When can a human still change what an agentic system does?

Opening question

At which points can a human still change the course of an agentic system’s activity?

Possible line of investigation

Start with a specific arrangement and trace its activity. Establish what it can access and change. Identify what humans can observe and the points where intervention remains effective.

Questions still open

• How much can be reversed after an action has propagated?
• Does an intervention stop one activity or also prevent further activity?
• What evidence would show that a nominal supervision arrangement actually supports control?

Scope of this page

This is an illustrative outline for a larger piece. No chapter has been written and no empirical conclusion is implied.

Connected thinking

NOTE-001 considers permission and intervention. CON-002 investigates observability.