I left a comment on the Nature commentary on this paper. Probably it should have been left here, so I'll repeat it.
Congratulations to Curry et al for managing to have an impact on what I now like to call the legacy publishing industry. It's taken far too long. It was 1997 when Seglen et al pointed out that the number of citations that an article gets is not detectably correlated with the impact factor of the journal in which it appears. I emphasized it again in 2003, in this journal [Nature].
After 20 years of pressure, it seems that something may, at last, be done about it. Not enough, of course. The impact factor has had such a corrupting effect on science that it should be forgotten altogether. Perhaps, when citation distributions are published, it will become so obvious that its mean is such an absurd measure that will happen. Expect that to take another 20 years.
The really interesting question, though, is whether Nature or Science will still be publishing original research 20 years hence. By then I expect they will be reduced to news journals, and research will be published in far cheaper journals, with fully open access. Even if that doesn't happen for good scientific reasons, it may well happen because universities can no longer afford the huge cost of Elsevier, Cell Press and NPG journals.