While working on the third edition of Designing With Web Standards, I decided to visit Alexa’s Top 100 US Sites to see how many of the top 100 use valid markup, how many nearly validate (i.e. would validate if not for an error or two), and which DOCTYPEs predominated. Even with a fistful of porn sites in the mix, it was dull work: click a link, load the home page, run a validation bookmarklet, record the result.
I had no expectations. I made no assumptions. I just clicked and tested.
Such tests tell us little
I make no claims about what I found. If all the home pages of the top 100 sites were valid, it would not mean that the pages beneath the home page level were valid, nor would it prove that the sites were authored semantically. (An HTML 4.0 table layout with no semantics can validate; so can a site composed entirely of non-semantic divs with presentational labels.)
Validation is not the be-all of standards-based design; it merely indicates that the markup, whatever its semantic quality may be, complies with the requirements of a particular standard. Conversely, lack of validation does not prove lack of interest in web standards: ads and other third-party content can wreck a once-valid template, as can later third-party development work.
Moreover, nothing causal or predictive can be determined from these results. If 25% of the top 100 sites validated in my test, it would not mean that 25% of all sites on the web validate.
And I got nothing like 25%.
Enough disclaimers. On with the test.
see the link for the test: