The problem there is that each page is essentially "generated in bulk". The system is given a file for the top half of the page (above the comic) and another file for the bottom half of the page (below the comic. It then generates each individual page, with the appropriate comic image, and updates the navigation links on each one.
It's easier and quicker than it looks- Curt wrote the script on a cocktail napkin 15 years ago, and it's only needed to be updated due to changes in server configurations in all that time.
But the upshot is, basically each page is, and can only be, essentially identical. We set it up so that each page has a spot for a unique "newspost" below the comic (where I usually put notes or holiday greetings and such) but that's visible to the reader.
It wouldn't be a terrible thing to replace all of those old posts with a dialogue transcript or keyword dump, but I'd rather not do it that way if we can help it. And even then, that assumes that Google or whoever would index it properly so that the comics can then actually be properly searched.
It might be possible to add the keywords to a sort of alt-text for each image, but I'd have to talk with Curt about that. And even if we could, without breaking the current comic setup, again, how do we know the search engines will even index it properly?