We all know that people smoked so they'd stay thin. While people who chose to smoke had a lot of delusional ideas, this particular aspect of smoking had some basis in reality. Recently, there's been coverage of a Science article which found a specific receptor (in mice) that could be a mechanism for this phenomenon and also a potential target for weight loss drug development.
Sample secondary coverage: http://www.medscape.com/viewarticle/744382
R. mentioned this to me, and when I finally remembered to go look at the coverage, it reminded me how annoying I found Kessler's focus on the 1980s as a time when adults in the US started gaining weight in a way they never had before.
The article at JSTOR above is a 1997 study based on a 1980s data series from WHO MONICA. The data was across many countries and it showed a consistent inverse correlation between smoking and BMI -- until populations started including a smaller fraction of smokers and a larger fraction of non-smokers [ETA: whoops. ex-smokers].
Kessler _completely_ ignored smoking as a possible factor in collective weight gain starting in the 1980s. Many, many things invalidate his argument, but this is arguably the worst of them all.