🚀 go-pugleaf

Jerry_LR_Chandler <chandler@helix.nih.gov> writes:

>Information theory (IT) presupposes a fixed (invariant?) probability 
>distribution.  From you perspective, what are you interested in distributing?
>What attributes of human beings do you select for analysis?  

Jerry,

   Information theory doesn't start to get interesting until one compares 
two different probability distributions (which could be measurements on 
the same physical system taken at two different temporal intervals.) 
Examples are the Kullback-Leibler measure of discrepancy (information.)


>Methods to integrate the dynamic interactions between the three IT 
>probability distributions may eventually be developed by mathematicians.
>(Indeed, if I read some of the earlier postings on this newsgroup 
>correctly, such methods are not completely intractable.   :-)  )

Well-defined information measures already exist for 3 or more dimensions. 
These are the "higher moments" that Don Foster was referring to. 
Perhaps Chris Hillman is lurking and would like to comment.

Cheers,
Bob