[austinjp]

Indeed. See Simspons's paradox, sub-grouping, and so on. There are deep problems in: the lessons we can learn from a population, vs how a specific individual will respond to a specific intervention; plus the filthy complexity of real life outside the lab.

Basically it boils down to this: we don't know where you as an individual lie on any of a huge number of bell-curves. Statistically, you're likely to be in the middle of all of them -- and yet it would be impossible to find an individual who actually is. So the more complex an intervention (and running is a very complex situation) the more trial and error you'll probably have to go through.

This is not a popular thing to say.