Imagine that, after being slapped around a bit by Guy Steele, the W3C announces that, effective immediately, the official format for HTML is no longer

-HTML-
    -HEAD-
        -TITLE-Bla-/TITLE-
    -/HEAD-
    -BODY-
         -IMG src="foo.bar"-
    -/BODY-
-/HTML-
-/CODE-

(think <> instead of --), but instead

<CODE>
(in-html (cons (in-head (in-title "Bla")) (cons (in-body (img :source 
+"foo.bar")) '()))
[download]

. Imagine also that, at the very same time, both AOL/Netscape and Microsoft have signed treaties promising that they'll follow the W3C specs by the letter. Now, if you've got a shitload of code floating around which extensively uses literal HTML (be it quoted, with here-docs, or using the __DATA__ facility), you're going to have a lot of editing to do... (and just try to cook up an HTML-to-SEXP translator script!). Whereas if you're using the CGI module's facilities, all you have to do is change a few functions here and there, and you're all set - and your CGI programs will be the only ones that work with the newest releases of Nutscrape and Exploder!