Indeed, the individual RVs are about the same size and temperature as a human being. One may therefore understand the problem by comparing it to the task of detecting a human being just by the heat the body radiates using a device located thousands of miles away.

As in phases II and IV, perhaps the most effective countermeasure in Phase III is the deployment of decoys. As long as the RVs remain in outer space, the decoys may be of very low density—in the form, for example, of metallic balloons. There is no drag, and only the surface characteristics of the decoys—shape, reflectivity, emissivity—need to mimic the real thing. The decoys are usually thought of as weighing one one-hundredth as much as the RV itself, so there could be about a hundred times as many of them.

Those who are optimistic about SDI's prospects generally believe that detection and analysis of the emissions from the RVs and the decoys will enable discrimination. To distinguish the RVs from the decoys, they propose the use of various means of "active" or "interactive" discrimination. One such technique involves the use of laser beams that would

transfer enough momentum to a lightweight balloon to change its velocity by an amount that could be detected by a Doppler laser radar. Another involves the use of a beam of high-energy, neutral hydrogen atoms that penetrate deeply into even the heavy RVs and interact very differently with balloons and decoys. It is also assumed that interceptors similar to those employed in the initial phases would also be used here. For both detectors and interceptors, however, the task is much harder in Phase III. As a result, less attention has been devoted to interactive discrimination, and none of the near-term deployment schemes focuses on it. An elaborate experiment, code-named "Delta 181," was conducted in February 1988, at a cost of $250 million, to collect information on how objects look in space, including potential decoys. Some important elements of this experiment failed, but large amounts of data were produced, which have yet to be fully analyzed.

Phase IV: Terminal Defense

Finally, the offensive RVs, hidden among a cloud of decoys, come to a point where they must reenter the atmosphere. Atmospheric drag then effectively slows down the balloon decoys, and the RVs must proceed on their own, unprotected during the last minutes of the flight—save, perhaps, for a few, much heavier, decoys.

The means proposed for intercepting the warheads that remain at this point are similar to those first proposed in the 1950s. In brief, ground-based radars, perhaps housed in movable vehicles and aided by an airborne optical system (AOS), would detect and track the incoming warheads. The radars would pass the information on to the computers, which would interpret it and then instruct interceptor rockets to lift off and proceed on a course toward the incoming RVs. When in range, the interceptor missiles would release a KKV, which would use a passive or active homing device to intercept and destroy the incoming RVs by impact or explosion nearby. This intercept system might be extended into the late midcourse regime if there were no cloud of decoys at that point. The current plan for accomplishing terminal defense includes two contemplated missile systems, the High Endoatmospheric Defense Interceptor (HEDI) and the Exoatmospheric Reentry Vehicle Interceptor System (ERIS). HEDI is designed as a high-acceleration weapon that would intercept any remaining warheads. The objective would be to destroy the warheads as high in the atmosphere as possible because they could be salvage-fused to explode on contact; an explosion at a low altitude,

even if not yet on its target, would still do severe damage. ERIS would also be ground-based but with slower acceleration and longer range; it could intercept warheads in late midcourse, just before they reenter the atmosphere.

Countermeasures against this type of defense have been known for a long time. They include exhausting and saturating the defenses, using heavy decoys, countersimulation, precursor nuclear bursts, and the like. If a terminal defense were the only layer deployed, such countermeasures would probably work. But if it were deployed as the last of a four-layer system, it could probably destroy the remnants of even a large attack, depending on how the other layers had functioned, of course.

Battle Management

In any defense system, the various elements have to be alerted and generally "told what to do" before they can even begin to perform their functions. Always difficult, this problem is compounded enormously in the case of the multilayered system contemplated under SDI. A master command, control, communications, and intelligence (C³ I) system, in this case usually referred to as a battle-management system, must not only alert and instruct the first layer and initiate the entire defensive process, but it must also somehow instruct each layer in sequence, "handing off" the problem from one layer to the next as the attack progresses. In addition to the usual problems such systems face, a special problem, called "kill assessment," assumes importance here. Many of the proposed defensive systems either leave the RVs and other objects more or less intact, but nonfunctional, or break them into large fragments that may still look threatening. The overall battle-management system must somehow determine which RVs have been put out of action—that is, it must make a kill assessment and pass this assessment on to succeeding layers in order not to waste ammunition over and over again on objects that are already out of the battle. (The problem of kill assessment has been combined by SDIO with surveillance, acquisition, and tracking in a program package called SATKA.) Large quantities of data from the sensors and the battle satellites must be rapidly processed and the results assimilated and communicated to the satellites and to ground controllers.

The entire supersystem, battle management included, must perform all of its functions correctly the first time under actual battle conditions, and it must have an overall sensitivity threshold such that the entire defensive process is certain to be set in motion when the alarm is real

while, in the event of a false alarm, nothing at all is done. And there will be many opportunities for false alarms, especially if both sides deploy space-based defenses. Under such circumstances there would be a great many objects in space, and plenty of launchings from enemy territory to set the alarms ringing. Even if only one side has space-based defenses, operating and replenishing them involves events that might trigger operation of the entire system in a sort of autoimmune-like reaction.

Another unique problem of battle-management derives from the extremely short response times following a launch warning. The response times for space-based defenses are very much shorter than those associated with ground-based BMD. The fifteen- to thirty-minute response time for the latter allows for at least the possibility of human intervention. The record shows that, so far, when false alarms have been sounded,^[17]    the human operators of the system have used the time available to determine that the alarms were indeed false, and thus avoided a catastrophic reaction. In the case of space-based defenses, response times drop to minutes or seconds or even less. Moreover, the entire system must be fully automated if it is to have any chance of responding in a timely manner. This obviously means that the false-alarm suppression mechanism must also be both purely automatic and much more finely set than it would ever need to be for ground-based terminal defenses of the classical kind.

The need for quick reaction poses the question of whether the defensive shield would have to operate automatically, without a human being "in the loop." SDIO officials have assured Congress that an "affirmative" human decision would be required before any lethal element of a space shield could be activated. Congress has mandated, in P.L. 100-180, that SDIO must not develop command-and-control systems that would make it possible "to initiate the directing of damaging or lethal fire except by affirmative human decision at an appropriate level of authority."^[18]    In order for an interception to be made successfully, however, the time for such "affirmative human decision" will inevitably be extremely short—on the order of minutes or even seconds. The "appropriate level of authority" is therefore likely to reside not with the commander-in-chief but with a designated subordinate, very likely a military officer directly in charge of the space shield who would have been "preprogrammed" to know what to do in a variety of contingencies.

