Rethinking Nuclear Deterrence

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

“We will complete a nuclear posture review that opens the door to deeper cuts and reduces the role of nuclear weapons.” - President Obama to United Nations, Wednesday, September 23, 2009

IS DETERRENCE, EVEN NUCLEAR DETERRENCE, A ZOMBIE SITUATION? Herman Kahn used to argue that there were three kinds of problems in the world: (1) those that are insoluble by human intervention; (2) those that are made worse by human intervention; and (3) those that are solvable by timely and sufficient human intervention. My argument is that the problem of how to achieve nuclear nonproliferation and nuclear disarmament is an insoluble problem as long as nuclear deterrence continues to provide the basis for military defense. Military defense as the primary means of achieving national security is also expensive (~$60,000 billion in current U.S. dollars consumed over the past 64-years). Essentially, we are playing a game of nuclear deterrence rationalized as a means of military defense that may represent a zombie situation (i.e. the game is unwinnable). Thus, if this is so, it is not only expensive to continue playing this game for capital might be allocated more productively to other activities for national security, it may be highly risky to continue with a nuclear posture that is no longer of strategic value. For example, what if, unlike the game of nuclear deterrence as a means for national defense that may no longer be considered of strategic value for national security, the problem of mitigating global warming was considered of strategic value for national security? That one engages in the enterprise of allocating capital to mitigate climate change does not depend whether it is a resultant of either natural or man-made forces. It is occurring. It is happening with great rapidity. It is unbelievably expensive in its impact on the world's economies. Significant reallocation of capital is required to adequately address this problem adequately in a timely fashion. The consequential systemic risk is very, very large. I would argue that a responsible calculation of this systemic risk (probabilistic forecast of occurrence and consequences, reduced to an estimated economic value) would be greater than that risk associated with either future conventional wars or potential unconventional (e.g. terrorist) violence (i.e. the supposed raison d'être of national defense) Thus, from a systemic risk perspective, if we are willing to support policies that result in the expenditure of ~$60,000 billion over the next 64-years for military defense as presently constellated (this is the present trajectory!), might we also be willing to support an equal or greater allocation of capital for the amelioration and mitigation of the very real, economic costs of global warming? This, of course, serves to illuminate the faulty thinking of some present national security policies. It is highly unlikely that the world’s economy is able to support two

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 1 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

equally strategic initiatives of ~$60,000 billion each over the next 64-years. Either future policy initiatives will continue to be ruled by non-analytical reasoning, convention (what has happened in the past serves as the primary guide to future choices; the world is ruled by stationarity), and convenient, often self-serving rhetoric, or conversely, sound analysis and considered, rational policy choices will begin to rule the day. If the problem of responsible and twenty-first century national security initiatives are to be addressed adequately in a timely fashion, the question on the table is: “Which shall we choose?” THE STATED OBJECT OF DETERRENCE IS TO AVERT WAR Deterrence as developed through national defense strategy, despite its overt military focus, touches practically everything and everyone. It’s success, or lack thereof, either establishes or destroys the platform for cultural stability and human development. Its cost of implementation affects the global economy and diverts capital from other uses. If too much capital is diverted from civilian needs, technical innovation, economic prosperity, and the means to improve general welfare of the nations will be harmed. DETERRENCE BASED ON NUCLEAR WEAPONS IS A FOOL’S GAME National defense in the U.S. and many other countries of the world, either directly or through nuclear umbrellas of nuclear states still relies on nuclear deterrence. This is based on the maintenance of large stockpiles of hydrogen-nuclear weapons and launchers capable of delivering a devastating counterattack against any opponent foolish enough to launch an attack. Nuclear weapons have not been used in the 64years since 1945 and there's never been a nuclear, or even a nonnuclear, war between two states that possess them. Yet, there is data that nuclear weapons do not deter war between states with them and those without nuclear weapons (e.g. Korea, Vietnam, Gulf War, Iraq War, Af-Pak War). The foundational assumption of the value of nuclear deterrence was originally based on MAD (mutual assured destruction), a strategy developed from economic game theory. This strategy assumed a Nash Equilibrium would be achieved between two players. It would be in neither player’s interest to engage in a First Use attack, if the result is a counterattack that destroys the attacking side. With the appearance of multiple parties possessing nuclear weapons, MAD morphed into various flavors of the same game: Assured Destruction, Massive Retaliation, Minimal Deterrence, etc. A growing number of nations now believe it is beneficial to possess nuclear weapons, or the capability to make them in the future, as needed, to deter aggression by adversaries with superior forces. The resulting counterattack, although not as decisive as in the original MAD game, would still be potentially catastrophic to the attacker.

