Wicken has a good example that can be adapted to illustrate this difference (1987:185). Suppose that I ask you to place a die on a table a thousand consecutive times, each time choosing the face you want to place upward. Shannon’s equation calculates the informa-tion on the resulting sequence as a function of the number of times a face actually turns up in comparison with the number of times it could be predicted to turn up, given the number of faces on the die. The uncertainty implied by the probability function reflects my in-ability to know in advance which choices you will make, not my ignorance about which die faces have already appeared. To imagine an analogous case for thermodynamic entropy, suppose that a thou-sand dice are cast all at once, and a measuring instrument records the total amount of light reflected from the die faces. We do not know how each individual die landed. After calibrating the light in-strument, however, we could figure out the average face count on the basis of the amount of light reflected. In this case the probability function reflects ignorance of the microstates, not ignorance of our choices in assembling a series of such states.