These are good points. The devil is in the details for how to implement stuff like this, and there are a host of other tradeoffs, e.g., seemingly optimal regimens that people despise and that are de-motivating are, in practice, far worse than theoretically less-good training that people will actually do.
If it was as straightforward as some of the "expertise" research suggests, the world would be teeming with super-beings. That it isn't should give one pause about the relative leverage of constructing the problem this way.