Pessimism of the Intellect, Optimism of the Will     Archive   Favorite posts   Twitter: @steve_hsu

Monday, December 18, 2006

Machine Dreams

We're on break right now, so I have time to do some deeper reading. I'd like to recommend the book Machine Dreams by the economist and intellectual historian Philip Mirowski. Mirowski ruffled quite a few feathers in economics with his earlier book More Heat than Light: Economics as Social Physics, Physics as Nature's Economics, in which he argued that much of the mathematical and conceptual framework of neoclassical economics was lifted from 19th century physics. In particular, he argued that maximization of utility was inspired by ideas about energy, and that market equilibrium was inspired by thermodynamics. Origins aside, whether these ideas are useful for the description of complex, nonlinear systems comprised of thinking participants is another question, addressed directly in this book.

Machine Dreams is, if anything, more ambitious than the earlier work. In it, Mirowski traces the influence of ideas concerning information and computation, largely developed by figures like von Neumann, Turing, Shannon and Szilard, on the field of economics since the 1930's. Mirowski refers to these individuals as cyborgs, and their area of interest as the cyborg sciences. He adopts an amusing tone throughout the 600 pages of his book, even as he delivers devastating blows to sacred cows of the economics orthodoxy.

This is a controversial book because it demolishes not just the conventional history of the discipline, but its foundational assumptions. For example, once you start thinking about the information processing requirements that each agent (or even the entire system) must satisfy to find the optimal neoclassical equilibrium points, you realize the task is impossible. In fact, in some cases it has been rigorously shown to be beyond the capability of any universal Turing machine. Certainly, it seems beyond the plausible capabilities of a primitive species like homo sapiens. Once this bounded rationality (see also here) is taken into account, the whole notion of optimality of market equilibrium becomes far-fetched and speculative. It cannot be justified in any formal sense, and therefore cries out for experimental justification, which is not to be found.

But there is more. Mirowski reveals that tenets of rationality and utility maximization were already at odds with results of game theory experiments conducted at RAND as early as the 1950s. He traces von Neumann's (vN's) influence on the discipline, which he claims has been deliberatedly ignored and obfuscated in the official histories. In his summary, he writes

This scientific titan [vN], who could only spare a vanishing fraction of his intellectual efforts upon a science he regarded as pitifully weak and underdeveloped has somehow ended up as the single most important figure in the history of 20th century economics. This mathematician who held neoclassical theory in utter contempt throughout his own lifetime has nonetheless so bewitched the neoclassical economists that they find themselves dreaming many of his formal models, and imperiously claiming them for their own. This polymath who prognosticated that "science and technology would shift from a past emphasis on subjects of motion, force and energy to a future emphasis on subjects of communications, organization, programming and control," was spot on the money.

Mirowski's book has drawn significant attention within the economics community. I suggest the following review by E. Roy Weintraub, a mathematical economist at Duke (Journal of Economic Behavior & Organization Vol. 53 (2004) 419–434):

Philip Mirowski is a singular historian of economics. Every one of his works has made a difference in our understanding of the development of economics. He is brash, uncompromising, and dedicated to producing magnificent historical studies. Nevertheless he raises his colleagues’ blood pressures because his work has both transcended and transformed the subdiscipline, and few are intellectually flexible enough to enjoy rethinking accepted ideas.

...In his new work, Machine Dreams: Economics Becomes a Cyborg Science, Mirowski (2002) reconstructs the history of neoclassical economics to the modern period, writing that history against and intertwined with the emergence of the cyborg sciences concerned with information—computer science, cybernetics, statistics, operations research, game theory, gaming and simulation, etc.—in the World War II and subsequent Cold War period. His major historiographic point, continuous with his past writing, is that one cannot construct a coherent narrative of the emergence of modern neoclassical economics in the postwar period without looking outside economics proper for both the dramatis personae, and the punch lines. His use of methods from history, sociology, and anthropology frames the best that Science Studies has to offer to help one construct a compelling narrative.

...Overall, this is the single most important totalizing narrative of the history of economics that we have had in the last twenty years. Important does not however mean that it will be loved. Mirowski’s writing is vivid and forceful. He enjoys the sound of words and the rhythm of the sentences he constructs. He writes with a musical fluidity. Yet to say that Mirowski is verbally facile, sharp-tongued, and acerbic hardly does justice to the blood he draws from rapier slashes, and cleaver smashes, to the august and famous. It is, given the author’s critical program, an angry book.

Withal, Mirowski is a scholarly treasure. There are few of us who make an immense difference, who see things differently and can maintain that vision through a sustained scholarly life, and who can show others how to make that vision their own. We need to take this book seriously.

For a meta-review of reviews, see Boland, and for a response by Mirowski to several critics, see here (contribution to a symposium on Machine Dreams published by the Journal of Economic Methodology).

blog comments powered by Disqus

Blog Archive

Labels

Web Statistics