I see that I do have some new readers. Gene Callahan says

Your “ridiculous” answer on obesity seems obviously correct to me — maybe insufficient, but fine as far as it goes — and the answer to the “puzzle” also obvious — adjust calories in and calories out until you are at a stable weight that you like. What is “ridiculous” about that?

The first issue is that saying one ate too much or too exercised too little is a non-answer. It simply takes the equation of motion and applies a normative frame to it.

I could say that rocks fall because they have too little resistance to the propensity to fall. This is a non-answer. It simply takes the analytical notion that resistance impedes motion and the observation that things fall and applies a normative frame to it.

How would we know if the resistance is too little? Because the rock fell. How would we know if the resistance is enough? Because the rock did not fall.

This tells us little, except that you are aware that things fall.

What we would like is something that takes parameter values and then tell us what result we should expect.

Now, the second part of Gene comment is slouching towards a model. He words it as instructions but if we recast it as a general framework we get something like this.

1. Weight change is the residual of the choice variables calories-in and calories-out
2. When people are unsatisfied with their weight the adjust one or both of these variables to make the residual positive or negative.
3. When they are satisfied with their weight they adjust one or both of these variables to make the residual zero.

This has the virtue of being the beginning of a model. It has the vice of being at odds with the facts.

Lets set aside for a moment the issue how would we know if people are satisfied with their weight because I think it is too emotionally loaded. Instead, lets investigate the dynamics of this model with exogenous shocks.

First, imagine a young teenager who is temporarily at some stable weight and so presumably is satisfied and has chosen a residual of zero. Then the teenager hits a growth spurt. Under our model the teenager’s weight does not change.

He or she will grow taller but will not grow heavier. And, so BMI will fall and at the square of height change. If at this new BMI the teenager is unsatisfied with his or her weight then he or she may alter calories-in and calories-out to achieve a new weight, but in the absence of such conscious alteration weight will be fixed as height extends and so BMI will fall.

Does this appear to be what’s going on? Are teenagers accidently getting thinner as they grow taller and then moving to correct that?

Second, imagine a woman who gets pregnant. Again, under our model her weight will not change. Thus in the absence of conscious alteration of her caloric balance her core body will lose at least as much weight as the fetus and placenta gain. Indeed, she will lose more because the fetus and the placenta produce metabolic waste.

Now she may become unsatisfied with the loss of core body weight and move to alter caloric balance but in the absence of such unsatisfaction she will lose core body weight. Moreover, once the fetus and placenta exit the residual will exogenously rise and her core weight will move back to its original amount.

Does this seem at all like the patterns we observe?

Not to give away the show, but one might suspect that there is a mechanism which works to adjust either calories-in or calories-out in response to growth spurts and pregnancy. Our questions might be, does such a mechanism exist? How does operate? Is it present even when these events are not occurring?