It's google's success undermining the assumptions behind pagerank (!) leading it to perform less well. Before google got so effective people put more work into

It's google's success undermining the assumptions behind pagerank (!) leading it to perform less well. Before google got so effective people put more work into maintaining collections of links to other sites; hand-curated omnibus directories like yahoo and dmoz or narrower fields. As search became better this was increasingly a poor use of time; you could, after all, just go and google stuff to find it again, so why bother maintaining a hand-curated set of favorite links. A consequence of this is that link-actions-taken-for-reasons-consistent-with-pagerank's-assumptions dropped, and reducing the signal-to-noise ratio insofar as pagerank is concerned; similarly link-actions-taken-for-reasons-contrary-to-pagerank's-assumptions increased (SEO), further dropping the signal-noise ratio. Even extremely robust SEO-detection techniques can only cut away the increase in noise; now that people are less incentivized to publicly post high-quality hand-curated link collections there's less for the algorithm to go on. This is why there's talk of twitter and fb vis-a-vis search; twitter and fb are where the info the search engines need currently lives. (!) I know they use additional techniques these days but whatever they're throwing into the mix doesn't seem to be counteracting the overall utility decay.