Get as nit-picky on the domain as you like; sometimes, it's still not enough to determine which "site" you're going to. Think of /~user/ directories, for example, or hosted subsites. Not everything that's distinct from its owner has its own domain.
The ideal split would actually require something like:
<link rel="Site Root" href="http://example.com.nyud.net:4561/~bob/thebigsite/oldv1/">
as metadata on the resulting page. (As an aside, does HN currently retrieve the resulting pages for any metadata when submitting? Doing an AJAX autofill of the
default submission title from the <title> tag might be a good idea.)
For most sites, "Site Root" would just be the same as the domain name. Perhaps, then, it could have that as the default assumption of any parsers, the actual tag just overriding it for the sites that need it.
Also, you know what already (mostly) provides the same information? The <base> tag! Oddly, something we've long dropped from our HTML vocabulary.