[nilirl]

Nice premise but unsatisfactory conclusion.

Experimentation in these contexts is difficult to make good inferences from. I'd like to have read more about how experiments are designed to work well with uncontrolled environments, how to make inference for short-term prediction, and how to plan multiple small experiments to usefully exhaust a search space.

Still, thank you for writing. I enjoyed it. I'd love recommendations for more procedural knowledge if you have any.