Over at Google Code they ran a survey, in December 2005, looking at a couple of webpages trying to find out which elements and their respective attributes are used most. And more importantly how they are used.
We took a sample of slightly over a billion documents, and looked at what elements were used on the most pages, what class names were used on the most pages, and so forth.
Pretty interesting read this Web Authoring Statistics study.
E.g. why would anyone use a <table>-tag and not put any <td> or <tr> inside? Beats me… Is it a remnant of MS ‘HTML’? Or someone deleting a table in a WYSIWYG environment? And there are more examples.