Caplan’s basic argument is that labeling isn’t a big problem and that a strong identity in terms of an “ism” or its equivalent can actually be productive. In short, putting on goggles can sometimes help you see things you might not have otherwise. Of course, Caplan recognizes the costs of labels too, but he doesn’t think you have “to salute the intellectual equivalent of the Swiss flag” to avoid/minimize them (I think this phrase is a great one, btw).
Wilkinson, on the other hand, thinks that labeling – or more accurately – politics makes us dumber and more callous. Thus he rejects the libertarian label others try to tag him with – and not just because, as he admits, he isn’t actually a libertarian (!).
My two cents is that labels are useful heuristic devices (mental short-cuts) that efficiently provide a huge batch of information. Yes, they can simplify or distort like all heuristics. But they can also convey a great deal of information so that a conversation can move forward without having to set up every piece on the board perfectly before the next rhetorical move. Whether as self-identification or trying to capture the views/thoughts of others, using labels, like other descriptive words that aggregate information, isn’t the problem. The problem is when we purposely distort the views/thoughts of others for our own ends or when we simplify when more granularity if required for the relevant engagement. But does a conversation partner really lose “15 IQ points” or so when he calls me a libertarian during a discussion of U.S. drug policy even though that fails to capture my entire perspective on drugs (should be legal to manufacture, sell, and use without permission or barrier except in the case of antibiotics which should be controlled AND morally wrong to use or use in excess those drugs that are generally inconsistent with human flourishing)?