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 2 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

Thus, nuclear deterrence provides an impetus for nuclear proliferation, potentially a more volatile brinksmanship defined decision-space, and the ever increasing possibility that nuclear weapons may fall into the hands of terrorists who may have nothing to lose by First Use, as no credible counterattack could be launched against them. The problem with much of the thinking concerning nuclear deterrence is that it is based on technological positivism and the logical fallacy of retrospective determin-

ism. It also presumes stationarity - past history is a guide to the future. QUESTIONING THE BASIS FOR NUCLEAR DETERRENCE Technological positivism presumes that the correct number of nuclear weapons analytically derived will achieve deterrence and extended deterrence objectives. The fallacy of retrospective determinism attributes nuclear deterrence, post ipso facto, as the sole or primary cause for no nuclear exchanges between nuclear powers over the past 64-years. The assumption of stationarity imagines that along with the other assumptions of nuclear deterrence: (1) approximate parity so that neither side has an advantage for First Use, (2) both sides have enough information that they can make informed, rational decisions as the game progresses, and (3) the players are rational (subscribe to rational choice theory), that (4) this strategy (of nuclear deterrence) can be employed indefinitely through time, into the indeterminate future. 1 Has nuclear deterrence encouraged an arms race in conventional weaponry and preparations for national defense? Must a state develop ever more lethal weapons so that a state doesn’t have to use the nuclear option? Has nuclear deterrence been partially responsible for nations collectively spending $60,000 billion on national defense over the past 64-years, and spending $1,500 billion additionally each year? Does this massive diversion of capital to military-based defense merely divert scarce capital from human development and economic growth? Does this diversion of capital make aggression more probable, rather than less so? Other question that need answering include: Is nuclear deterrence, and its overdetermined extended deterrence, a risk factor in economic recovery, addressing climate change, protecting cyberspace, and succeeding at disarmament and nonproliferation efforts? Have nuclear weapons, collectively along with nuclear deterrence doctrine, become a doomsday machine it is long past time to unplug? For example, if a Probabilistic Risk Assessment (PRA) of the continued use of the nuclear deterrence was performed, 2 would these be the outcomes:

1

The Economic Games Behind Nuclear Deterrence http://www.pdfcoke.com/doc/20228926/.

Probabilistic Risk Assessment (PRA) is an analytical process that begins with two system design counterfactuals: (1) the magnitude (severity) of the potential adverse consequences of system failures; and (2) the likelihood (probability) of the occurrence of each potential consequence. 2

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 3 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

Consequences of deterrence failure are unacceptably large; Probability of deterrence failure is beyond an acceptable risk level; Nuclear deterrence is not appropriate for cyberspace; Nuclear deterrence is a barrier for disarmament and nonproliferation; Nuclear deterrence provides an impetus for increasing defense budgets. QUESTIONING THE RISKS OF NUCLEAR DETERRENCE In the risk assessment of the behavior of complex, non-linear systems (all complex systems are non-linear) we encounter a number of practical/philosophical limitations in any probability risk analysis: All associated events that may occur in complex (engineered and natural) systems may not be anticipated due to the presence of system emergence. Emergence is a characteristic of all complex systems due to their nonlinearity; the system will always be capable of surprise. All the system’s attributes and variables may not be fully known, as the system’s behavior evolves over time and unique combinations of inputs/outputs feedback to the system in unanticipated sequences of commands over a long operating history that involves human input where mistakes and arbitrarily determined (outside standard operating procedures) may occur at any time over many years of operation; Consequences will tend to follow a power law distribution rather than a normal distribution. That is, consequences may be immaterial over a very wide range of system and subsystem inputs/failures, but rise to exceedingly catastrophic consequences once a tipping point is reached or exceeded; 3 and In calculating probabilities of events that may occur, we must concede that any probability we predict must be highly uncertain due to the fact that we have not accumulated a full complement of data over the predicted time period. Even if we have collected this periodic data, we run into issues related to assumptions about stationarity (assuming that the future is as ‘similar’ to the past as may be required for historical data to have relevance in calculating forecasts) and Black Swan events (events that cannot be predicted with any accuracy because they cannot be adequately specified as individual events, but, as a class of potential nut imaginable events, they will almost always occur with near certainty) that PRA methods were

All systems have a tipping point, a set of stresses (an overload beyond a threshold rate of change of inputs) beyond which they breakdown (loose complexity and cease to function within normal ranges) and sometimes collapse (recovery is uncertain). As failure proceeds, moments of contingency arise. 3

