It's replacing a wholly-incorrect model with a potentially flawed, but partially-validated one. I don't think we have the body of knowledge right now to precisely model how a particular diet works vs. another, but we have observed that low-carb diets improve health and wellness.

This isn't to say we should all ditch our bread and pray at the Church of Paleo. But with our current state of knowledge, I think it's more intellectually dishonest to cling to provably false "scientific" models of how different diets affect the body, than to just go with something that works (but we don't really know why). The scientific method dictates that theories must be made to fit observation, not the other way around.

Consider this: a theory that the world was round prior to any evidence of such has no more validity than the flat-world hypothesis. Post hoc, you may find out that the Earth is in fact round, but the knowledge was gained from the observation, not the theory. So call it religion if you want, but it's actually more scientific than the alternative.

This particular "religion" is backed by studies that say it's the most effective way to eat to lose weight. Granted, the only studies I know about are the ones Taubes cites, but is there evidence to the contrary?

It seems to work for a lot of people and the science behind it makes sense to me. I don't see why it should be discouraged.