We need new rigorously controlled experimental studies to tease out the causation patterns suggested by correlations observed in observational studies of human diet. The way to test a causal hypothesis is always, at bottom, to do a controlled experiment.[1] So we will tease out the effects of diet on different people by finding experimental volunteers and subjecting the volunteers to controlled diets, such as those planned for some of the experiments described in this interesting article.

This is very difficult to do, as almost all human beings eat when they feel like eating WHAT they feel like eating. Earlier human experiments on effects of diet in the 1970s actually required the experiment subjects to live in the laboratory long-term, and to have every gram of everything they ate during the experiment measured exactly by experiment team assistants. Even at that, those experiments came up with few clear conclusions, perhaps because the experiments weren't lengthy enough or didn't include enough subjects for strong inferences. Now the experimenting begins again. Whether the currently hotly debated hypotheses about human diet win or lose, it's important to put the hypotheses to the test of a rigorous experimental study to advance human knowledge.

[1] http://escholarship.org/uc/item/6hb3k0nz