[wyager]

> Nutrition research seems to change as often as JavaScript framework of the month.

Nutrition “science”, like most health-related “sciences”, has approximately zero predictive power. A pretty reasonable strategy is to completely ignore whatever helpful advice the FDA or whomever is currently spouting (food pyramid!) and come up with some plausible first-principles model that matches available anecdata, like trying to match a plausible far-ancestral diet.

[amanaplanacanal]

Agreed. The problem is that doing actual controlled studies is really hard. You have to take a group of people and randomly assign them the diets you want to study, essentially locking them up for the duration. It could possibly be done, if one of our billionaires took an interest in this problem.