“The research, by industry body PRS for Music, showed the most pirated pop songs tend to be those at the top of the music charts. There was little evidence that file-sharing sites helped unsigned and new bands find an audience, it found. … There was no evidence of the Long Tail operating.” (BBC, see also WaPo, Billboard, Idolator, etc.)
This shouldn’t be a surprise ever since the 2006 Columbia University study that showed pretty convincingly that popularity tends to breed popularity whether on the Internet or not: When facing a big list of music, even if you have sampled each song, most people are apt to decide that the best ones are the ones other people also like. (Snobs will immediately get their sneers on here, but as I say in my book I think this is mostly a reflection of a desire for connectedness, and a certain faith in the wisdom of crowds, both of which are positive traits albeit with potential for abuse.)
It’s also notable that the Big Champagne study found that most people followed this pattern because otherwise they were overwhelmed by choice (you’ve probably run across Barry Schwartz on that paradox). All these factors make it seem to me (perhaps self-servingly) that the obsolescence of the critic in the file-sharing age has been exaggerated. Yes, there’s some changing of personnel at the gates and the way critics play their roles but people still need to outsource some of the search for needles to a designated haystack sifter.
What’s more the ensuing exchange of information and opinion is the primary way that these choices become meaningful. As one of the researchers, Andrew Bud, told The Register: “… it’s through people chatting to each other and seeing the music talked about in the media. That’s what culture is.”
PS: A caveat here. Zoilus friend Sean Michaels points out to me that big data errors were found in the same group’s study last year of music sales online. If they get the relatively clear info about legal sales wrong, one does have to wonder how well they crunched the much more nebulous info on file-shared tracks. I haven’t had time to examine the data; I am just pointing it out.