by Peter Byrne
Mutual assured destruction was based on terrifying both sides into building large enough arsenals that it would became rational not to use them. Of course, if a comparative advantage was reached by either side, it would be tempting to preemptively wipe out the enemy. Consequently, the hydrogen bomb-wielding players were trapped in a feedback loop of suspicion and weaponeering as it is not rational to expect the enemy not to strike first when he is afraid that you will strike first.
The decision matrix for assured destruction ignored Rapoport’s admonition to look outside the box, in this case at the only truly rational utilities: avoiding war through disarmament or, better yet, never having built the first atom bomb!2 Obviously, the big winners in MAD were weapons manufacturers and operations researchers.3
It gradually became clear to game theory practitioners that the theory was more valuable for modeling tactical scenarios, and making cost-benefit analyses of weapons systems, than it was useful as a consistently rational guide to deciding whether or not to strike first, or second, in a nuclear stand-off. A remarkable fact that was not always ignored by Cold Warriors on both sides of the superpower divide was that repetitive plays of Prisoner’s Dilemma allow the same two players to gain information about each other’s decision-making patterns, thereby reducing the risks tied to cooperation, and making multilateral disarmament an agreeably rational option.
Of great note: modern incarnations of game theory in evolutionary biology show that in life’s contest Nature tends to reward group altruism—cooperative behavior—more than individual selfishness and competition.4 However, in matters of military planning (then as now), cooperative options are easily, and unfortunately, conflated with weakness and irrationality and assigned low utility values.
Everett, who was the quintessential Cold War technocrat, subscribed to the prevalent notion among his generation of operations researchers that, “We prepare for war to prevent war.” Values that might have questioned the basis for this rationale were banished from the morally closed world of operations research.
Rapoport observed:
The strategist defends nightmare visions of the world as a ‘realistic’ vision, forgetting that any vision of the world is compounded of elements which one has selected for observation. The strategist sees what he has selected to see.5
Or as the satirical songwriter, Tom Lehrer, wrote in the early 1960s,
‘Once the rockets are up, who cares where they come down? That’s not my department,’ says Werner von Braun.
8 von Neumann’s Legacy
The model nature is quite apparent in the newest theories, as in nuclear physics, and particularly those fields outside of physics such as the Theory of Games, various economic models, etc., where the degree of applicability of the models is still a matter of considerable doubt. However, when a theory is highly successful and becomes firmly established, the model tends to become identified with ‘reality’ itself, and the model nature of the theory becomes obscured…. It should be clearly recognized that causality is a property of a model, and not a property of experience.
Hugh Everett, III (1956)1
In March 1957, Everett, who was by then a heavyweight in the Weapons System Evaluation Group (WSEG) at the Pentagon, attended the Third Conference on Games at Fine Hall. The meeting was dedicated to the memory of von Neumann, who had died a few weeks before at the age of 53 from bone cancer. Dozens of mathematicians and economists flocked to the heady affair, sponsored by the Office of Naval Research. The participants hailed from Ivy League universities, and such military contractors as RAND, General Electric Corporation, and Hughes Aircraft. The roster of attendees was a cross-section of the best minds working for the “military industrial complex,” which President Dwight D. Eisenhower was soon to call a threat to democracy.2
Olaf Helmer of RAND presented a striking paper, “The Game-Theoretical Approach to Organization Theory.” Having learned the lessons of Prisoner’s Dilemma, Helmer concluded:
The theory of games has in my opinion reached a state of near stagnation with regard to its applicability to the real world…. [W]hat it cannot generally do is to predict the behavior, even of rational players with known utility preferences, with regard to their cooperative options, insofar as these depend on the players’ attitudes toward their fellow players and toward any behavior patterns which they may have observed in past plays.3
A decade before, Wiener had made a similar comment in Cybernetics:
Naturally, von Neumann’s picture of the player as a completely intelligent, completely ruthless person is an abstraction and a perversion of the facts. It is rare to find a large number of thoroughly clever and unprincipled persons playing a game together.4
But von Neumann-style game theory pointed the way toward the invention of more useful computational techniques for modeling complex situations in real life (including socio-economic and war simulations), and Everett was one of von Neumann’s legatees, in more ways than one.
Rich, socially flamboyant, and supernaturally smart in life, von Neumann remained a demigod long after his untimely death. At the Princeton game theory conference, Morgenstern eulogized his recently departed friend and colleague: “There was hardly another contemporary whose mind encompassed so much and who made such a significant contribution to anything he touched.”5
Gifted on the left side of his brain, von Neumann had been the first to see that the equations of quantum mechanics could be expressed in a highly abstract “Hilbert space” of infinite dimensions. While axiomatizing the new quantum physics in the early 1930s, he postulated a widely accepted solution to the vexing problem of discontinuity in the measurement of quantum objects: “wave function collapse.” (As von Neumann lay dying in Princeton, Everett was thinking up a way to debunk that postulate.) In computer science, von Neumann was one of the inventors of “stored programming,” i.e. software. The 50-ton digital computers that he designed for weapons research burned through vacuum tubes like tissue paper, crashed every few minutes, and were among the most elegant machines every built.
