On the cessation of US Nuclear Testing
I grow daily to honor facts more and more, and theory less and less. A fact, it seems to me, is a great thing — a sentence printed, if not by God, then at least by the Devil. — Thomas Carlyle, 1836
"...three, two, one, zero-time," announced a calm voice. We watched the TV monitors. The helicopter-mounted camera showed a circle of dust spreading from ground zero, or , "GZ," as we call it, and another mounted on an instrument trailer began to shake. Then nothing. Twenty minutes of it.
"It’s working its way to the top," someone said, referring to the fracturing of the rock. The seismometer trace went wild, and suddenly the ground around GZ slumped, forming the subsidence crater. Other than the mad rush for the next six hours to make sure we got reliable data, that was that. A contained underground test not of a nuclear weapon, but of an experimental nuclear explosive. The public little appreciates that there is sometimes a difference.
The difference is that a nuclear weapon is designed and manufactured so that it can be stored for a long time and, if need be, delivered to a target and exploded. An experimental nuclear explosive may be a complicated, jury-rigged affair, requiring such delicate handling, and containing such limited life-time parts, that we can barely make it work at all even in a controlled environment like the Nevada Test Site. Such a device is not necessarily intended to be "weaponized," as the military calls it. So, why do we bother?
Well, given that the recent dissolutions of the Warsaw Pact and the Union of Soviet Socialist Republics have been accompanied by the re-emergence of regional conflicts together with the proliferation of nuclear weapons and delivery systems, I think it fair to say that peace is not yet at hand. We therefore need to maintain a reasonable capability to deter aggression. It has been argued elsewhere that deterrence is the last third of what it takes to make peace. Here I would like to discuss the role of nuclear testing in maintaining that deterrence.
I consider the nuclear weapons program to be engaged in maintaining and improving the capability, reliability, safety, and security of the nuclear explosives packages in the United States nuclear weapons stockpile. Because these four objectives are incompatible, we must make compromises among them.
Capability is the ability of a nuclear explosive package in a weapon to carry out a mission designated by the Department of Defense, and may be specified in terms of many parameters, including the familiar "kilotons of equivalent explosive yield." Reliability refers to the confidence that the nuclear explosive package will indeed go off when properly triggered. Designed-in reliability is in conflict with safety, the confidence that the nuclear explosive package will not go off or release toxic substances unless it is properly triggered. Safety is related to security, the confidence that nobody but the authorized people can trigger the nuclear explosive package, or exploit it for the nuclear weapons grade materials it contains.
Clearly, the safest nuclear weapon is one which cannot explode under any circumstances. However, it is also the least capable and reliable. Conversely, the most reliable nuclear weapon is one that explodes easily, but nobody would want to handle it. Traditionally, we have designed for safety and security by making weapons that are difficult to explode, and we have confirmed that they are reliable and capable by nuclear weapons testing. This testing includes detonating versions of the explosives package deep underground during the design and development phases of a new weapon (which I’ll call "developmental" testing), underground detonations of sample weapons taken from the stockpile to confirm that they still work after their components have aged ("stockpile confidence" testing), and underground tests of developmental weapons and stockpile weapons to confirm that various safety features or methods of disablement work as intended ("safety testing").
Eventually, some components age to the extent that they must be replaced. This often occurs over a long enough time that manufacturing techniques, available materials, and acceptable safety (including environmental safety and health considerations in manufacturing) and security criteria may also have changed, making it impossible to replace the components, or to re-manufacture a previous design. In other words, we cannot "stay in place" with old weapons and stop designing new ones. There will always be such a "stockpile stewardship" role for a nuclear weapons design program as long as we rely on nuclear weapons as part of our national defense.
You might expect that we could make our design trade-offs using computer simulations. However, the biggest computers available today can be brought to their knees (as it were) by calculations that, by the standards of my lab, are relatively trivial. Consider that the physics in a nuclear device must be calculated for spatial and time scales ranging from those of bulk matter to individual atomic nuclei. Since this is (and probably always will be) impossible, our computer codes rely on approximations. Now, the validity of many of these approximations can only be determined experimentally, i.e., by testing. And some of that testing must be nuclear (our fourth and final category, "weapons physics" testing).
If it seems odd that we would like more data after forty-seven years of nuclear testing, remember that nuclear explosives create a local environment in which good data is difficult to get. Our measuring instruments need to collect their inputs, produce their signals, and get those signals up their cables to our recording shacks while the instruments are being vaporized, and the cables are being crushed. Thus, it is only recently that we have been able to field good enough instrumentation to answer some of our long-standing questions.
