I think the biggest mistake we make is thinking that we can control for one variable and have all the answers. Seems like that's your point. The only time complexity like this gets "solved" is if we break it completely. Then there's chaos and we don't know what's next. Not much of a solution. The bomb calorimeter is a useful tool but it only supplies one datapoint. Knowing the number of calories tells us nothing about the composition of those calories and how they affect insulin, fat storage, and hunger. But that's a start. Combining calories with other measures is the way out of the confusion, not abjuring all science. Science preserves a lot of hard earned knowledge gained from trial and error. Throwing it all out simply takes us back to the drawing board and we don't have that luxury in a world of 8 billion people.