I will not write the prefaces, I will go straight to the point. No one asked to check the validity of the html code of several terribly popular sites?
With the help of a useful menu item in Opera browser, the main pages of these sites were checked:
1. google.com (English version)
2. yandex.ru
3. rambler.ru
4. bing.com
5. mail.ru
6. vkontakte.ru
7. facebook.com (Russian version)
8. twitter.com
9. wikipedia.org
10. artlebedev.ru
That is, 4 well-known search engines, 5 popular sites and 1 more just for joke.
google.com
47 errorsWe can say that Google is keeping up with the times - only here and in Wikipedia HTML 5 is used. But the combination of HTML 5 and outdated tags and parameters (font, bgcolor, etc.) looks strange. Because of this, the validator and displays an error in half the cases. The second half is ampersands in javascript inserts like onblur = "google && google.fade && google.fade ()".
yandex.ru
44 errorsVirtually all errors due to closing tags. No, they are not absent, on the contrary - there are no opening tags. Well, the little things are still: there is no alternative text for the image, ampersands (& instead of & amp; in the links), there is no indication of the type of script. At the same time, there are only 4 errors on the minimalist ya.ru: the document type is not specified, the identifier is duplicated, the autocomplete attribute and the search field type are used (the validator defined the page as HTML 4.01 Transitional).
')
rambler.ru
57 errorsAlmost the same picture as Yandex: ampersands and closing tags in the absence of opening ones. The easy version of the Rambler r0.ru validation passes.
bing.com
4 errorsThe record for the number of errors among the four search engines. All four errors are invalid nested tags (a div inside a, p inside a span).
mail.ru
236 errorsAlso a kind of record, but now for the largest number of errors in this test. Among the errors, the most notable are the same ampersands, the absence of quotes when defining tag parameters and the lack of the type parameter for scripts. I don’t know whether the validator works this way, or because of the absence of this parameter, but he also considered the tags inside the lines in the script for errors.
vkontakte.ru
69 errorsThe biggest variety of errors: ampersands, quotes, tags in the wrong place, unclosed tags, missing required parameters, present unnecessary parameters, etc.
facebook.com
39 errorsAlmost the same picture as VKontakte, only supplemented with duplicate identifiers and novelties from HTML 5 (the autocomplete parameter; the page uses XHTML 1.0 Strict).
twitter.com
15 mistakesThere is an original error: spaces in the name parameter. Well and not in those places and not those meta tags and links to CSS.
wikipedia.org
0 errorsThe only valid main page is Wikipedia. Given that everything else is commercially available, and Wikipedia is an open source, there is something to think about ... In addition, HTML 5 is used.
artlebedev.ru
25 mistakesThe famous page of the famous designer with a mat in the title and the phrase "<NB> The best validator is a browser </ NB>" when trying to validate directly. There is no doctype on the page, there is a closing tag to the link (which nobody did before), no image captions, no script type or CSS specified, again ampersands ... In general, no rolls.