Moreover, in the days of the Cold War, developmental testing often edged out weapons physics testing, or relegated it to so-called "add-ons" — devices bolted on to the main test package that used some of its energy to drive physics experiments. This was especially true after atmospheric testing was banned for environmental and public health reasons. The underground tests which followed were much less frequent, because digging those deep holes is time-consuming and expensive. Now our leadership proposes to edge-out weapons physics testing again — this time with safety testing. I favor safety tests, but I think it only prudent to do the physics tests as well, so as to better maintain our ability to make design trade-offs during an extended (and possibly permanent) moratorium.
So, what will happen when we stop testing? Not much, at first. But eventually, it will become time to replace or retrofit an old warhead in the stockpile. An experienced designer, one who has done several tests, will have developed an intuition for how far he or she can trust the computer simulations. Based on that intuition, he or she will make the trade-offs among capability, reliability, safety, and security. But if only inexperienced designers are left (due to the retirements and deaths of the old guard), then the trade-offs will have to be weighted in favor of capability and reliability (which for many applications means bigger, cruder bombs). Safety and security will have to take a back seat. And that is provided the inexperienced designers even know where those trade-off points are and which directions to go from them.
On the other hand, some of my older colleagues who recall the test ban of the early 1960’s fear precisely the opposite situation from the prudence I described above. They fear that without the confrontation with reality provided by testing, the eventual retrofit of our stockpiled weapons will become a game of "Liar’s Poker," in which those with the greatest genius for self-promotion get their ideas included in the stockpile, and that those people may not necessarily be those with the soundest judgment.
If all this seems a little arcane, one of my colleagues puts it all in perspective with the following example: Imagine a shop full of young vocational school graduates who have studied auto mechanics, but who have never actually worked on a car. Perhaps you might let them tune up your car’s engine. But you’d feel a little nervous if they had to redo your car’s entire braking system because you won’t let them use asbestos brake pads anymore.
As indicated above, nuclear explosives testing has been considered a necessary part of the nuclear weapons design program. However, in an attempt to limit nuclear proliferation and reduce international tension, our government has decided to limit further the number and explosive yields of US nuclear tests, and to place another moratorium (we had one in the early 1960’s) on such testing after 1996. 
To get more specific, the "Hatfield Amendment" requires that we stop testing for nine months, restart, do fifteen tests in three years, and then stop for good. The right idea, perhaps, but the wrong approach. A better approach would be for Congress and the White House to ask the National Laboratories, "How many tests do you need (a) to meet military requirements for safety-related retrofits to the remaining stockpiled weapons, (b) to establish confidence in the safety and reliability of the remaining stockpile, and (c) to benchmark key weapons physics parameters, before shelving the test program?" That would at least allow an orderly and sensible shutdown. It would allow us to put the nuclear weapons test program to sleep rather than to kill it.
It would still mean giving up something. The nuclear weapons program could include effort on new designs that achieve unprecedented safety, but whose reliability, capability, and shelf-life have yet to be established. Since an extended testing program would be required to certify such weapons for the stockpile, super-safe designs would not be pursued. Similarly, new designs that achieve unprecedented shelf-life would also be abandoned. Discontinuing the test program means that we must focus on understanding only the open questions regarding what is in the stockpile now, and discontinue new design efforts to improve safety and reliability. In other words, the trade-off for the Laboratory management, the government, and the American people can be put as a question: "Do you want a nuclear test ban sooner — with current levels of nuclear weapon safety and reliability frozen in, or later — with potentially higher levels of safety and reliability?" This of course raises the question, "Is the remaining stockpile going to be safe and reliable enough?"
Well, it probably is. We have gone forty-seven years without a disaster, and we have had very few dud tests. A real disaster would be for us to shut down without doing the weapons physics and stockpile confidence tests that we need to do, rather than just the tests we can do under current law. Our current law is simply an act of Congress, performed without consulting official representatives or nuclear weapons experts from the National Laboratories, the DOE, or the DOD, and signed by President Bush because it was attached to funding for the Superconducting Super Collider (SSC) which would have provided jobs in Texas. Another case of common sense and national security taking a back seat to election year politics.
The nuclear testing moratorium was foreshadowed by the National Defense Authorization Act of Fiscal Year 1989, which mandated a Nuclear Test Ban Readiness Program (NTBRP). In the event of a test ban, the NTBRP was intended (1) to assure that the United States remains able to detect and identify potential problems in stockpile reliability and safety in existing designs of nuclear weapons, (2) to assure that specific materials, components, processes, and personnel are available for the re-manufacture of existing nuclear weapons, or the substitution of alternative nuclear warheads, and (3) to maintain a vigorous research program in areas related to nuclear weapons science and engineering so that the United States maintains a base of technical knowledge about nuclear weapons design and effects. Meeting objectives (1) and (2) is primarily a problem in the management of technology. I would like to see us have a rational program of weapons physics tests, a period of calibration to emerging technologies, to insure that we can meet the third.
