I thought the hard part was getting the machine to act. It wasn’t.
The harder problem was deciding what had to survive when the machine changed, the session ended, the model was replaced, the runner disappeared, or the evidence disagreed with the worker.
Read the GhostMesh thesis