At 2:50 PM +0300 10/23/00, Sampo A Syreeni wrote:
On Sun, 22 Oct 2000, Tim May wrote:
The book I recommended a week or two ago, Judea Pearl's "Causality," is much more advanced in its mathematics. (But the math is important if one is actually trying to construct the causality diagrams Pearl is talking about.)
Would it be too much to ask you to recant the main point made? It sounds pretty interesting...
I'll recount it, but not recant it. Think of spacetime diagrams, a la the lightcones of Minkowski diagrams. Events A and B precede Event C. The same kind of diagrams obviously apply in ordinary events, without regard to the speed of light. A directed acyclig graph (DAG) of various events, some in the "causal chain" leading to some Event C, some outside the causal chain. Pearl addresses Bayesian networks in terms of DAGs and provides tools for analyzing when events actually "cause" other events. Of great interest for deciding when, for example, some drug test produces meaningful results, when some legal proof of causality is being challenged, etc. Pearl doesn't produce some magical formula for separating causes from non-causes, just a bunch of theorems and corollaries which may be useful in policy analysis, experiment design, and just thinking about the world. I think of it as a kind of "network analysis," akin to tools for circuit analysis (like Kirchoff's Law, for example). More discussion is of course available at Amazon or in search engines. --Tim May -- ---------:---------:---------:---------:---------:---------:---------:---- Timothy C. May | Crypto Anarchy: encryption, digital money, ComSec 3DES: 831-728-0152 | anonymous networks, digital pseudonyms, zero W.A.S.T.E.: Corralitos, CA | knowledge, reputations, information markets, "Cyphernomicon" | black markets, collapse of governments.