By including filtering options based on the article level metrics data PLoS has been building, the new engine allows smarter, crowdsourced searches in addition to the traditional indexed content type. For example, consider this (malaria, sorted by "relevance") versus this (malaria, crowdsourced by most views).
But there's a more important reason why this is important. This is embodied by the Scholarly Kitchen crew banging on about how scientists shouldn't waste their time messing about with social media and should get back in the lab and do some research. But it's also typified by majority of scientists I talk to about this asking, reasonably enough, What's in it for me? Incorporating socially-generated data such as article level metrics into scientific workflows answers that question. By contributing to social datasets, you improve the quality of them for yourself and for everybody else. And that's a good enough reason for scientists to engage with social tools.