Questions

It seems that the longer an agent exists, the greater the chance for a mesa-optimizer to appear. Or does the agent somehow reduce this probability during its existence?

What exactly do agents compete for in models? What constitutes an ecological niche in their relationships?

How does agency shift when past agentic models transfer control of the system to other agents?