00100	Conversations with Minsky, Papert and Hewitt.
00200	
00300	I had some difficulty maintaining the usefulness of predicate
00400	calculus representations of knowledge.  The important criterion
00500	seems to be epistemological adequacy.  Namely, it is not clear
00600	that the procedural representations of info favored by them
00700	can be used to represent the information actually transmitted
00800	between humans.
00900	
01000		A key problem to try to axiomatize is that of attending
01100	the Japan AI conference in all its gory detail including:
01200	
01300		1. What other people will or might do if induced.  Represent
01400	them by parallel indeterminate programs.  (Side thought: Suppose we
01500	have parallel indeterminate programs with <before> and <after>
01600	statements.  In general, a set of such programs will be inconsistent,
01700	but under suitable conditions they will be consistent and replacable
01800	by programs with a specific synchronizing mechanism, a la Manna
01900	and Ashcroft.  It may be easier to write, and prove the properties
02000	of the programs with the before and after.)
02100	
02200		2. The location of information in documents and in people
02300	who will say it when asked.
02400	
02500		3. What we know now and what we know about information
02600	becoming available when needed.
02700	
02800		4. The form of the program for going to Japan and the 
02900	pseudo-proof that it will work.