A death bed convert to Catholicism, von Neumann had long struggled with the mysteries at the heart of physics and socio-economics. Uncertainty was his sworn enemy, whether it appeared in the nature of the electron, or in the nature of society. He evinced no moral uncertainty, however, about designing the implosion method for the plutonium bomb that destroyed Nagasaki in August 1945. And along with his fellow hydrogen bomb-engineers, Edward Teller and John Wheeler, von Neumann was certain that preparing to wage nuclear war was a historical necessity.
Life magazine recalled in his obituary:
After the Axis had been destroyed, von Neumann urged that the U.S. immediately build even more powerful atomic weapons and use them before the Soviets could develop nuclear weapons of their own. It was not an emotional crusade, von Neumann, like others, had coldly reasoned that the world had grown too small to permit nations to conduct their affairs independently of one another. He held that world government was inevitable—and the sooner the better. But he also believed it could never be established while Soviet Communism dominated half the globe. A famous von Neumann observation at the time: ‘With the Russians it is not a question of whether but of when. If you say “why not bomb them tomorrow,” I say “why not today?” If you say “today at 5 o’clock,” I say “why not one o’clock.”’6
Physicist-historian Kenneth Ford has a more subtle recollection:
I heard von Neumann advocate for the development of the H-bomb on the grounds that it was necessitated by inherent inaccuracy in ICBMs [intercontinental ballistic missiles]. He wanted a multimegaton weapon, he said, not in order to inflict orders of magnitude more damage, but to take out a target if the missile missed it by 5 or 10 miles. Despite his brilliance, he did not at that time envision the pinpoint accuracy of ICBMs that would later be developed.7
As a powerful advocate for hydrogen bomb testing, von Neumann told Lewis Lichtenstein Strauss, the head of the Atomic Energy Commission: “Th
e present fear and vague talk regarding the adverse world-wide effects of general radioactive contamination are all geared to the concept that any general damage to life must be excluded.” (In his first project at the Pentagon, Everett was to scientifically disprove von Neumann’s low assessment of the lethal effects of radioactive fall-out.)
von Neumann remarked of fighting a nuclear war:
Every worthwhile activity has a price, both in terms of certain damage and of potential damage-of risks—and the only relevant question is, whether the price is worth paying…. For the U.S. it is.8
Super-feisty Hermann Kahn put the “price” worth paying for victory in a nuclear exchange with the Soviets at 60 million American dead.
Writing to Strauss in 1951, von Neumann revealed his Freudian side:
The preliminaries of war are to some extent a mutually self-excitatory process, where the actions of either side stimulate the actions of the other side…. As the conflict’s ‘foreplay’ progresses, the original aggression, and its motivation, become increasingly obscured. [I think] the US-USSR conflict will probably lead to an armed ‘total’ collision, and that a maximum rate of armament is therefore imperative.9
Such a ridiculously sexualized, Manichaean comment would be amusing, excepting that during much of the 1950s, the de facto strategy of the Strategic Air Command under General Curtis LeMay was to “preventatively” launch everything in its nuclear arsenal in what Kahn disapprovingly termed a “Wargasm.”
As the arms race heated up, von Neumann sold his services as a consultant to the Central Intelligence Agency, International Business Machines, Standard Oil, and Convair (a strategic bomber and ballistic missile manufacturer).10 Racing through the revolving door, he headed the “Teapot Committee,” a collection of military and scientific experts that, in 1953, recommended going full speed ahead on the production of ballistic missiles capable of carrying nuclear warheads. Eisenhower appointed him to the Atomic Energy Commission, where he oversaw the nation’s atomic energy projects. And he helped design SAGE (Semi-Automatic Ground Environment), the top-secret computer system designed to detect a nuclear attack by the Soviets.11
Although Everett barely knew von Neumann, he was an aficionado of his work. He aimed to teach digital computers how to play war according to von Neumann. And he devoted his doctoral thesis to a proof that one of the axioms of quantum mechanics (wave function collapse), per von Neumann, was flat out wrong.
So Long, Suckers
One of Everett’s favorite recreations was a zero-sum board game called So Long, Suckers, invented in 1950 by Princeton graduate students, including Nash and Shapley. Similar to Prisoner’s Dilemma, it was also known as Fuck You, Buddy. The game encourages chip-trading players to cooperate until it becomes more convenient to deceive and murder each other. Everett himself viewed life itself as a winner-takes-all game, and playing mind games was, for him, an acceptable way to optimize his personal utilities—suckers beware.
Kuhn reflects, “Everett had a good deal of ego even as a graduate student. We were all kind of an elite group, though, so it would have been hard for him to play mind games with us.” He lost track of the young game theorist after he, “disappeared inside a classified organization [WSEG].”
