Taube's theory (about cholesterol, but I think it applies here as well) is that it's "where the light is". Things which are measurable are optimized for even if

Taube's theory (about cholesterol, but I think it applies here as well) is that it's "where the light is". Things which are measurable are optimized for even if they're pragmatically pointless. So a study comes along that finds a correlation between sodium intake and cardiovascular disease. Totally possible. The USDA looks at those numbers and says, we're eating Xmg of salt, we should be eating Ymg. They make a recommendation, years go by, and they look at the numbers again and, hey, salt consumption is down to Ymg! Success! What about mortality? Who knows. That's not their department.