A few years ago I was talking with a friend about using access frequency (popularity) as a factor in ranking web pages. He pointed out that this works well for dead trees materials, too.

He used to go to the university library and start down a row of bound math journals. He’d pull one out and look at the non-bound edge. If there was a section where the page edges were dirty, that was a paper that lots of people had read. So he would read it. Then go to the next volume. Going purely by popularity lead him to top papers in areas that he might not have looked at otherwise.

Interestingly, this only works when the materials are shared, like in a library.