.comment-link {margin-left:.6em;}

-Through my eyes-

28 January 2005

Agent Compromises

Suppose you want a set of software agents to solve something using genetic algorithms. There are tradeoffs involved here.

The more dumber your agents are, the more you need them in quantity and the more time you need.
The more slicker/smarter the agents are, the lesser quantity you need and the lesser time you need.

Keeping 'diseconomies of scale' on the side, in the dumb agent scenarios, the more you have them the better.

However, it seems intuitively that with slicker agents only so many can solve a problem efficiently and after a while quantity makes on difference on the efficiency of the problem solving activity. Too many cooks spoil the broth in the scenario of smarter agents I guess.


Post a Comment

Links to this post:

Create a Link

<< Home