Monday, November 12, 2012

David Warsh reflects on Howard Raiffa's memoir and career

On the anniversary of the Cuban Missile crisis, David Warsh reads some books, ending up with Howard Raiffa's memoir, about which he has this to say:

High drama, great stuff.  But I wanted something a little closer to life as we had lived it in the decades since that narrow escape. I found it in Memoir: Analytical Roots of a Decision Scientist, by Howard Raiffa. His name is hardly a household word. Raiffa is one of those intellectuals who in the years since World War II fundamentally transformed the world in which we live, as one of the foremost pioneers of the application of mathematics to business – not just game theory, but Bayesian statistical decision analysis as well.
Aas it happens, between 1968 and 1975 Raiffa also organized, and then  administered the joint US- Soviet think-tank known as the International Institute for Applied Systems Analysis. IASSA (pronounced YASSA), itself hardly a household word, is one of the durable outgrowths of the Cuban missile crisis and, in many ways, a symbol of the strange interregnum of the 1970s known as d├ętente. Last week in Vienna, IASSA celebrated its fortieth anniversary having been, among other things, the cradle of climate modeling.
It is not the career you would have expected from a rangy kid in the Bronx who loved basketball better than schoolbooks.  At thirteen, Raiffa played in a high school championship game inMadison Square Garden. At sixteen, in 1940, however, he met Estelle Schwartz, whom he would marry five years later. She was a student at Manhattan’s High School of Music and Art; to stay close, he enrolled in the City College of New York. He loaded up on math courses; his basketball aspirations shifted from playing to coaching. Three semesters later, he enlisted, eventually becoming an air traffic control officer working on radar for the Army Air Corps. When peace came, he was sent toJapan.
After mustering out, Raiffa switched to theUniversity of Michigan.  He turned out to be a much better student that he had thought. Since the engineering profession was said to be generally anti-Semitic; he resolved to become an actuary. Estelle Raiffa obtained a masters degree in elementary education, teaching autoworkers’ kids at Willow Run, near Ann Arbor.  Six years later they were still there.  Raiffa had run into Kenneth Arrow, Robert Tucker, John Nash, Abraham Wald, Lawrence Klein and Robert Solow. He had become a mathematician, working in the borderlands of economics, psychology and the new field of operations research.
Abraham Wald was killed in a air crash; to replace him, Columbia hired Raiffa in 1952.  (He turned down what would have been a better-paying offer to work with Claude Shannon at Bell Labs).   Columbia had been a hotbed of statistics since before the war; it was teaching there that Raiffa began his conversion to the Bayesian approach, gradually learning to update initial beliefs with objective new information as it arrived, just as Rev. Thomas Bayes, the early eighteenth-century amateur mathematician, had first maintained should be done when thinking probabilistically.
Having been trained in the verities of classical (Neyman-Pearson) statistics, Raiffa’s colleagues complained: “Look, Howard [he says they would say], what are you trying to do?  Introduce squishy judgmental psychological stuff into something which we think is a science?”  In reply, he quoted the legendary Leonard Jimmie Savage (by then Savage had joined Milton Friedman and Allen Wallis at the University of Chicago): “Yes, I would rather build an edifice on the shifting sands of subjective probabilities than build upon a void.”
(For an illuminating account of this revolution in statistical thinking, including a chapter on what Raiffa did next, see The Theory that Would Not Die: How Bayes Rule Cracked the Enigma Code, Hunted Down Russian Submarines, & Emerged Triumphant from Two Centuries of Controversy, by Sharon Bertsch McGrayne, Yale University Press, 2011.  It is at least a very useful complement to Nate Silver’s The Signal and the Noise – Why So Many Predictions Fail but Some Don’t, Penguin, 2012, and in some ways a much better book.)
It was when HarvardBusinessSchoolhired him in 1957 that Raiffa’s major phase began. By then, Games and Decisions: Introduction and Critical Survey, the text he had written withDuncan Luce, had appeared and opened a portal through which game theorists and economists would pass in the next generation. (Its original title had been Conflict, Cooperation, and Conciliation.)  A new department of statistics was forming, one that included Frederick Mosteller, Raiffa, John Pratt and, for a time, the ancient Greek scholar Robert Schlaifer. In the next few years Raiffa’s students included Richard Zeckhauser, Robert Wilson, Michael Spence. Edith Stokey, Roger Myerson, Eric Maskin (at least three future Nobel laureates among them), and even a young Lawrence Summers, who pronounced Games and Decisions among the most eye-opening books he had ever read.  B-school students called Raiffa “Mr. Decision Tree.”  He started an institute to intensively train forty professors in the new methods of game theory, decision analysis, and operations research. Several subsequently became influential business school deans: Lawrence Fouraker at Harvard; Robert Jaedicke at Stanford; Donald Jacobs at Northwestern.