The battle-management systems must be designed and programmed so as to avoid the difficulties and to solve these problems. It has been estimated by experts that the necessary software program would involve

ten million or more lines of code. SDI opponents point out that no such program has ever been constructed and that experience would indicate that even if it could be built, it would be rife with untestable and undetectable errors.^[19]    Proponents say the software could be assembled in smaller pieces, which could probably be tested adequately or otherwise made "fault-tolerant." In such a fast-moving field, there is no sure way to predict the outcome of this issue. Clearly, however, the experts are hardly in agreement that the battle-management problem can be solved. One acknowledged expert, Frederick P. Brooks, has said he sees "no reason why we could not build the kind of software system that SDI requires with the software engineering technology that we have today."^[20]    Others share the view bluntly expressed by Robert Taylor, formerly the director of computer research programs for DARPA, and currently head of the research center at the Digital Equipment Corporation: "I think it's pretty clear that it can't be done. The goals of the SDI put demands on software that are just absurd in terms of the state of our knowledge."^[21]    After examining the problem in detail, the Office of Technology Assessment concluded that even if the system could be designed and built, it would be highly prone to failure: "In OTA's judgment, there would be a significant probability (i.e., one large enough to take seriously) that the first (and presumably only) time the BMD system were used in a real war, it would suffer a catastrophic failure"—defined in the study as "a decline of 90 percent or more in system performance." The OTA also found that no existing software system (such as the long-distance telephone network or the Aegis ship-defense system) provides an adequate model for developing, testing, producing, or maintaining the software required for a BMD system.^[22]   

The Testing Problem

One problem with building the sort of system required by SDI is that it could not be tested in a fully realistic fashion. Individual components could, of course, be tested against real U.S. or imaginary Soviet equipment and countermeasures, and the operation of the total supersystem, as well as each of the various layers, could be simulated with computers. A special project intended to do just that (called the National Test Bed) is under construction. In it, the various components of the system will be electronically connected to each other as they are developed, so that the whole problem can be dealt with in an incremental fashion. On the one hand, the ability to make such simulations and use them to predict

real world events reliably is rapidly improving. On the other hand, the complexity and unpredictability of events in a real battle are truly unprecedented. It seems very likely, therefore, that in the race between the continually growing complexity of the problem and the ability to simulate it accurately—a race which is inevitable in this sort of man-against-man contest—complexity would always be ahead.

A major human paradox is also at work here. If a comprehensive defense were deployed, many defense researchers would be eager to announce the discovery of grave flaws and weaknesses in it. These analysts are apt to be the same ones who tend to find some grave Soviet threat just around the corner (claims that the Soviets are on the verge of decisive breakthroughs in strategic defense technology are current examples). Even if a robust shield is built, and even if it could be reasonably expected to work as advertised, critics will come forward with claims that it is unreliable and that the other costly activities needed to maintain the old-fashioned kind of deterrence must still be supported. The irony is that those who will find the flaws in even a robust strategic defense are apt to be the same researchers who are now urging the United States to proceed with SDI before the Soviets do it first. They alternately encourage and denounce because they have a professional interest in all forms of military technology. They take pride in being able to find the flaws in existing systems and in figuring out how to get beyond the current state of the art. Although they will readily admit that nontechnical policy considerations must ultimately determine what is to be deployed, they tend to support political decisions favoring the development and deployment of the weapons systems with which they are, or could be, associated. They are thus the natural allies of those who are predisposed to see a never-ending Soviet threat that requires continuing improvements in the technology of warfare. A space shield, no matter how elaborate, cannot be protected from zealous defense researchers who want to breach the next frontier lest the adversary do so first.

The Aps Study

Shortly after SDI was announced, the leadership of the American Physical Society (APS) decided to undertake some sort of major study designed to elucidate the physics issues involved in the Fletcher Committee's report. After some hesitation on the part of this leadership, a plan eventually emerged that called for the establishment of a special ad hoc group to study and report on what was currently known about the

sciences and technologies underlying the various directed-energy weapons (DEWs) then under serious consideration by the SDIO.

The study group was cochaired by two distinguished physicists, Nicolaas Bloembergen, a Nobel laureate on the faculty of Harvard, and Kumar Patel, a senior official at the AT&T Bell Laboratories. They were joined by fifteen others, all directly involved in one or more of the sciences and technologies concerned. A review committee (including York) chaired by George Pake, president of the Xerox Corporation Laboratories, was also established. Among the members of the two groups was Charles Townes, who had received a Nobel Prize for inventing the laser technology in the first place. Also included were three experienced researchers from the nuclear-weapons laboratories (two from Sandia and one from Lawrence Livermore), one from the Air Force Weapons Laboratory at Albuquerque, one from the U.S. Military Academy at West Point, one from the Jasons group (which for many years had been studying most of the advanced ideas relevant to the various schemes), and other experts from a number of first-rank academic and industrial laboratories, all engaged in work in science and technology relevant to the overall project. Perhaps most important in this context, several of the members had direct and deep experience in directing the conversion of advanced scientific ideas into large, complex engineering projects. Such experience is not common among physicists generally, but it is essential in understanding the practical problems and time scales that are always involved in the adaptation of advanced physical principles into working devices and systems.

A few of those involved in the study had publicly indicated their opposition to the Strategic Defense Initiative as it eventually took shape, but most had taken no position on the matter, and all were prepared from the beginning to make a serious and politically unbiased study of it. The anti-SDI biases that had already become evident from a poll of the members of the National Academy of Sciences, in the actions of so many academic physicists in pledging to boycott SDI, and in such public interest groups as the Union of Concerned Scientists (UCS) and the Federation of American Scientists (FAS), were carefully and deliberately avoided in setting up this particular study and review.

The APS team worked closely with the president's science advisor (then still Keyworth) and the SDIO staff in elaborating the scope of the work and the means for carrying it out. This cooperation continued throughout the entire study period. The study group was given classified briefings concerning the work then in progress, including projects at the

nuclear-weapons laboratories, but the final version of the report itself was written so as to be entirely unclassified and thus available for the widest possible distribution and use. Because of these arrangements, the final report was submitted for clearance to the SDIO before being published in April 1987.

