Why this works for agents

• An agent that is really a FSA is simple.
• It doesn't have much memory (it just has its current state).
• However, it can do a lot of complex things.
• This gives the illusion of intelligence.
• Buckland thinks that this is a nice way of encapsulating code, avoiding spaghetti code.
• I'm not sure about that, but hey why not.
• Randomness and empty transitions can help.
• Note that agents can send input to each other, so you can get emergent behaviour.