Tagging is useful (?)

Read that in two ways, as declaration or interrogation. Plenty of people claim tagging is useful, it seems a reasonable thing to say, but… where’s the empirical evidence for this? Pet peeve: people who argue dialectically (without saying that’s what they are doing). People like Clay Shirky, who as far as I can tell has no research to back up his multi-faceted statements. By research I mean following a scientific, empirical approach (where empirical means ‘grounded in real-world data’). The obvious problem with this approach — and you can see how hard it is from how *few* scientific articles in computer science follow it — is that it requires one to put oneself on the line. When you argue empirically, you open up to attack: it now becomes clear what your validity claims are (such claims are nonexistent in dialectical argument). Once clear, they can be criticized: ‘your subjects aren’t representative’, ‘that construct isn’t appropriate’ etc.