In the words of the report itself, it "concentrated on the physical basis of high intensity lasers and energetic particle beams as well as beam control and propagation. Further, the issues of target acquisition, discrimination, beam-material interactions, lethality, power sources, and survival were studied."^[23]    The study and the report did not review any of the several varieties of kinetic-kill weapons (KKW) then being proposed. The members of the study group and the government officials directly involved agreed that this limitation would lead to a better and more effective result. The study also did not attempt to estimate the cost of the programs that would be necessary to achieve any particular level of defense capability. And most important, as the review committee noted, the study "does not, and could not, address the global question of whether and on what scale or at what pace the United States should proceed with programs intended to create strategic defenses."^[24]    The decision to eschew consideration of strategic and policy issues was reached by the members of the study group themselves, their primary rationale being the simple conviction that consideration of these issues would needlessly undermine the credibility, and hence the value, of the rest of the work. All of these limitations were thoroughly discussed with the governmental parties as well, and everyone involved understood what they meant and intended.

The study group's work is summarized in one general finding and twenty-six specific conclusions about as many different elements of the total DEW program. (Several of these specific conclusions were quoted earlier in this chapter in the section on SDI architecture.) The general finding focuses on the current status of knowledge in the sciences involved, with particular attention to the most important gaps in it:

Although substantial progress has been made in many technologies of DEW over the last two decades, the Study Group finds significant gaps in the scientific and engineering understanding of many issues associated with the development of these technologies. Successful resolution of these issues is critical for the extrapolation to performance levels that would be required in an effective ballistic missile defense system. At present, there is insufficient information to decide whether the required extrapolations can or cannot be achieved. Most crucial elements required for a DEW system need improvements of several orders of magnitude. Because the elements are inter-related,

the improvements must be achieved in a mutually consistent manner. We estimate that even in the best of circumstances, a decade or more of intensive research would be required to provide the technical knowledge needed for an informed decision about the potential effectiveness and survivability of directed energy weapon systems. In addition, the important issues of overall system integration and effectiveness depend critically upon information that, to our knowledge, does not yet exist.^[25]   

Unfortunately, in publicly presenting the report, the APS council allowed the strongly anti-SDI views of some of its members to be expressed as though they reflected the views of the study group. In a statement issued only one day after the report itself was made public, the council observed that "the SDI program should not be a controlling factor in U.S. security planning and the process of arms control" and argued against early deployment of any SDI components.^[26]    This bit of advice was a nonsequitur at best, inasmuch as the elements of the early deployment schemes then being considered most seriously were based on the use of KKWs, which had been specifically excluded from the review.

The Controversy Over The APS Report

Given the highly charged political atmosphere surrounding SDI, it is not surprising that the manner in which the council presented the report stimulated some very strong reactions to the report itself. Fortunately, neither the president's science advisor nor the senior officials of SDIO confused the council's statements with the report itself. Although they clearly would have preferred a more optimistic tone, they accepted the report with reasonably good grace. Shortly after the report was issued, Louis C. Marquet, deputy director of SDIO for technology, declared in a press interview: "I think, frankly, that they carried this study out in a very responsible fashion. … I frankly think that both of us gave each other A's. … There was nothing in their report which says we're completely out of our minds, that something is beyond the laws of physics."^[27]    A handful of the most fanatical advocates of the project had a different response. Ordinarily, the opinions of such a small group would not matter very much to the general public. But given the highly charged politics and exuberant rhetoric that have surrounded SDI from the beginning, their reaction had at least a short-lived impact. The core members of the group were Lowell Wood of the Livermore Laboratory; Gregory H. Canavan of the Los Alamos Laboratory; Angelo Codevilla,

a fellow at the Hoover Institution and formerly a member of Senator Wallop's staff; and Frederick Seitz, a former president of the National Academy of Sciences and then chairman of the official SDIO advisory committee and of the unofficial Marshall Institute study group. A few congressmen well known for their strong advocacy of the president's initiative worked closely with the group.

The counterattack on the APS study was launched at a special seminar and press conference staged by some of SDI's congressional advocates. At these events Seitz said that the report was not worthy of serious consideration: "I know of no precedent, in my long association with the American Physical Society, for the issuance of so seriously flawed a document as this." It contains "numerous errors, inconsistencies and unrealistic assumptions," he said, that are, "as far as we can tell, always in one direction—such as to make the plan for defending the American people against a Soviet nuclear attack seem more difficult than it really is."^[28]    The thrust of this remark, to the effect that the APS report reflected the putative political biases of its authors, also permeated the remarks of the others who spoke at that seminar and press conference. Seitz also criticized the report for ignoring kinetic-energy weapons, despite the clear understanding on the part of the study group and the relevant government officials that this type of weapon had been omitted deliberately. Later, Seitz elaborated on his views by saying that the whole experience reminded him of the 1930s when the German scientists adjusted their scientific views to conform to the demands of the Nazi leadership.^[29]    Wood added that the executive summary of the report was written "with a political goal in mind,"^[30]    and that he and Canavan noted that none of the six "very senior and eminent reviewers" had worked on the technological areas at issue "for at least a quarter of a century." They added that two of the six had publicly opposed the SDI before the report came out and two others had privately expressed reservations about it.^[31]    Wood later changed this to "five of whom had taken public positions against SDI."

This same group also found fault with a number of technical details in the report and presented papers describing its flaws at various congressional seminars, hearings, and press conferences already cited. The APS study group considered all of the charges, conceded that the report contained some ambiguities, and issued a measured but firm reply: "On the whole, we stand by the findings of the Report, and we consider the arguments posed in these two [Wood and Canavan] papers to be without merit."^[32]    The details of these charges and replies are not easily summarized

or paraphrased, but perhaps the flavor of the situation can be conveyed with the following examples.

One of the issues in controversy was the prospect for chemical laser weapons. Wood and Canavan charged that the study group had misstated the facts. In particular, they noted that the executive summary of the APS report had asserted that such lasers had been tested to date only at a power level of 200 kw, whereas in fact lasers of this type had been operated at more than a megawatt. Thus, the technology of chemical lasers did not have to be extrapolated nearly as much as the APS study had estimated in order to achieve the brightness needed for space-defense applications.

