Matt writes a post which displays American diet composition in 1970 vs 2008.

He concludes

If everyone ran an hour a day at eight miles per hour, that would actually make up for the increase, but obviously that’s not what’s happening.

The experimental and historical evidence suggests that this is not the case. The experimental evidence says that if we induce people to exercise more and let them eat freely, they will on average not lose weight. Most people will change weight but some will lose and some will gain.

As always, no one disputes that wedging will result in weight loss. That is driving calories-in and calories-out in opposite directions will lower the caloric content of the body. Excluding water there is a rough relationship between caloric content and mass. For most fat, which is our primary concern in obesity, the relation is about 3500 calories per pound.

The tendency of all animals is to try to get calories-in and calories-out to move in harmony. If you have to wedge then that means that this system has failed. At a minimum we would like to know why.

Historically people worked a lot more than they do now. They also ate a lot more than they do now. From the year 1400 to 1970 average calories expended fell dramatically but so did average caloric intake. Obesity was never a severe problem. The system did not fail.

Then from 1970 to 2010 average calories expended actually rose but calories consumed rose more and obesity exploded. The system failed.

If you go and look at the actual graph though, you can see Gary Taubes’s thesis on display. This is natural since the dataset used to make the graph is one of Gary’s favorites. First you see meat falling, then fat falling/stalling, then sugar falling as various healthy eating theories rose to prominence.

The only thing that rises consistently is grains. Gary insinuates and sometimes outright says that obesity was caused by the encouragement international health authorities for people to eat more grain. The naturally tendency is for people to eat more meat as calories become easier to obtain. He suggests that consciously overriding this mechanism led to an excess of insulin and possibly deficiency of peptide YY, which are key regulators of caloric balance.

I am skeptical of this theory, but it is one that at least recognizes the underlying theoretical problems.

Oh and just for uber-nerds, Andrea Jezovit, the chart creator says

By calculating such food losses, the USDA data closely approximates the amount of food that actually makes its way from the farm into the average American stomach.

This is hotly debated. The general charge is that the USDA loss adjustments are outdated at best and methodologically suspect in the first place. The charge that comes up the most is that the USDA is overestimating fat consumption because they are not properly accounting for the way fat is used as a cooking medium.