The problem is that it takes giant leaps backwards, too. See the discussion in other comments about margarine, saturated fats, and the links to heart disease. The scientific recommendations that were made promoted the very problem they were trying to address. As others have pointed out, this problem may largely be attributed to misinterpretation and miscommunication (rather than bad science). But since the vast, vast majority of people get their scientific information from schools or media rather than reading papers directly (which isn't a bad thing, since scientific papers are not written for interpretation by non-scientists), that's not a helpful distinction.
It does move backwards. But that's good, too (for science, at least). Sometimes, when people generate hypotheses, and those hypotheses are wrong, they still seem to work when tested. That's just statistics. So someone tried something, it worked, and they reported it. Eventually other people will try it too, and will realize that the original hypothesis was wrong. And thus it gets corrected. But there's no way to avoid that apart from not sharing ideas, and it's vastly preferable to the other option—that wrong ideas never get corrected in the first place.
The problem is that it takes giant leaps backwards, too. See the discussion in other comments about margarine, saturated fats, and the links to heart disease. The scientific recommendations that were made promoted the very problem they were trying to address. As others have pointed out, this problem may largely be attributed to misinterpretation and miscommunication (rather than bad science). But since the vast, vast majority of people get their scientific information from schools or media rather than reading papers directly (which isn't a bad thing, since scientific papers are not written for interpretation by non-scientists), that's not a helpful distinction.