The APS group replied by first noting that its original draft had actually presented a higher value for the laser power levels already achieved experimentally, but that in its last full meeting, the group was informed by SDIO officials that for reasons of security the conclusion must read 200 kw. Even so, the statement in the report was not as misleading as it might at first appear because the higher power levels, while actually achieved, were produced in lasers whose output beams were substantially less well focused than those needed for the ultimate application, and the particular laser involved in the higher power experiments "cannot be scaled to significantly higher powers" in the range of those needed for the ultimate application.^[33]   

Wood and Canavan charged that the report's conclusion that a ground-based excimer laser would require a larger source of power—1 gigawatt (1 billion w)—was inconsistent with its own calculations, which should have led to the conclusion that only 6 mw (6 million w) were needed. The study group responded by pointing out that the critics confused the calculations for impulse kill with those for thermal kill. For an impulse kill, the report estimates that a 100-megajoule laser would be required: Canavan arrives at a lower requirement by assuming a smaller diameter spot on the target, based on a much shorter engagement range and a larger mirror than the study group considered workable. For a thermal kill, where energy efficiency is much lower, the APS found that power of 1 gigawatt (gw) would be necessary—a calculation not disputed in the Canavan study. Finally, the group noted, SDIO's own actions in relegating the excimer laser to a "back up" technology indicated disagreement with Canavan's assertion that excimer lasers "could be legitimately scaled in a single step to the levels required, from present modules."^[34]   

Wood and Canavan claimed that a 4 mw free-electron laser would be

adequate for purposes of strategic defense, and could operate at 40 percent efficiency. Thus, only 10 mw of delivered power would be required, rather than the 1 gw the APS calculated. The study group replied that Wood and Canavan reached their conclusion by assuming that the efficiencies achieved in an experimental "ETA accelerator" at Los Alamos could be achieved in FELs configured for the more demanding requirements of boost-phase intercept. "To our knowledge," the group reflected, "no one—including the SDIO itself and the laboratories, including Los Alamos and Livermore, building the FELs—currently predicts efficiencies for such devices that are anywhere close to those obtained on ETA."^[35]   

Another issue raised by Wood and Canavan concerns the power supply needed for space-based satellites. The APS report noted that nuclear reactors would probably be required to provide "station keeping" or "housekeeping" power for satellites carrying surveillance and directed-energy kill mechanisms. Wood and Canavan contended that the study group had greatly exaggerated the need for electrical power by assuming that a continuous supply of between 100,000 and 700,000 watts would be needed for each of the one hundred satellites in the hypothesized architecture. Although that level of power could only be met by nuclear reactors, "the power needed for satellite housekeeping," according to Wood and Canavan, "is not hundreds of thousands of watts, not for any of the satellites considered in any of the 'baseline architectures' by the SDI program." The satellites under actual consideration, they claimed, would require only a few thousand watts, which are "routinely supplied" by solar photovoltaic cells and storage batteries. Higher levels of power would be needed, they pointed out, only if the satellites were to include radar units, "but the SDI has no plans for radars on any of the satellites in any of its baseline architectures."^[36]   

The study group replied that the upper limits of its estimates did in fact take account of the power requirements for satellites bearing radar units,^[37]    evidently on the assumption that in some proposed configurations, radar units would be mounted on at least some of the satellites. The lower limit, the study group acknowledged, was "about a factor of two higher than the estimates provided for us by SDIO officials … not an unreasonable disagreement on such a speculative engineering project." The SDIO's lower limit of 50,000 watts, the group noted, was itself considerably higher than the several thousand watts posited by Wood and Canavan.^[38]   

Did the APS estimates grossly exaggerate the power needs defined by

the SDIO? Since its inception, SDIO has been cooperating with NASA and the Department of Energy in a joint project known as "SP-100" (for Space Power - 100 kw). SDIO's interest in the project reflects the finding of the Fletcher Committee that "the overall success of certain concepts is highly dependent upon the ability to generate tremendous amounts of electrical power."^[39]    In its 1986 report to Congress, SDIO observed that SP-100 "is the cornerstone of the research and technology effort seeking long-term continuous power supplies." The project was needed both "to provide moderate continuous power levels for a variety of projected SDIO needs" and for other civil and military space missions.^[40]   

In order to determine the official administration view of this matter, Rep. Edward Markey (D., Mass.) asked the Department of Energy to assess the objection raised by Wood and Canavan to this part of the APS report. On behalf of the department, Under Secretary Joseph F. Salgado replied that he had to "take strong exception to Dr. Wood's claims. There may be some people who have the same view as Dr. Wood, both within the external to SDIO, but the official documents provided to us and the decisions reached in concert with the Director of SDIO indicate that higher power levels are required." In a detailed commentary attached to the letter, the department noted that it "strongly disagrees with Dr. Wood's contentions regarding SDIO's 'housekeeping' power requirements," because current SDI studies indicate requirements 'from a few 10's of kWe to over 100 kWe for non-burst power duty cycles." The commentary noted that "the overall SDI architecture is still under development, and therefore the power requirements cannot be precisely defined," adding that "history would indicate that the power requirements will rise with changing mission requirements." Although some SDI staff specialists had expressed interest in lower power levels, "SP-100 technology was baselined in all of the recently sponsored SDI space power system architecture studies for providing 'housekeeping' power."^[41]   

Whatever the eventual outcome with respect to power requirements for a space-based system involving directed-energy weapons, the SDIO's statements and the view expressed by the Department of Energy made it obvious that Wood and Canavan, not the APS report, grossly misstated the officially adopted parameters of the SDIO with respect to projected power levels.

These exchanges are typical of the controversy between Wood and Canavan on the one hand and the majority of informed specialists on the other. The latter are more skeptical about the prospects for strategic defense. The details of the arguments put forward by Wood and Canavan

are frequently accompanied by polemics and ad hominem remarks that serve only to confuse the nonexpert and make the argument as a whole hard to follow. Although polemics are not unknown in exchanges among partisans on such issues—and anti-SDI forces have used similar tactics—in this case the attack misfired and only weakened the optimists' position in the eyes of other technical specialists.

