Abstract
A shift in focus, of the sort recently advocated by David Wallace, towards consideration of work in nonequilibrium statistical mechanics has the potential for far-reaching consequences in the way we think about the foundations of statistical mechanics. In particular, consideration of the approach to equilibrium helps to pick out appropriate equilibrium measures, measures that are picked out by the dynamics as "natural' measures for systems in equilibrium. Consideration of the rationale for using such measures reveals that the scope of their legitimate employment is much more limited than an appeal to a Principle of Indifference would suggest. These points are illustrated by use of a toy model that I call the parabola gadget.