As CronoDAS has pointed out, the field of nutrition science has a long history of releasing bad studies leading to conclusions that were later proven false. Here are some of the stupid things which researchers have done, which have lead to bad dietary advice:

1. Inferring what's good or bad for humans from its effect on lab rats
2. Studying a diet's effect on one particular disease, then using it to argue for its effects on overall health
3. Conducting a controlled study, losing half the sample to dropouts and non-compliance, and pretending that the group that dropped out did so for a reason other than because the diet was hurting them
4. Measuring weight loss, but failing to distinguish between loss of fat and loss of muscle.

I believe Taubes is correct, and that the idea that low-fat diets are healthy is due to errors 3 and 4. What convinced me was hearing a bodybuilder talk about fat and muscle weight as separate things, and connecting that observation to the "rebound effect" - that is, the observation that people who lose weight on low-fat diets tend to gain it back plus extra. My interpretation of this is that some of the weight they lose is muscle, not fat, and this lowers their metabolism. This makes sense because maintaining muscle requires protein, and low-fat diets are usually also low-protein diets; the most common protein-rich foods, meat and eggs, are also high in fat.

This also explains why vegetarians don't have the same problem with low-fat diets: they're specifically instructed to be careful about their protein intake.