If you do run your robot again I would be very interested if you could
generate a simple list of document titles and their corresponding
document id's (or URL's). We have a powerful spires database here,
interfaced to the web, which we could easily import such a file into to
great a VERONICA like index of the web. I think that would be pretty
useful (unless someone is already doing it??).
One other problem to add to you list.....many documents are probably
only accessible by giving a "keyword" . Unless you can write a robot
which can successfully guess all possible keywords, you cannot
gaurantee to be able to traverse the whole web.