The Proposal For Early Deployment

Very soon after Reagan launched the Strategic Defense Initiative, proposals for early deployment rose insistently in certain defense-oriented quarters. Indeed, the early deployment of the High Frontier version of the SDI was being strongly advocated by Graham and his supporters in and out of government even before the March 1983 speech. Given the fundamentally political nature of the whole affair, and especially the intrinsically political objectives of those who advocated early deployment, it is not surprising that this particular subissue evoked some especially strident controversy. Indeed, even within the defense establishment, the meaning of the term "early deployment" was itself debated. Clearly, in the minds of some of the more politically oriented advocates, the drive for early deployment was stimulated primarily by the desire to get as much of the project as possible committed during the Reagan administration in order to make it difficult—even, they hoped, impossible—for a future president to back away from it. But if the purpose of the drive for early deployment was clear to the politicians, it was not so clear to the technologists, including those who are sympathetic to the basic idea.

To the authors of the Marshall Institute report, early deployment meant abandoning what they called a "business-as-usual" approach to the initiative. Deployment of a partial "three-layer defense" system based on KKVs could, they said, begin in seven years, or in 1994, if the government used "streamlined management and procurement procedures," while under "business as usual" conditions, "deployment of the full 3-layer defense cannot commence until the late 1990s."^[42]   

The highly respected defense expert Robert R. Everett, who was chairman of the Defense Science Board's Strategic Defense Milestone Panel and had been president of the Mitre Corporation, saw matters differently. In a 1987 memorandum to Under Secretary of Defense Richard Godwin, he remarked—plaintively, it seemed—that "the term early deployment, which is sometimes heard, appears to mean only that a first

phase would necessarily be earlier than later phases and not earlier than previously suggested. In any event, current plans and decisions deal only with continued research and development, and deployment will come later."^[43]   

Despite this uncertainty about what, if anything, early deployment might mean, plans for going ahead with the deployment of "phase I" of a combined ground and space multilayer defense system were continuously being elaborated. As of the fall of 1987 the Defense Acquisition Board advanced six SDI technologies to the "demonstration and validation phase," recognizing their potential roles in a first-phase deployment:^[44]   

Oddly, as of summer 1987 HEDI was not on the Defense Acquisition Board's list of phase I projects. As SDIO director Lieutenant General James A. Abrahamson, Jr., put it, "We think the other layers have more

advantage and contribute more to the stability equation and deterrence equation."^[45]    HEDI is included in most other unofficial versions of early deployment, including that described by the Marshall Institute (it is, in fact, the recommended third layer; without it there would be only a two-layer system).

In April 1988 the SDIO accepted the Everett panel's recommendation calling for the reordering of Phase I priorities to emphasize space surveillance (in the form of the BSTS and SSTS projects and communication). The panel recommended against inclusion of space-based interceptors and called instead for consideration of an initial deployment of one hundred ground-based, long-range interceptor missiles similar to the proposed ERIS missiles, but possibly larger in order to provide greater range. These missiles would be deployed at the existing Safeguard site in Grand Forks, North Dakota, and would be intended to provide site defense for a missile field, as contemplated in the ABM Treaty, and only a very thin area defense. This proposal resembles the "Accidental Launch Protection System" proposed for "debate and serious exploration" by Sen. Sam Nunn (D., Ga.) in January 1988. Notably absent from either the Everett panel recommendation or the Nunn proposal was any endorsement of the early deployment of space-based kill systems. As a result of the adoption of the Everett panel recommendations, research on neutral-particle-beam devices for midcourse discrimination and the testing of the space-based "Alpha" chemical laser will probably be delayed.^[46]   

On the basis of the Everett report, the Defense Acquisition Board recommended in June 1988 that the plan for Phase I deployment in the late 1990s—which the board had recommended, and Secretary Weinberger had approved, a year earlier—should be reassessed. Just prior to receiving that recommendation, the under secretary of defense for acquisition, Robert Costello, reportedly sent a memorandum to Abrahamson on May 27, 1988, outlining new, more modest objectives for SDI, indicating, as a report in Science magazine observed, "that the program is being brought more tightly under the control of the civilian managers of the Defense Department." Previously, as the report noted, Abrahamson "had broad authority to set the goals and structure of the program and he reported directly to the secretary of defense." In keeping with Secretary of Defense Frank C. Carlucci's decision that the planned Pentagon budget would need to be cut by $300 billion over the next five years, SDI was coming to be regarded not as a sacrosanct program but one

that would have to be compared with other research priorities.^[47]    Nevertheless, SDIO planning continues to assume that deployment, in several phases, will take place as research continues (see tables 1 and 2).

Indeed, the more exuberant technological optimists continue to believe that the technology is now in hand to build a ballistic missile defense that would have significant strategic benefits. Indeed, the thrust of the High Frontier study was that this technological plateau had already been reached in 1982. The Marshall Institute report, reportedly written with the benefit of briefings by SDIO staff members, expresses no doubts whatever that a three-layer system "93 percent effective" against a "threat cloud" of ten thousand warheads and a hundred thousand decoys could be built in seven years for $121 billion and operated for between $10 billion and $15 billion per year.^[48]   

The Everett panel reached a very different set of conclusions. Reporting some months later than the Marshall Institute did, this panel had been asked by the under secretary of defense for acquisition to review the SDI program and to comment on the state of SDI technology, systems design, costs, organization, management, and readiness to move toward deployment. (The Everett memorandum cited above was sent on the occasion of the completion of this review.) "Much remains to be done," the panel concluded, "before a confident decision can be made to proceed." It also observed that "a number of significant technological problems remain to be solved. Cost estimates are, therefore, highly uncertain." It went on to list "the principal pieces of missing technology": (1) the technology for the survivability of the SBKKV bus, (2) targeting the rocket hard body (i.e., the booster) in the presence of the rocket plume, (3) the ability of the passive infrared detectors on the probe and the SSTS to discriminate anything but the most primitive decoys and debris, and (4) the technology for the manufacture of the very large IR focal planes (i.e., the basic component of the infrared telescopes that determine the effectiveness of the probe, SSTS, and improved BSTS)—by no means a trivial list. The panel's report goes on to note that there "is little information on how objects look in space or how rockets look in boost phase. Component and systems design are proceeding on the basis of assumptions and calculations which may or may not prove reliable."^[49]    It should be noted that all these cautionary remarks came from people who were, in the main, sympathetic to the project. Critics could easily suggest a number of other major missing pieces, but this list is nevertheless formidable as it stands.

