Hypothesis possibly reconciling link between impulse control and weight, strong heritability of both, resistance to experimental intervention, and society scale shifts in weight:

Body weight is largely determined by the 'set point' to which the body's metabolism returns, hence resistance to intervention. This set point can be influenced through lifestyle, hence link to impulse control and changes across time/cultures. However this influence can only be exerted either a) during development and/or b) over longer time scales than are generally used in experiments.

This should be easy enough to test. Are there any relevant data on e.g. people raised in non-obesity ridden cultures and then introduced to one? Or on interventions with obese adolescents?l