I've no idea how good it actually was, but it felt revolutionary at the time because the only native computer interfaces we'd ever seen up to then were:
Plus it was fast...
I'm a librarian and you are very right that these are important things to consider.
However, most librarians that I know are committed to getting the information to the people who need it (in academic libraries, our faculty and students). You will find that libraries, including Harvard, are already investing in systems to make stuff available and easily searchable from a wide variety of sources: including all those in-house ad-hoc repositories you mention, open access sites like arXiv and pay-for material (where we can afford it).
What worries many librarians I know, is whether someone like Google can do it better, or whether we (with local knowledge of our academics) can add enough value to make it worthwhile to keep us. I think the jury is still out on that one, but it won't be for lack of effort on supporting our users
I've seen the MST3K and the original version of both these and while there is some plot cut, in no way does it come close to saving these turkeys. They are both bad films and MST3K makes for a fun evening.
I can see where you're coming from. I've just tried my first conversation with Cleverbot, and while any judge would have be off their head to vote it human, it is at least entertainingly weird and I can see why some humans I know would get scored higher...
Me: How are you?
Cb: Fine. How are you?
Me: Not too bad. bit of a cough.
Cb: Do you hear voices in your head?
That's wonderfully surreal...
According to that article there are approx 10^123 positions in the game-tree. Given that there are about 10^78 atoms in the visible universe, building enough memory to store all the outcomes of a chess game is tricky...