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 4 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

not designed to address adequately. 4 For example, the SOARCA study due in 2010 does not calculate the consequences or or probability of a terrorist attack on a nuclear power plant, nor does it do a PRA for accidents with spent fuel rods awaiting longterm storage off-site. 5 In the end, models arrived at through PRA or any other analytical methods may be wonderful tools at thinking through engineering and design issues in comprehensive and systematic ways to minimize risk. However, these models can end up offering false policy guidance and misplaced certainty, if one imagines that the numerical precision resulting from PRA are predictive, rather than merely descriptive. It is highly unlikely that reality can be adequately predicted through the use of PRA. Especially with Black Swan events, the asymmetry of these events allows one only to be confident that something will go wrong (i.e. that our analysis is incomplete and probably not completely trustworthy), not about what we believe is deterministically right concerning the PRA. That is because typically our error rate grows exponentially as we project into the far future. Small errors become very large; potentially many magnitudes off the mark. In using PRA, we must avoid misplaced concreteness where we make the mistake of confusing our model with the physical reality it was meant to describe. An expensive example of misplaced concreteness, is the most recent collapse of the global financial markets that were largely caused by Wall Street’s difficulties with toxic assets beginning in mid-2007 and escalating to a full blown crisis through most of 2008 and into 2009. The toxic assets in question were financially engineered collateralized debt obligations (CDOs). The construction of these derivative financial products were based on underlying home mortgages, some subprime mortgages (risky credits) and some prime home mortgages (better credits). Wall Street firms hired PhD physicists to design these derivatives based on complex risk models, similar to the risk models used in PRA. What occurred in 2008, according to the risk models that Wall Street spent an estimated $2 billion developing, should never, ever have happened: the chances of the derivatives loosing their value as they did and becoming toxic assets was calculated as less than once in a a billion years (i.e one in a billion probability of occurrence). Also, the consequences of such an event occurring were calculated to be negligible as the risks were believed to be so well hedged (financially engineered). Yet, these derivates became toxic, in many cases in less than 5-7 years. The consequences have also Nassim Nicholas Taleb, The Black Swan: The Impact of the Highly Improbably (New York: Random House: 2007). 4

United States Nuclear Regulatory Commission http://www.nrc.gov/about-nrc/regulatory/research/soar/faqs.html. 5

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 5 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

not been negligible. The U.S. government and the U.S. Federal Reserve System have allocated $17,489 billion of capital in the form of grants, loans, guarantees, and tax relief to address the meltdown of financial markets and to restart the domestic economy. Wall Street’s misplaced concreteness in its use of CDO’s (engineered based on PRA-type risk models) has cost the global economy more than $50,000 million in lost value of financial assets to date. The moral to this tale is that even the most sophisticated risk models, developed at a cost of a couple of billion dollars can get systemic risk very wrong, and a Black Swan event (in this case the meltdown of the sub-prime market was the precipitating event), unanticipated and unplanned for, can become catastrophic in its consequences and may occur with a probability closer to 1.00 (P<1.0) than the often small probabilities for an event occurring calculated under standard PRA methodology. The above catastrophic failure of the use of PRA-type models in the financial industry does not suggest that these models do not have value or that the results produced from such models should not be weighed in decision-making for allocating scarce capital. Only that PRA results should not be mistaken as predictive of future reality. The value of their results is descriptive. One example of an engineering and design problem where the non-use of PRA-type models is potentially catastrophically expensive is in developing potential consequences and probabilities regarding the failure of nuclear nonproliferation programs. For example, if the possibility that nuclear deterrence would fail once every 100,000weapon years (years where more than 100 operational hydrogen weapons are possessed by more than one nuclear armed state), the risk would approximate the risk of skydiving each year for 100,000 years. However, the consequence of chute failure would not be one person dead, but many millions dead. The cost would not be the cost of a destroyed parachute and bloody clothes, but potentially $1,000 billion in lost GDP over a multi-year period, or much, much more in a worst-case scenario (loss of two or more billion human lives and as much as $100,000 billion casualty loss in a worst-case scenario) rather than an average case scenario. If the PRA results were that the probability of nuclear deterrence failing was once in 100 weapon-years, then this would approximate the risk of making 3 parachute jumps daily for 100 years, with potential consequences as above in an average case scenario. Another way to think about this level of risk is that it corresponds to playing Russian Roulette six times during this 100-year period, or once every 16.7 elapsed years. But each additional 16.7 year period, another bullet is added to the cylinder (the revolver’s cylinder holds six bullets). On the other hand, if the PRA calculated probability of a deterrence failure was once in fifty years, that would mean that every year of delay in addressing (and fixing) this policy engineering problem related to a potential failure in nuclear deterrence would result in an opportunity cost of $20 bil-

