Oh no! All the previous posts died in a fire. All burnt up. Ashes.

Yoni proposed an experiment to measure the relationship between caloric intake and weight gain per individual. It’s not easy, but not insanely difficult. I’m going to start doing it on September 1, running through the rest of 2012.

**Steps**

- Determine current weight and daily caloric intake X that maintains said weight.
- Eat Y calories beyond maintenance for 14 days.
- Eat maintenance calories for 7 days.
- Average the weight readings for the last 4 days of those 7 to see how much weight was gained or lost.
- Diet back to the starting weight.
- Do 1-5 four times, with Y = X+500, X-500, X+1000, and X-1000.

**Helpful**

- Exercise consistently.
- Measure body fat, too (Withings, calipers).
- Try to maintain similar macronutrient ratios (carbs, fat, protein).
- Track other interesting things, like sleep (Zeo), caloric expenditure (BodyMedia armband), cognitive performance (Quantified Mind), mood, productivity, and blood tests.

Classical theory says that the relationship is supposed to be linear with exercise fixed. That is: **f(Y) = c*(Y-X)** where f(Y) is the weight change for caloric intake Y compared to maintenance intake X, and c is a scaling constant, like 3500 calories per pound. Energy in minus energy out is weight change, right?

Yoni’s prediction is that the relationship will be: **f(Y) = c*max(0, abs(Y-X)-Z)** where f(Y) is the weight change for caloric intake Y, abs(Y-X) is the absolute value of the difference between actual intake and maintenance intake, Z is the “metabolic buffering” constant (see later) and c is a scaling constant. The metabolic buffering constant is the degree to which the body adjusts energy expenditure up or down by varying heat production and NEAT (“non-exercise activity thermogenesis”, or more simply, “fidgeting”) in response to changes in dietary intake. His hypothesis is that c and Z are individual-specific, but they have typical values across the population, and that beyond the buffering effect, the relationship will be linear with individual-specific scaling that has low variance across the population (that is, c is nearly constant between people, but is a function of their current bodyweight, or rather, their current intake levels X).

Then there’s the set point theory, which claims that it’s easier to return to the original weight to get away from it. This could apply to either the classical theory or the metabolic buffering hypothesis.

I know of at least one other hypothesis, which is that weight gain would be controlled less by how much one eats and more by how much insulin and maybe ghrelin are released from the foods eaten. I think that this can’t be tested without another experiment that controls for insulin levels, but perhaps you can suggest a way that we could figure this out in this experiment which doesn’t require me to eat like I’m diabetic for four months, which I don’t know how to do accurately.

- What would be your own predictions?
- Please criticize this experimental design and propose improvements.
- Do you want to participate, or know anyone who would?
- What else would be cool to ask/answer/do?

I’m going to find out for myself what calories do to my weight, and I’ll share the data on this blog. I’m planning to continue eating mostly paleo / Bulletproof / low-ish carb and working out a few times a week.

When tryna gain muscle you need a caloric surplus. Usually +500 from maintenance but a lot of bullets will eat 3000-5000 calories a day.