It took four months, but I finished reading Gary Taubes' masterwork of science journalism, Good Calories, Bad Calories. If what he writes is correct, this could be the most important popular health and diet book written in a very long time.
The first half of the book is a long and careful investigation into the science behind diet and the "diseases of civilization": obesity, type-2 diabetes, heart disease, cancer. The conclusion is grim: all our common wisdom is poorly founded. "Everyone knows" that people get fat from eating too much and not exercising enough, that eating fat makes you fatter, and that eating cholesterol causes blocked arteries. But those statements are all scientific hypotheses, not axioms, and it turns out the experimental science doesn't confirm them.
Taubes looks in detail at the science and politics behind current government recommendations: low fat diets, exercise and calorie restriction for weight management, etc. And frequently finds problems with the science like inconvenient data ignored, alternative studies dismissed, and personal bias overcoming good science. He also dives deep into the politics. One of the most shocking things is the way public health officials have basically run a giant nutrition experiment, advising Americans to eat low fat diets with no solid evidence that's actually healthier.
The last part of the book examines an alternate dietary hypothesis: that carbohydrates are the source of our health problems, not fats. There's research that indicates that the insulin response from eating refined carbohydrates causes your body to store that sugar in fat cells rather than using it as energy, leaving you still hungry but fatter. This metabolic response seems like it could be very interesting for understanding both obesity and type-2 diabetes. If Taubes' book encourages more research in that area, it's a success.
The best thing about Good Calories, Bad Calories, and also the most demanding, is how thorough the journalism is. This isn't some hot screed selling a fad diet, it's a boring, meticulously footnoted examination of the actual original science journal articles that led us to our current dietary understanding. It makes for slow reading and a frustratingly muddy final story. But it also feels true, much truer than the platitudes, moralism, and oversimplification that is the usual diet discourse.
Definitely worth reading if you are interested in diet and metabolism, or just interested in a detailed story about public policy and science. See also his (much shorter) 2002 NYT article What if It's All Been a Big Fat Lie, this somewhat negative review of the book, and this positive review.
Update: it's going against the spirit of the book to synopsize it like this, but here's the author's summary of his conclusions in a short 10 item list.