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 6 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

lion (2% x $1,000 billion average case consequence from the failure of nuclear deterrence). 6 That is $200 billion in opportunity costs over a decade of doing nothing other than harboring the delusion that deterrence will work forever. A significant problem with all these calculations of probabilities and their consequences is that no PRA has ever been conducted on the potential failure of nuclear deterrence, to the best of our knowledge. It’s unlikely that a secret PRA exists in that the only value of such a study would be its public knowledge. That’s comparable to asserting that deterrence will never fail more than once in a billion years. Somehow, this assertion seem highly unlikely, even as “no mathematical model can capture the intricacies of human psychology.... as [we] must guess what other people are going to do. But their eventual actions will depend on what they think [we] are going to do.”7 INVENTING A NEW DETERRENCE STRATEGY: WHY IS IT SO HARD? Might there be perverse economic incentives to maintain nuclear deterrence? What if the maintenance of nuclear deterrence is a similar situation to the CDO meltdown on Wall Street that destroyed $50,000 billion in asset values? Systemic risk was/is not being managed.8 The systemic risk of nuclear weapons must be costed and included in defense budget discussions. Otherwise, it may appear that nuclear deterrence produces a positive economic return to those countries that hold nuclear weapons (or the capacity to quickly assemble them). This assumption is mistaken thinking. Nuclear weapons, if their systemic risk is included in their costs (e.g. systemic risk insurance premiums) might be understood as unbelievably expensive and dis-economic (as well as morally and ethically untenable). 9 But, as long as states mis-account for their true cost, nuclear weapons look inexpensive and economically attractive from a strategic perspective. Maybe that is why they’re so hard to get rid of. Instead of the horse and buggy that they are, states continue to mistake them for a shiny new technological wonder. The downside to not including the economic cost of managing systemic risk in calculating the cost of nuclear weapons are that: (a) systemic risk may not be managed See Martin E. Hellman, “Risk Analysis of Nuclear Deterrence,” The Bent of Tau Beta Pi, The Engineering Honor Society (Spring 2008) 14-22; Martin Hellman, “Soaring, Cryptography and Nuclear Weapons” (October 21, 2008) at http://nuclearrisk.org/soaring_article.php. 6

Quoting Emanuel Derman, former Goldman Sachs executive, ‘quant,’ physics PhD, and professor at Columbia University in Pablo Triana, Lecturing Birds on Flying: Can Mathematical Theories Destroy the Financial Markets? (NY: Wiley, 2009). 7

8

See “Securitization of Risk” at http://www.pdfcoke.com/doc/20747708/.

9

“Costing Systemic Risk from Doomsday Machines” http://www.pdfcoke.com/doc/19772476/

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 7 of 8

R E T H I N K I N G

N U C L E A R

D E T E R R E N C E

D O C T R I N E

adequately as its costs are not included in allocating capital to national defense; (b) budgets drive wasteful, zero-sum activities even as no deterrence or national defense or national security value-added may be achieved; (c) capital is misallocated as an improper discount rate is often chosen that may make deterrence/defense projects look more attractive than they actually are in practice. Not managing systemic risk is ultimately destructive of capital and national security. Certeris paribus (all things being equal) does not apply to nuclear deterrence as accounting for the systemic risk of nuclear weapons is a world-changer. Even limited nuclear war is presently believed to cause global destruction of ecosystem services and potential for world-wide famines. The use of nuclear weapons is predicated on hetacomb - the belief that the deaths of many are necessary to save a privileged few. That some utility may result from nuclear deterrence doctrine does not belie that nuclear weapons hang like a Sword of Damocles over all the peoples of the earth and the earth’s life-sustaining environment. It is long past the time to examine the feasibility for: (a) developing an updated deterrence framework; (b) deterrence that does not depend on nuclear weapons; and (c) an alternate methodology for allocating capital to deterrence based on this new framework. Otherwise, the global economy is set to spend yet another ~$60,000 billion (in current US dollars) for national defense predicated on nuclear deterrence (either directly or under the nuclear umbrellas of nuclear states) over the next 64-years. Can the world afford this allocation of capital today in the face of known national security risks? Should the world afford nuclear deterrence at all, given this doctrine’s systemic risks and that rational choice theory upon which nuclear doctrine rests has been apply proven to be a highly deficient description of human behavior (e.g. 2008 financial markets meltdown)?

LYLE A. BRECHT 410.963.8680 DRAFT 2.4 CAPITAL MARKETS RESEARCH --- Sunday, January 3, 2010

Page 8 of 8