"Most predictions are wrong, so we should ignore them all" seems bad advice. Maybe we evaluate them, question the assumptions, check the reasoning, and validate the data rather than ignoring.