[grey_whiskers]

People often make the mistake of using "crossstream" (inferences from a population of sequencess of a given length, sometimes called "weak") computations to model "downstream" (inferences along a given sequence, sometimes called "strong") computations.

Couple of off-the-wall observations here.

1) Why does this bring the word "ergodic" to mind?
(look it up on www.m-w.com...)

2) This also seems like the way I have been told some
drug companies and others investigate the effects
of chemical additives on organisms--overdosing
on a given chemical all at once to simulate
cumulative, long-term exposure.

Cheers!

[Doctor Stochastic]

In an ergodic system, the crossstream (note the artificially induced sss) and downstream properties are the same. Coin tossing games are not that way. The properties of 10 games of matching pennies with 10 tosses are not the same as that of a single game of 100 tosses. (Trivially, one cannot be behind 11 points in a 10 toss game.)