Lucas Gonze has a great, thorough write-up of the Webjay playlist popularity metric he developed. (Webjay was an innovative music playlist sharing site that Lucas created—both Webjay and Lucas are now part of the Yahoo Music site / service.)

Popularity ranking schemes represent one major approach to highlighting important or "relevant" items that would otherwise be lost in a large set of items. In Webjay's case, the popularity metric was used to highlight the "best" (most popular) playlists.

Lucas' post covers a number of interesting details that I think are relevant to anyone developing either a popularity algorithm or a user interface for highlighting popular items. I also like how Webjay defined rank in terms of observed information (aka, so-called implicit metadata), rather than relying on users to honestly input "votes."