I subscribe the "calories in < calories out = weight loss" philosophy, but I understand when people do not. It's not easy to measure calories in and calories out.
Food calories in labels are the actual chemical calories in the food. As in, if you burn them, how much energy do they release. The problem is obvious. You don't burn food for energy, and it is perfectly natural that different people have different efficiencies in converting food energy. 1mg of fat is 9kcal in the label, but perhaps I use it at 50% efficiency and you use it at 40% efficiency. It makes for wildly different dietary outcomes.
Calories out suffer from the same problem. You can estimate how much is my basal energy quota or how much energy I use in running a mile. However, I've never seen the variance of this measurement anywhere. There is a variance, and it is acceptable to assume it may be relevant.
If you add to that the glycemic properties of food -- as in the evolution of blood sugar over time upon food intake -- and again, its variance across individuals, and there's a whole lot of unacknowledged error in the simple philosophy.
How to tackle modelling error? As ever in engineering. Measure input, output, and do a regression analysis.
Food calories in labels are the actual chemical calories in the food. As in, if you burn them, how much energy do they release. The problem is obvious. You don't burn food for energy, and it is perfectly natural that different people have different efficiencies in converting food energy. 1mg of fat is 9kcal in the label, but perhaps I use it at 50% efficiency and you use it at 40% efficiency. It makes for wildly different dietary outcomes.
Calories out suffer from the same problem. You can estimate how much is my basal energy quota or how much energy I use in running a mile. However, I've never seen the variance of this measurement anywhere. There is a variance, and it is acceptable to assume it may be relevant.
If you add to that the glycemic properties of food -- as in the evolution of blood sugar over time upon food intake -- and again, its variance across individuals, and there's a whole lot of unacknowledged error in the simple philosophy.
How to tackle modelling error? As ever in engineering. Measure input, output, and do a regression analysis.