It was in 1966 that President Lyndon Johnson asked National Security Advisor McGeorge Bundy to investigate whether some new form of high level cooperation with the Soviets might be possible. The missile crisis had demonstrated the value of the teletype connection linking the White House and the Kremlin that was the “hotline” that Thomas Schelling had proposed in 1958.  Stanley Kubrick’s Dr. Strangelove was making vivid to moviegoers the dangers of mutual incomprehension.  Perhaps a bricks-and-mortar East-West center for high-level social scientists working together on problems experienced by communist and capitalist societies alike – energy, water resources, food and agriculture, population studies, urban policies – would enhance chances for peace. (It subsequently turned out that Francis Bator, Bundy’s deputy at the NSC, had thought up IASSA and planted the idea with the president, much as W.W. Rostow had first imagined the UN’s Economic Commission for Europe as a bridge between East and West, over which economist Gunnar Myrdal would preside for a decade in Geneva in the 1950s.) In early 1968, Raiffa agreed to take on the task.
The Soviets agreed. Newly-elected President Richard Nixon signed on. Viennawas chosen. What emerged, when the heavily renovated old palace outside of Vienna finally opened its doors in 1973, was a thinking-person’s version of what by then truly was a household word:  the Club of Rome.  That self-proclaimed “group of world citizens, sharing a common concern for the future of humanity” had grabbed headlines in the early 1970s with a bootleg computer model designed to demonstrate the need  for “lifeboat ethics” – a Dr. Strangelove script for the newspapers.
Clearly some sort of deeper dispassionate long-term thinking was needed about where the planet was headed.  So Raiffa populated the Vienna installation with first-rate intellects willing to take one- or two-year appointments in order to get things started, including Arrow, George Dantzig, Tjalling Koopmans, William Nordhaus, Alan Manne — and Donella and Dennis Meadows, principal authors of Club of Rome’s Limits to Growth. Raiffa hired talented Soviets; decreed that seminars would be conducted only in English, anticipating a trend that has since spread around the world; and encouraged the development of small environmental models, as opposed to the behemoths that Soviet planners preferred. “Howard was a latter-day George Washington,” says Mark Thomson, who served as Raiffa’s executive assistant in those years. “He was devoted in every way possible to the overall cause.”  Bundy later said that IASSA had succeeded only “because Howard didn’t know that it was impossible.”
IASSA slowed down some when Raiffa returned to Harvard in 1975, after three years as director, and soon thereafter founded its Project on Negotiation. (His 1985 Art and Science of Negotiation: How to Resolve Conflicts and Get the Best Out of Bargaining is today Raiffa’s best-read book. Harvard Law Professor Roger Fisher, who died, at 90, in August, soon joined the Negotiation project and later  sold 8 million copies of Getting to Yes: Negotiating Agreement without Giving In, cowritten with William Ury. Events accelerated; IASSA’s appointments became less spectacular. There was a fracas during the Reagan administration over whether it had become a nest of Soviet spies.  Institutes for Advanced Study in Princeton and Berlin impinged on its turf.  Its most successful project, climate modeling, was essentially spun out, to the International Energy Workshop and the UN’sIntergovernmental Panel on Climate Change.
Still, if the YouTube selections from the celebration in Vienna last week are any guide, IASSA remains a vital intellectual center, sponsoring systems analysis work on an array of interesting problems, ready to play a part whenever the next global crisis – food? water? – becomes acute.  The Raiffas attended, despite a twenty-year-long battle with Parkinson’s Disease that has reduced his mobility, before flying home for his friend Fisher’s memorial service.
Decision analysis is now firmly established. The carefully–designed controlled experiments for which Daniel Kahneman (and, by extension, his late research partner Amos Tversky) received the 2002 Nobel Prize in Economics, were predicated on an extensive body of body of prior work. It was after receiving Carnegie-Mellon University’s Dickson Prize, one of a handful of such awards that exist within the penumbra of the Nobel, that Raiffa undertook his Memoir. In its closing pages, he envisages departments of decision science within universities. These, he says, would offer instruction to undergraduates (Societal Risk Analysis, for example, Organizational Design and Structures of Constitutions); perform research on decision-making at once empirical, normative and prescriptive; and train a new breed of specialists: decision advisors, or DAs, equipped to help decision makers confront the intricate choices they must make. It all takes time, of course. Big mistakes still occur (remember Saddam Hussein’s weapons of mass destruction?). But what a long way we have come in the half centuty since the Cuban missile crisis.


Unknown said...

For students at Harvard/MIT interested in Raiffa's approach to the analysis of decision-making, I can highly recommend Richard Zeckhauser's API-302

Joseph Malkevitch said...

Was the reference to "Robert Tucker" meant to be Albert W. Tucker, who made contributions to game theory and linear programming?