Project insiders had other misgivings about early deployment, as evidenced

(table continued on next page)

(table continued from previous page)

by a report drafted by three analysts at the Livermore Laboratory.^[50]    The report points out that the optimistic projections (e.g., the Marshall report, in particular) are based on calculations involving the characteristics of the Soviet SS-18, the most ponderous and vulnerable of all current Soviet missiles. Changes in the Soviet force structure, now under way, will greatly complicate the problem of interception. The report also stated that possible future modifications of Soviet forces, including some that are less extreme than those projected in the Fletcher report, make interception by SBKKVs of the type now being developed impossible. For example, the Livermore report concludes that whereas 20,000 interceptors in orbit could intercept 90 percent of the currently deployed enemy RVs during the combined boost and post-boost phases, about 100,000 interceptors would be required by the mid-1990s, given the DOD's projections. The projected threat consists of missiles and warheads whose development and deployment must have been well established even before Reagan's "Star Wars" speech. The so-called near-term responsive threat (that is, the threat designed and planned in response to the speech) is widely estimated to be such that the kind of SBKKVs the United States now knows how to make would have virtually no intercept capabilities at all.

The issues raised by the Livermore study may be further clarified by analyzing them in terms of the fraction of SBKKVs able to reach their targets after a suitable launch warning. According to the Livermore group, in the case of Soviet SS-18s intercepted by SBKKVs accelerated to 6 km per second, only 2.5 percent of the KKVs would be available for boost phase interception, 13 percent for interception before completion of the post-boost phase. (This latter figure agrees, at least roughly, with the conclusion presented in the Marshall report; i.e., that a force of 11,000 SBKKVs can handle an attack launched by 1,600 ICBMs.) For missiles like the U.S. MX and the Soviet SS-24 (a new ICBM now being deployed), these percentages fall to 1.3 percent and 9.5 percent respectively. For missiles like the current U.S. Minuteman and the Soviet SS-25 (yet another new missile) they drop to 1.6 percent and 2.3 percent respectively. In other words, fewer than one-fifth as many SBKKVs can reach an SS-25 as can reach an SS-18. For the Fletcher report's so-called fast-burn booster, these percentages are both zero.^[51]   

There are still further difficulties that can easily be derived from the Livermore analysis. Recall that in the case of the SS-18, 13 percent of the SBKKVs are able to reach at least the post-boost vehicles, or buses, before they finish off-loading their RVs, but only 2.5 percent can reach

the boosters before burnout. This means that 80 percent of the interceptions ([13 — 2.5]/13) take place during the post-boost phase. Three important problems result. First, the bus is smaller, tougher, and much easier to decoy than the booster. The demands on the SBKKV are therefore much more severe than they are in the boost phase. Second, the rocket exhaust during the post-boost phase is far dimmer than in the boost phase. This means that the surveillance and tracking system, both in the BSTS and on the KKV itself, will require much greater extrapolations from "off-the-shelf" items than the optimists usually imply. Third, and probably most important, the buses are continuously discharging RVs all during the post-boost phase. Even if an SBKKV succeeds in destroying a bus, on average, more than half of its RVs will already be en route to the targets. And to make things even harder for the defense, the attacker could redesign the post-boost system so that all the RVs were released simultaneously rather than serially, thus greatly reducing the engagement time. A Senate staff report echoed these findings in even sharper terms:

Based on our briefings, it appears that SDIO is designing its Phase I space defense against an optimistic version of the Soviet threat. They appear to take a relaxed view of the smorgasbord of response options available to the Soviets to counter SDI. Phase I is being designed to address the Soviet threat of the mid 1990's, yet it probably will not even begin to be deployed until the late 1990's and will have the bulk of its deployment life in the following decade. The far more sophisticated threat environment of that later period would appear easily capable of defeating the Phase I system. In short, Phase I likely would be obsolete the day it was deployed.^[52]   

These skeptical assessments are important not because they indicate there is anything especially defective in the way optimists tend to present the prospects for successful boost-phase interception, but because these complexities are typical of all elements of SDI. Without exception, the various components under development in the project all turn out to be (even ignoring countermeasures) much more difficult and much further from current capabilities than simplistic claims make them appear.

The Livermore report went on to point out—as have other proponents of strategic defense in general—that changes the Soviets are already making to take account of possible U.S. defenses present less of a danger to the United States than the current Soviet offense poses. The Soviets have had to reduce the size and lethality of their forces in order to assure penetration of potential U.S. defenses. Thus, the Livermore analysts say, "the deployed defense has resulted in significant reductions

without firing a shot."^[53]    This effect, in fact, parallels what the Soviets have achieved with their (enormously expensive) air defenses. U.S. responses to the U.S.S.R.'s continuing improvements in its air defenses over the years have involved the substitution of penetration aids (e.g., standoff missiles, electronic countermeasures, etc.) for some of the explosive megatonnage previously carried by bombers. Thus, the Soviets have achieved a considerable reduction in the force of a potential U.S. attack "without firing a shot." SDI puts the United States in a similar position. The big question, of course, is whether the construction of extremely expensive defenses is the best way to accomplish such a reduction. Negotiated arms control may be much easier to achieve and is obviously a good deal less expensive.

Clearly, there is a very wide diversity of views, even within the defense community, about the status of the relevant R&D; about the meaning and prospects of what is already known and in hand; and about the feasibility, cost, and value of any kind of early deployment. In every instance—and even among those who favor SDI—the better-informed and more competent the group, the more cautious and hedged the claims.

A More Modest Proposal: ALPS

A much more limited defensive deployment was raised as a possibility meriting consideration. In a speech in January 1988 Senator Nunn proposed the "Accidental Launch Protection System" (ALPS), reminiscent of the 1967 proposal of a "thin" Sentinel system designed to defend against an attack from a small nuclear power such as China or from an accidental launch. ALPS calls on the defense planners to "seriously explore the development of a limited system for protecting against accidental and unauthorized launchers."^[54]    Such a system might involve the basing of one hundred ERIS launchers at Grand Forks, North Dakota, the site of the one decommissioned ABM system allowed under the ABM Treaty. An upgraded version of the old perimeter acquisition radar might be used, possibly along with AOS sensors. The ERIS missiles, as previously noted, are designed to intercept warheads in late midcourse, relying on built-in infrared sensors to home in on targets, once guided to the vicinity by data from other sensors.

