CGI scripts do allow dynamic generation of documents, but are a crippled low-level way to do that, and do NOT allow browsing of meta-document. Any tactics that the user would want to use as browsing a meta-document must be foreseen by the CGI script writer so the reader can use them; and if by chance, the script is made expressive enough to allow the same power as full meta-browsing, this would be in a new ad-hoc undocumented unreliable programming language with a slow remote interface that the reader doesn't have the time to learn, so can only use trivial tactics.
The Web is but yet another kind of coarse-grained client/server crap, where the reader is on a terminal (that displays a layout text document from HTML input, instead of a raw character array from VT100 codes) and is denied any computation power, while only the server-writer can put anything to the document. To achieve meta-browsing through the WWW, we oughto rewrite a full reflective OS on the server side, that must cope with a stupid horrible unexpressive HTML language as its only interface. Yuck.
Also, using URLs is like having to work with pointers, not objects, in an environment with an unreliable buggy moving GC. It just can't work, so that a reliable meta-browsing server on the WWW cannot include meta-browsable external references.
To sum it up, the WWW is NOT a meta-browsing system, just a "hyper-browsing" system. "Hyper-browsing" only allows browsing in multilinear ways, which is not better (though often faster) than a book with a good index. The rare cases of interaction in the WWW are very limited, and the reader is completely subject to the publisher as for this interaction.
On the opposite side, Tunes is a "Meta-browsing" system, that allows arbitrary extraction of information from the document view as a semantically meaningful object instead of a meaningless syntactical stream of bits.
About hypertext publishing systems and more, see the corresponding Wiki page.