If X's purpose is still relevant, then retaining and teaching about X's original purpose provides greater incentive for learning and teaching X, making X more likely to be retained. But if X's original purpose is not still relevant, then retaining knowledge of the original purpose is a disincentive to learn and teach X itself, making X less likely to be retained. So, given that X is still taught, learning that its original purpose is known is evidence that it is still relevant; whereas learning that it is not known is evidence that it is not still relevant.

If you are using the model of memetic selection, then useful things Xs are unlikely to have true explanations of why they are useful attached to them, but the most virulent ones. Sometimes they are the same, but obviously often they aren't. After all Robin Hanson gets a lot of low hanging fruit showing us how for example school isn't about learning etc.

Sometimes the most persistent combination would be a behavior or practice without an explicit explanation at all.