During his long career in academia and in private operations research, Kuhn did important work in the field of linear programming, which is a method of programming zero-sum, two person games.12 Linear programming is used to find the optimal solutions to complex problems in economics and operations research, problems that are subject to a large number of constraints. The method can calculate solutions for optimizing the delivery of medical supplies to all sides fighting a civil war, or to budgeting dollars between ballistic missiles, planes, and bombs for maximum punch in a nuclear strike.13
Kuhn, now in his eighties, is sensitive to the tendency of some Cold War historians to emphasize the malign applications of game theory and computerized optimization techniques over their more benign uses. Although Kuhn accepted Office of Naval Research research funding, he avoided working directly for the military. He says,
It is embarrassing when something you do for purely mathematical reasons is turned around and used for dirty applications. I was like a tailor who makes a coat with six arms one day, a coat with three arms the next day, a coat with seven arms the following day, and then a coat with two arms: if somebody puts on the two-armed coat, so be it.14
Everett, on the other hand, did not share Kuhn’s qualms; he gladly churned out two-armed coats on demand for the war machine.
In 1963, Everett resurfaced in Kuhn’s world, making a splash when a professional journal, Operations Research, published his “Generalized Lagrange Multiplier Method for Solving Problems of Optimum Allocation of Resources.” The “Everett Algorithm” improved upon a centuries-old optimization technique called the Lagrange multiplier method. A creation of the information age, it simplified the solution of hugely complex logistics problems. (Everett discovered the algorithm under the influence of a few beers while visiting Niels Bohr in Copenhagen in 1959.)
The beauty of the algorithm is that it breaks large, intractable optimization problems into smaller ones that can then be solved. Whereas game theory deals with how people make interdependent decisions under certain conditions, the Everett Algorithm tells you how to calculate a range of consequences (“prices”) for making those decisions in the real world when you expend a specific amount of a resource to overcome a constraint. It can be applied to logistics problems in which there are large numbers of alternative solutions and variously configured obstacles to those solutions. Problems of this type include maximizing the efficiency of assigning nuclear bomb targets, scheduling just-in-time manufacturing runs, allocating bus routes to most efficiently desegregate school systems, or projecting results of funding specific foreign and domestic policies.
The algorithm was the cornerstone of Everett’s career in operations research; but it did not drop on him totally out of the blue sky.
Kuhn recounted,
In 1951, Tucker and I published a paper which extended linear programming as a very useful and important research tool for doing efficient optimization for all kinds of business. Our approach was effectively a Lagrange multiplier approach. Later, Hugh, working at WSEG, had this bright idea about Lagrange multipliers which, in effect, we had previously generalized in our paper. Traditional Lagrange multipliers applied to situations with equality constraints. We had inequality constraints, and so did Hugh.
Kuhn said that Everett went a step further than he and Tucker; and he enthusiastically quoted from Everett’s innovative paper:
While the use of Lagrange multipliers does not guarantee that a solution will necessarily be found for all problems, it is ‘fail-safe’ in the sense that any solution found by their use is a true solution. Since the method is so simple compared to other available methods it is often worth trying first, and succeeds in a surprising fraction of these cases. They are particularly well suited to the solution of problems of allocating resources among a set of independent activities.
“Everett’s Lambda theorem is a lovely idea,” Kuhn continued. “It has very little mathematical content, but it has extreme practical application in operations research.”15
In his path breaking paper, Everett noted that his multiplier method could be applied to
a problem which often arises in military operations research [which] is the optimum allocation of given stocks of several weapons types of differing characteristics to a diverse set of independent targets. For such problems it is often crucial to account for the fact that weapons can be delivered only in integral numbers.16
He described the solution matrix clinically:
In this case, the cells are the individual [nuclear bomb] targets. A strategy for a cell is a -tuple of integers, one for each weapon type, representing the number of that type of weapon to allocate to that target. The payoff in a cell is the expected destruction to the target of the given weapon allocation
, and the resource functions are simply the numbers of weapons allocated themselves…. The method has been employed in WSEG for several years for solving both production and military allocation problems, and has been quite successful.17
Now that we have introduced Everett as a game theorist, optimization expert, and nuclear war planner, it is time to back-track and see what he was doing in quantum mechanics during his second year at Princeton. But, first, we must note that an off-shoot of game theory—decision theory—is now playing a major role in the international debate about the validity of Everett’s Many Worlds Interpretation. Philosophers and physicists at University of Oxford and elsewhere are looking toward the utilitarian and probabilistic structures embedded in decision theory as an argument that it is rational to believe that Everett’s multiple universes exist.
Whether or not Everett ruminated about decision theory, per se, while constructing his interpretation of quantum mechanics is not known, but the structure of his branching universe resembles a decision tree—to which the interplay of information and probability and rationality is fundamental.
BOOK 3
QUANTUM WORLD
9 Quantum Everett