February 26, 2006 - 22:54
Projects:
As I was drafting my comment here in response to PG's blog entry, another thought (irrelevant to PG's thread) occurred to me:
Is there an agent in a CA?
We all know that a defining characteristic of a CA (or an emergent system) is that there is no conductor or director. But what about an agent? What is "doing" a CA? Is the computer the agent? If you read Wolfram's formal definition of a CA, you might begin to think that the rules are the agent. The rules themselves might be the "dynamically interacting rule-based agents" (funny concept there...are CA rules fundamentally "rule-based""?) that are the hallmark of an agent-based model. If so, aren't CAs just a subset of agent-based models?
But what if there is no agent in a CA?
Or could we consider the agent to be some sort of an "invisible hand" that implements the rules? Is this "invisible hand" the automata of a cellular automata? Could this be similar to what Smith (and others) have conventionally thought of as an invisible hand?