11/04/2011

Statistics, Sociology, Physics

Had a read of the "Statistics and Physical Theories" chapter by Theodore Porter in The Cambridge History of Science (vol 5) today as some relief from adverse selection/signalling/screening models.  I found it near why I was sitting in the library and I've been recently talking to an art historian friend, Ruth Harriss, about this kind of thing in relation to Xenakis's philosophy of music (more to follow on this later).

It's quite an interesting chapter - he points out that statistics as a subject area was originally a human science, attached to census analysis &c and only gradually became attached to probability theory before being used in statistical physics (the etymology of 'statistics' is statisticum collegium: 'council of state').  He also talks for a while about the history of statistical regularities being seen as synonymous with 'social laws', and arguments over the impossibility of human agency in the face of such regularities through the 19th century.

More interesting, however, is the section Reversibility, Recurrence and the Direction of Time where the "reversibility paradox" is discussed.  The reversibility paradox is that, under the statistical physics interpretation, the second law of thermodynamics (law of increase in entropy in an isolated system) is based purely on the fact that high entropy states are more probable than low entropy states.  In which case there is always a small but positive probability that entropy might spontaneously decrease in a given system.  The recurrence paradox is that any ergodic system should eventually return very nearly to their original state - again, suggesting that entropy should at some point (however far in the future) decrease.  A spontaneous decrease in entropy is never observed however - hence the paradox.

This also raises more philosophical problems, as the law of entropy increase is the physical basis for an 'arrow of time':

"Perhaps the universe simply began in an extremely improbable state.  Possibly, however, it is so huge that from time to time vast areas of it fluctuate by chance into highly improbable states.  These define the conditions in which life is possible . . . Or perhaps our experience of the direction of time is defined by increasing entropy."

This, of course, is all pre-Big Bang theory, and pre-Schroedinger's definition of life as a process of 'negative entropy'.  Lastly, Porter repeats Boltzmann's point that even a small volume of gas would only separate into hydrogen and oxygen about once in (10^10)^10 years.