And that's why we can't have nice things, er, why we will never have valid HTML on a significant percentage of websites: browsers are historically very lenient with HTML errors (because otherwise they wouldn't be able to show 90% of all sites), and no one uses HTML validators to check if their HTML actually conforms to the spec. It's a chicken and egg problem really: the browsers can't be more strict because there are so many broken sites, and the sites won't be fixed because the browsers aren't strict enough.
I did a quick check, and most of the errors it reports are "unknown attribute" or "element such-and-such not allowed here". Those "errors" would be allowed anyway for forward compatibility, and aren't really a big deal.
IMO the validator's definition of "invalid HTML" is just too strict; it should only count parse errors and completely non-sensible things. And the specification is also too strict at times; on my own website I have "Element style not allowed as child of element div in this context." This is because on some pages it adds a few rules that apply only to that page and this is easiest with Jekyll. I suppose I could hack around things to "properly" insert it in the head, but this works for all browsers and has for decades and why shouldn't it, so why bother?
If the specification doesn't match reality, then maybe the specification should change...
