You like causal questions in messy logs.
Confounding, off-policy evaluation, calibration, placebo tests. You have shipped something that ran unattended and told you the truth when it broke. You would rather report a negative result than tune it away.
Write with something you built