It's interesting to note that the 5th highest ranked publication is arXiv. For those who aren't familiar with it, arxiv.org is an open-access repository of academic papers, mostly in quantitative science. In my field (computer science) it is standard practice to deposit a copy of one's papers in arxiv before submitting them for publication, and arxiv is the place to find the latest research.<p>There is currently a lot of hand-wringing in academia about open access publications. Everyone wants it, and it is trivial to switch a field to it (machine learning has done so, for the most part), but it requires the leaders in the field to lead the change and they are normally too invested in the status quo. What the high ranking of arxiv suggests to me is that while people maintain lip service to the idea that the (mostly closed) publications are important and the maintain the definite version of a publication, the reality is that no-one gives a damn and goes to arxiv when they want to read something.