Such a system might be deployed in compliance with the ABM Treaty, except for changes that could be negotiated, perhaps in the Standing Consultative Committee, to allow, for example, for mobile radars or

airborne sensors. The developer of ERIS, the Lockheed Corporation, estimates that such a deployment would cost $3.55 billion, although other observers put the estimate at more than $16 billion, taking into account auxiliary systems as well as the cost of ten years of operation and support.^[55]   

This system might be effective against a small or accidental launch of ICBMs, but it would not be effective against shorter-range, depressed-trajectory missiles launched by a submarine, which have a much shorter flight time. Protection against shorter-range missiles would require additional deployments at several sites, in violation of the ABM Treaty. ALPS could not provide protection from bombs delivered by more conventional methods, including air attack. Furthermore, a serious problem with reliance on ERIS is that, according to U.S. intelligence data, Soviet nuclear warheads are now salvage-fused so as to detonate on impact. If one such warhead were intercepted by an ERIS missile, the resulting explosion would complicate and perhaps prevent the interception of any remaining warheads released by the same accidentally launched missile. As Aviation Week & Space Technology reported: "Remaining ERIS missiles would be unable to find their way to the other warheads lofted by an accidentally launched missile. They would have been rendered 'dumb and blind' by the first blast's radiation, which would paralyze the microelectronics indispensable to the system's command and control."^[56]   

From "Smart Rocks" To "Brilliant Pebbles": Another Supposed Last Move

Another proposal, put by Teller and Wood to Reagan, George Bush, and other high-ranking government officials at a classified White House briefing in July 1988,^[57]    would change the emphasis of SDI research from "smart rocks" housed on large satellites (as proposed for Phase I) to "brilliant pebbles" (technically known as "singlets," or small, self-contained kinetic interceptors) orbited in very large numbers to intercept a missile attack.

This proposal amounts to a resurrection of the BAMBI project of the 1950s discussed in the previous chapter, updated to take advantage of the very considerable progress made since then in miniaturization and computer power. According to Wood, each of the "pebbles" would contain its own optical sensors in the form of silicon microchips ("eyes

which look out for targets"), a small but very high performance computer on the level of a CRAY-1 supercomputer ("a brain to recognize targets"), and fuel for propulsion ("legs to execute the brain's target seeking commands"). Each device would weigh only between 1.5 to 2.5 kg, and cost, when produced in quantity, "about $20,000 per pebble, assembled and tested," or "perhaps $50,000 each in the early 1990s." The cost of deploying each weapon, including launch costs, would be around $100,000. To suppress roughly one thousand Soviet ICBMs—"with maximum clustering of mobile launchers in the worst case imaginable"—would require about one hundred thousand of these brilliant pebbles. The total cost of such a deployment would be $10 billion, or, if actual production costs prove higher than estimates by the usual multipliers, $30–$50 billion—still, in Wood's words, "an eminently affordable strategic defense system." Furthermore, because these "Stingers-in-the-sky" would rely only on nonsensitive technologies, the system "could be fully shared with the Soviet Union with minimal complications for national security."^[58]   

A similar concept was put forward earlier by the physicist Richard L. Garwin, who noted that for "mid-course intercept, a microminiature homing kill vehicle, a 'hornet,' could be a very effective defensive weapon." Garwin was careful to point out, however, that even a vast swarm of such hornets could be rendered less effective by countermeasures. Decoys would confuse the sensors. Offensive RVs could be enclosed in relatively large balloons, making targeting of the enclosed warheads more difficult. As a result, "there would be little kill or kill assessment." Space mines could be orbited to shadow and destroy the hornets—and the mines could be even cheaper to deploy than the hornets.^[59]   

Garwin is skeptical of the new proposal not only because it leaves such potential countermeasures out of account but also because it is being advanced with the same overblown confidence that characterized earlier schemes: "The audacious moniker, 'Brilliant pebbles;' the argument that because there is no law of physics known to the proponent to prevent the desired performance, then it is achievable; the breath-takingly optimistic schedules for capabilities that would have enormous commercial and military significance if they could be achieved—have all been seen before from the same source."^[60]   

Indeed, all that can be said at present about this ingenious idea is that it may or may not prove workable and cost-effective after research and testing, which will surely take much longer than anticipated. Even if it

proves feasible, however, it will hardly dumbfound a determined adversary. If all else fails, even a belt of sand looped around the equator would do considerable damage over time to these pebbles, or hornets—and, of course, to every other vehicle in space. To guard against such countermeasures, the defender might be able to add shielding to his weapons, but that would make them considerably heavier and add to the cost, whereupon the adversary would devise different countermeasures, and so on, ad infinitum—so long as those who are asked to pay the costs are gulled into believing there can be a last move in a technological arms race.

What Might Come Of This Research? Four Possibilities

What are the possible outcomes of the current program? We will here consider just four possibilities. In each we explore what would happen if the current R&D program clearly pointed to one of the following possible outcomes: (1) There is great promise ; it begins to look as though Reagan's vision of a truly effective space shield may eventually be fulfilled. (2) There is some promise ; it begins to seem possible that a somewhat effective, but not impenetrable, system can be designed that meets the "Nitze criteria" of survivability and cost-effectiveness at the margin. (3) There is little promise ; the situation continues to look the same as it does today, that is, very dubious at best. (4) There is no promise ; it soon becomes clear that the whole thing is a wild goose chase.

Case 1: Great Promise

Suppose that in the next few years the current R&D program shows that there is a substantial possibility of eventually building a strategic defense that would be reliable; survivable; cost-effective (even in the light of the currently foreseeable chain of responses and counterresponses); and on top of all that, leakproof, or very nearly so. If a majority of researchers and key political leaders become convinced of its ultimate feasibility, the resulting course of action is easy to imagine. The United States would, and probably should, go ahead with an accelerated effort, arguments about transition problems and costs notwithstanding. Even in this case (one seen as extremely unlikely by well-informed analysts), the political authorities would still be well advised to make no other changes in strategic policy until it becomes completely clear that this wondrous

outcome could be achieved soon. This admonition applies equally to U.S. arms-control efforts, alliance arrangements, and plans for modernizing the strategic forces—assuming, of course, that these policies are all currently correct as they stand. To the extent that the remainder of U.S. strategic policies make sense now, they will continue to do so for at least the foreseeable future and should not be substantially changed on the basis of a very improbable, even if highly desirable, outcome for SDI. Even if the researchers' most optimistic expectations are fulfilled, developing and deploying a multilayered defense will take decades. In the interim, it would be most imprudent to behave as though the outcome were a foregone conclusion.

