If you wanted to run a survey on the way that web pages are put together – number of img elements, amount that rel=”nofollow” is used etc. – how would you do it? Ask your webby chums? Write a program to read through a few sites automatically? Or use Google web authoring statistics? Probably the latter, I would guess.
It’s on the brain-numbing side of nerdy, but there are some fascinating snippets of information in there. There are also some frightening things – particularly the number of Microsoft Office psuedo-HTML elements. It takes me back to my days wrestling with Frontpage *shudder*.