A problem of convergence

[From Bill Powers (920608.1400)]

David Chapman (920606) and Phil Agre --

I've read some papers that Agre sent me (much thanks!). I'm left wondering
what to do. If I come over and try to play your game on your field, I'll be
too ignorant to say anything interesting to you. If we stand on our
respective fields and throw fastballs at each other, it won't be much of a
game. Although it would be a dirty shame, it's possible that our worlds
simply don't intersect over a large enough area to support a dialogue. I
hope that this isn't true.

I think that HPCT redefines some fundamental problems in modeling behavior.
It doesn't substitute for analysis of symbol-handling systems, and it
certainly has some family resemblances to the guts of Pengi etc. But I just
know too little of your interests to serve as an interpreter in any way
that would be useful to you. You're hell-bent-for-leather in a certain
direction, and if I try to get in the way I'll just get trampled.

At the risk of sounding standoffish or condescending, I think there's only
one practical way to find convergences between your approach and mine, and
that is for you (two) to learn HPCT. If you could take the time to pretend
to be students again for a while, and learn the basic principles behind
HPCT, you would be by far the best judges of its meaning in your own work.
I know how that sounds and I apologize for the implied hubris. But I just
can't think of any other way to go about this that won't lead to endless
misunderstandings and arguments about words. In Agre's articles I can see
opportunities for hundreds of frustrating interchanges. But I would greatly
value your participation on CSGnet if we could bypass all that stuff.

As to your suggestions about possible publication of the Gatherings
simulation, thanks, David, but it's moot. A paper on this model by McPhail,
Powers, and Tucker will appear as the lead article in the next issue of
Computational Sociology (if I have the title right).


Bill P.


CC: CSGnet

Well, in fact my interests have gone off in entirely different
directions since the Pengi work, and really I don't think about that
stuff any more. (I think the same is true of Phil.) There's still
plenty of people in AI interested in these sorts of issues, but we
aren't them. Sorry about that...