Interesting approach especially the idea of treating failures as state signals rather than exceptions. That’s a subtle but important shift from traditional orchestration thinking.
I agree with your critique of DAG-centric systems: they work well when the world is deterministic, but real production environments are adversarial, noisy, and stateful. An OODA-style control loop feels closer to how robust distributed systems behave under uncertainty
No comments yet.