Google Code Analyzes the Markup of a Billion Pages

The folks at Google Code have published some interesting analyses of HTML markup statistics and trends, complete with nice bar charts for us visual learners (Google Code: Web Authoring Statistics):

In December 2005 we did an analysis of a sample of slightly over a billion documents, extracting information about popular class names, elements, attributes, and related metadata. The results we found are available below. We hope this is of use!

They have separate reports covering a large number of categories (elements in pages, attributes in elements, classes, metadata, the body element, link rel microformats, the a element, custom markup from web editing tools, and more). Each of the separate mini-reports is well worth reading, with a lot of amusing “stupid markup tricks” revealed.
(via reddit: what’s new online)