page of the paper, insisting that the probability of its being a real face (or a representation of a face) was minuscule. But a deputy managing editor replied that either it was or it wasn’t, so the odds were 50-50! I hoped he was kidding, but decided it would be wiser not to ask.


9. Jaynes, “Information Theory,” pp. 620, 621.


10. This is, of course, the basis for teachers grading on a “curve,” the bell-shaped curve or Gaussian distribution that represents equal probability of all the microstates.


11. David Wolpert, interview at NASA Ames Research Center, July 18, 2005.


12. Ibid.


13. Another decision-theory system was worked out by the statistician Abraham Wald, but the story of the similarities and differences between Wald’s and Savage’s approaches goes far beyond the scope of this discussion. If you’re interested, you might want to consult a paper exploring some of these issues: Nicola Giocoli, “Savage vs. Wald: Was Bayesian Decision Theory the Only Available Alternative for Postwar Economics?” Available online at


14. David Wolpert, interview at NASA Ames Research Center, July 18, 2005.


15. Strictly speaking, it’s not the temperature of an individual, it’s the temperature that the external scientist assigns to the individual, Wolpert points out. Just like in statistical physics, the temperature is a measure of what the external scientists infer concerning the molecules in a room (since a single molecule doesn’t have any particular temperature—temperature is a property of the distribution of velocities of a set of molecules).


16. Wolpert, interview at NASA Ames, July 18, 2005.



1. Orson Scott Card, Ender’s Game, TOR, New York, 1991, p. 125. Thanks to my niece Marguerite Shaffer for calling Ender’s Game to my attention.


2. Ibid. p. 238.


3. Joshua Greene, interview in San Francisco, Calif., April 17, 2004.

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement