An "agent" is a bit of code that has a goal and travels around the net in fulfillment of that goal. It gathers information as it goes and reports back to its creator on its progress. After it fulfills its goal it may stay around as in information provider.
When two (or more) such agents meet a "conversation" will ensue in which each will try to communicate its needs and its services. If they can discover a mutually beneficial relationship they will exchange information. Each must have the possibility of gaining from the transaction, though in a given transaction not all will benefit.
The goal is to build a swarm of a billion or so such agents, each negotiating with each other. This is an experiment to see what develops from such a swarm. Will something like intelligent behavior emerge? Will anything emerge? The project consists of designing and building the swarm and then observing its behavior.
Note that an individual agent is not intelligent. We seek intelligence in the swarm.
An agent must
Optional
eg.
A place to begin looking
Last Updated: May 5, 2002