Case 2: Some Promise

Suppose it soon becomes widely apparent both inside and outside the defense establishment that the current program may eventually lead to strategic defenses that are reliable, survivable, and cost-effective, even in the face of the first round of likely countermeasures—but, alas, clearly far from perfect. That is, suppose it remains as obvious as it is today that even if a BMD system fulfills the Nitze criteria, there would still be so many pathways through it and around it that the United States and its allies would continue to be threatened with great, probably total, destruction. Most analysts believe that even this more modest outcome is very unlikely; but, conceivably, it could come to pass. Let us suppose it does.

Then, as in Case 1, the United States would, and under certain conditions probably should, go ahead with the program on an accelerating basis. The usefulness of cost-effective but imperfect defenses has been much discussed. In general, even quite imperfect defenses can make preemptive attacks more uncertain and more difficult. If, for example, active defenses are deployed to protect the retaliatory forces, the attacker must increase the size of his strike in order to bring the potential result back to the level that had existed before the deployment of defenses, and he cannot be fully confident of doing so even then. (So-called preferential defenses, to be described in the next chapter, in which only certain specific, but unidentified, units of the retaliatory forces are in fact defended, greatly exacerbate this problem.) The same consideration applies to the defender's command-and-control system. If an attacker believes he knows where the defender's vital control units are located, then, in the absence of defenses, he can at least calculate that a certain

level of attack would destroy them all. In the presence of even partially effective defenses, he can no longer be sure of doing so. Given the size of today's forces, even a relatively small remnant is easily sufficient to threaten annihilation of the attacker's cities and population, and this remains so even if the attacker has the same imperfect defenses. Thus, improving the survival chances of even a modest remnant of the defender's forces reinforces deterrence.

In sum, the potential value of imperfect but robust and cost-effective defense lies in the fact that such defenses would in general reinforce deterrence. Thus, the probability of an attack would be reduced even if the defenses could not adequately blunt the attack on cities and population. There are, however, alternate means for reinforcing deterrence. Besides, simply meeting the Nitze criteria is not enough by itself to justify building active defenses. These alternative means in general are also based on increasing the survivability of the various elements of the deterrent forces, including especially the command-and-control system. For the immediate future, the most promising appear to involve better protection of national-command authorities and the further application of mobility, dispersal, and other forms of deception to render retaliatory forces untargetable. Active defenses, if they are to be deployed, would in general have to be cheaper than the alternatives to make them worthwhile.

But if strategic defenses were to hold promise of eventually surpassing the Nitze criteria by a substantial margin, and not merely meeting them on an equal cost basis, the United States would probably choose to go ahead with them even if there were other ways to make offensive forces more survivable. In such a case, the deployment of strategic defenses might well lead to a world in which there was progressively more emphasis on defense than on offense. Some sort of useful transition away from the current dreadful situation—in which peace is based mainly on the threat of mutual suicide—might occur. Such a transition is too far off and too speculative to foresee its details, but it might include such things as substantial arms reductions, a defense-protected build-down, and a general shift away from the current high-strung nuclear confrontation with all the extraordinary dangers inherent in it.

Such an outcome also seems to us and to most defense analysts to be very unlikely. But the possibility, however slight, that it might emerge is one of the important reasons that many of the defense experts who oppose the current SDI program, with all the rhetoric and politics that have surrounded it, endorse a substantial, continuing research program

in all areas of defense technology, including its newest and most exotic forms.

Case 3: Little Promise

Suppose things continue as they are. Avid proponents continue to project great results and SDI activists in the defense establishment try to lock the program into the political system by making major spending commitments and promising some sort of early deployment. The opponents of the program in the wider technical community, and the doubters inside the defense establishment, both of whom form substantial majorities in their respective milieus, continue to see little promise for fulfilling even the Nitze criteria—except, possibly, for the special sub-case of ground-based, hard-point terminal defenses. What then may we expect to happen?

This is, perhaps, the most likely case, at least for the near term. It is also the one most susceptible to the play of domestic and international political forces. We see two distinct possibilities if this case prevails. One will apply as long as the United States is governed by a president who has the same passions for this approach to strategic issues as Ronald Reagan. In this particular scenario, the project will continue at about the same expenditure level it reached in 1988—roughly $4 billion per year. Attempts to increase this expenditure or to commit the nation to early deployment of space-based systems will likely fail. The present organization, built largely so that the R&D program could be rapidly expanded and made to evolve quickly into an early deployment, will live on even so.

The other scenario will apply if the United States is governed by a president who hasn't Reagan's passion in the matter. In that case, the project and its supporting organization will evolve as in Case 4 below.

Case 4: No Promise

Suppose the view of the opposing majority takes hold in the post-Reagan years. The force of the president's idiosyncratic desires, in addition to the personal loyalty to Reagan of the majority of the Pentagon staff, were surely major factors supporting the SDI and the SDIO. It is therefore at least possible if not in fact probable that without Reagan in the White House the entire edifice will collapse. The package of programs called the SDI may be disaggregated and treated as they were before

the March 1983 speech. Even during Reagan's tenure, the funding for strategic-defense programs as a whole did not grow very much beyond what was called for in the plans and programs in place before the speech. Total expenditures for SDI for FY1984–89 will be just over $17 billion, compared with $14 billion projected by the DOD before SDI was declared. The really big differences before and after the speech were that the U.S. and international bodies politic were flooded with grand rhetorical promises about a radically new future, and that a special office, the SDIO, reporting directly to the secretary of defense and by-passing all normal staff offices, was set up. Neither of these two changes is essential or normal to the conduct of an R&D program of the type likely under either Case 3 or Case 4. With this peculiarly dedicated president out of office, they could therefore both easily disappear. Such a denouement is what many observers expect. It is also what most SDI contractors fear and (privately) prepare for.

The technological difficulties in the path of SDI are formidable, but they are by no means the only ones. The strategic issues that surround a commitment to develop and deploy a shield in space involve still more complex uncertainties. We examine these in the next chapter.