Google's PageRank is precomputed relevancy ranking, where the heavy lifting of actual relevancy ranking is done by us humans. Why is this important? I was re-reading A new comparison between conventional indexing (MEDLARS) and automatic text processing (SMART), which lays out how computerized indexing can beat the best manual indexing by:
(It's more complicated than that, but you get the idea.) Relevancy ranking is the hardest part of the indexing job, as there are no clear-cut algorithms for relevancy ranking with both excellent precision and excellent recall (getting all of the documents you want and none of the documents you don't want). Google's PageRank works around the difficulty of relevancy ranking by handing the hardest part the ranking of individual documents to us humans. You can get good results from proper metadata, but metadata is useful only in environments where no one has interest in gaming the metadata (I wonder if it should be called "The Semantic Intranet"? That's where Semantic Web technologies really make sense to me.)
The original paper is worth a read, especially if you work on software that incorporates search and these days, I suspect that almost any non-embedded program could grow to a point where it incorporates a search mechanism (and an email client, and a web browser you get the point).