On Sun, Jan 19, 2014 at 01:58:43PM -0800, Edgar L. Owen wrote:
> Russell,
> I agree that your model here is "theoretical" and does NOT apply to the 
> actual reality of decision making organisms such as humans. My comments DO 
> apply to the real world.
> Rational agent theory properly applies to only extremely limited and 
> non-representative cases in the real world. Specifically it applies to 
> simple well definable games, checkers would be an example, with well 
> defined 'best outcomes', and well defined rules that all rational agents 
> are able to understand in the same way with no ambiguity. But the vast 
> preponderance of decision making situations in the real world are not thus 
> well defined and rational agent theory does not apply for the reasons I 
> explained in my previous post.


> Even in a well defined game situation it is quite possible to act 
> rationally and NOT make the optimal move. E.g. in chess or Go it is 
> impossible to know what a true optimal move is because optimal moves are 
> not computable except in the far end game. Nevertheless an agent can act 
> rationally by choosing the best move he can compute having a limited 
> understanding of the game.

Such situations are handled by "bounded rationality theory". A
fully rational player in chess or go would have god-like computational prowess.

> We can understand this better by noting that an IRrational agent is NOT one 
> that is unable to compute an optimal move from the knowledge he has (if he 
> is a novice at Go he will simply be unable to compute the best move from 
> complete knowledge of the board and the rules even IF he is rational). An 
> IRrational agent CAN compute an optimal move but rather chooses a 
> SUBoptimal move after computing an optimal move because he has some e.g. 
> other agenda than winning... Or an IRrational agency computes with faulty 
> logic. So there are two types of IRrational agent. All the other decision 
> making is rational.....

Or computing the winning outcome takes too long, so it is better to
make some decision rather than none at all. Think Chess with a clock.

Or being predictable allows competitors to exploit you via arbitrage,
or whatever.

... and so on ... 

Many more than two types of irrational agent.


Prof Russell Standish                  Phone 0425 253119 (mobile)
Principal, High Performance Coders
Visiting Professor of Mathematics      hpco...@hpcoders.com.au
University of New South Wales          http://www.hpcoders.com.au

You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to