But you're right to be skeptical in general. Nutrition research is in such scientific wilderness that it's hard to consider any guidance at all reputable in any way.
So it's tough to filter through a wide variety of literature to find the reputable material. I try to focus on meta-analyses that cover a lot of ground by authors I consider reputable (and aren't selling a diet themselves).
There's a meta-analysis from 2014, examining a wide variety of studies of diets (ranging from fad to institutionally recommended). It concluded:
"A diet of minimally processed foods close to nature, predominantly plants, is decisively associated with health promotion and disease prevention."
I take your logical point about how "don't overeat" begs the question, but this is where Davidson's charity is helpful. The best way to read "not too much"--if you follow Davidson's advice and assume Pollan is rational--would be to read it as a shorthand for something like the following recommendation:
Calibrate how much you eat to avoid obesity, because obesity is correlated with significantly lower health outcomes over the long term.
Caveat, some people should shoot for a BMI < 35, some should shoot for a BMI < 27.5, so precise general advice is probably impossible.
Tautology or not, it's actually pretty easy not to eat too much, considering that our bodies are pretty good at telling us when we've had enough.
All you have to do is eat a moderate amount, wait twenty or thirty minutes (for your stomach to start digestion, send signals to your brain, etc), see if you are still hungry, and eat a bit more if you are. If you reach the point where you feel 'full' you've probably overeaten. Not complicated, although many people, myself included, don't always take the time to listen to their bodies or have the self control to stop eating when they are chowing down on something delicious, even though know they've probably had enough.