The emerging technologies I have in mind can function as "acceptable surrogates", not for development, confidence, or safety testing, but insofar as some aspects of weapons physics are concerned. They include laser fusion and advanced computing technology, among others.
Laser fusion targets have only one thing in common with thermonuclear explosives: they’re supposed to produce energy via thermonuclear fusion, the same process that powers the sun. The fusion part is scaled way down so that Inertial Confinement Fusion, or ICF as it is called, can be used in principle to generate electric power. The idea behind ICF is that multiple laser beams impinge on a little ball of thermonuclear "fuel" causing it to become hot and dense enough to undergo a bit of fusion before it disassembles. ICF offers one arena in which some nuclear weapons physics issues can be explored in the laboratory, rather than underground, provided that we benchmark the relevant weapons physics parameters with nuclear tests before we abandon the nuclear test program. Otherwise, we risk making the "vigorous research program" of the 1989 National Defense Authorization impossible to achieve.
Another part of the changeover to a test-free nuclear weapons program involves the kind of computer simulations that we use. Techniques in massively parallel computing, distributed computing, and scientific visualization are only now emerging. Since we hope that the refined predictive capability of new codes using these techniques will partially make up for the lack of nuclear weapons testing, maybe we should validate the new codes themselves with a few tests.
Finally, several emerging technologies, which I will decline to mention here, can help us to examine the effects of nuclear radiation on critical military and civilian systems even in the absence of nuclear testing. Once again, a bit of calibration may be in order.
Unless we make the change to a test-free nuclear weapons program carefully, the "vigorous research program" of the 1989 Defense Authorization may not be so vigorous. Now in a democracy like ours, the good people leave a program when they get bored. And whether or not you believe we should keep our nuclear weapons, surely it is not in the best interest of national security or nuclear safety to have them designed by second-raters.
In any event, a "test-free" world is one in which only those nuclear tests are performed which have explosive yields below the seismic detectability threshold. If the United States stops nuclear testing altogether, the quality of its arsenal will gradually decline to a lower level of safety, security, reliability, and capability than we have today. The world will not necessarily become a safer place — it may only seem that way as proliferant nations play catch-up with tests camouflaged by ordinary seismic noise. Perhaps we might consider a regime in which a limited number of such sub-threshold tests is tolerated. These would of necessity be physics tests rather than weapons tests because of their limited yields, but such confrontations with reality would still serve to keep our nuclear design program and our capability to maintain our deterrent intact until it is genuinely no longer needed — for reasons pleasant or otherwise to contemplate.
- See "Obscenity and Peace," at this website.
- Hugh Gusterson has correctly identified fielding a nuclear test as an "initiation rite" for a new designer. (See his book, currently entitled Testing Times: A Nuclear Weapons Lab at the End of the Cold War, University of California Press, forthcoming.) However, many who read his analyses forget that every human institution has initiation rites, and that all human knowledge is cultural (i.e., "tribal"). In other words, nuclear designers are no more "tribal" than anyone else. Moreover, this particular rite has its roots in common sense — if you absolutely must count on something to work in a critical situation, you try to test it first. And if you must count on someone to do a job, you try to determine that they can — whether the job is fixing your car’s brakes or your country’s weaponry.
- A number of respected analysts, including David Kay, Secretary General of the Uranium Institute (London), believe that nuclear testing is of such great symbolic significance that the United States must stop if we are to have a credible voice in the effort to prevent the further proliferation of nuclear weapons. Of course, other analysts disagree. I think that our refraining from nuclear weapons testing will have little effect on nuclear weapons proliferation directly. However, it may contribute to nuclear weapons gradually becoming less important in the daily affairs of the world — the "relativization of the nuclear question" as Brian Hehir put it recently — and that I regard as a good thing, for as long as it lasts.
- Now the Clinton administration continues to politicize nuclear weapons testing. It has instructed the National Laboratories to prepare to do one nuclear test in response to the recent (1993) test by the People’s Republic of China. Such an isolated test comes nowhere near constituting the well-thought-out nuclear test series that I would like to see before we continue an extended moratorium.
- It’s probably also in the national interest to keep nuclear design programs alive in some form at both Los Alamos and Lawrence Livermore National Laboratories. The programs could continue to act as critics of each other, avoiding mistakes that "groupthink" could foster, especially when such groupthink cannot be confronted by the reality of nuclear testing.