In it, he criticizes, not so much the recent Nobel Memorial Prize to Aumann and Schilling (no, it's not really a Nobel Prize, it's a Bank of Sweden Prize) but the notion that game theory is useful in economics. He says:
Game theory is no doubt wonderful for telling stories. However, it flunks the main test of any scientific theory: The ability to make empirically testable predictions. In most real-life situations, many different outcomes -- from full cooperation to near-disastrous conflict -- are consistent with the game-theory version of rationality.
And I'm afraid that, in a development that also results in the immediate collapse of my metaphor in the first paragraph, both Cowan and Roberts have missed the catch. Mandel is making a category error in the whole argument: game theory is not a "scientific theory", it's a "mathematical theory", and the two are completely different orders of knowledge. It's a silly error, and an error of essence, not accident.
My point, however, isn't just to point out an error by two eminent bloggers, or make fun of a person writing for Business Week — although the temptation to do so when reading a column by a guy who happens to point out that he's a "journalist with a Ph.D." is pretty intense — but to use this as an example to talk about the same category error that happens over and over again, notable in talk about Intelligent Design, or about the status of atheism.
The error is actually simple in this case: "game theory" is a mathematical construct, not a theory in the sense of prediction and confirmation or falsification. His argument that game theory is not predictive is like making the argument that arithmetic is not predictive, or noting that euclidian geometry breaks down in the real world. von Neumann and Morganstern started with some basic observations and developed a mathematical theory based on combinatorics and probability which has turned out to be useful in thinking about other situations.
The category error lies in confusing the mathematical formalism with the use economists happen to make of it --- just like it would be a category error to claim that geometry is a dead end because it doesn't predict what a surveyor sees accurately.
To me, the thing that really struck me about this category error, though, is how common this becomes. In intelligent design, the ID theorists claim that "Darwin's theory of evolution" is false because it can't predict everything or can't explain everything — a standard that fits mathematical constructs, not natural science.
On the other hand, the devout atheists assert that notions of Diety must be false because they can't be confirmed with science; a category error for a number of reasons, but most specifically because it has logical issues in that a Deity could create a universe in which all experiments to confirm or deny the Diety's existence are true, false, or equivocal. (For a fuller discussion of this, see Steven Brams book Superior Beings: If They Exist, How Would We Know? : Game Theoretic Implications on Omniscience, Omnipotence, Immortality and Incomprehensibility.)
There are different kinds of knowing, different kinds of "truth": mathmatical, scientific, metaphysical.
How often do category errors among the kinds of truth lead to human suffering?
Update: revised "accident" for "argument" above: an error.