I decided to see if the popular pages use correct HTML (XHTML etc) code. My validation tool was W3C validator.
I checked portals only. Don't have a source showing which pages are most popular, so I had to guess. More sites - possibly later. Also, portal sites are generated dynamicaly, so validation results change.
I started from google. Simple page design, should be correct. And? Suprise. No type definition and other erors. No doctype produces 0 errors. 51 errors with default settings.
Then, Yahoo. More compilcated page, more changes for errors. And more errors - 253.
Next page, MSN, identifies itself as XHTML strict. It doesn't validate, however. 22 errors, mostly not serious and easy to fix. So why not fixed?
BBC page code looks strange. It has a big number of empty lines, a number of empty tags and so on. Number of errors - moderate. Only 65.
CNN has page with more errors (106). Many of them are about missing picture descriptions. It's easier to read than the previous one, however.
So it looks that it's hard to find correct HTML on portal pages. Most of the errors are not serious and don't require much work, usually just changing one or two characters will fix it. It looks that no-one really cares...