Just noticed an O'Reilly Radar post by Peter Brantley, in which he interviews Google's Ramanathan V. Guha about Google Custom Search. Brantley thinks this is a powerful tool that holds a lot of promise, if certain issues can be addressed (the predictable privacy, of course). I had heard about this tool awhile back, and had thought about playing with it a bit, but it seemed way too big a project for me at the time -- it's sort of, at its base, like creating a big list of all the sites that you think are the best on a particular subject, no small task -- but now I'm rethinking my dismissing it so quickly.
I looked at the example Brantley included of the Cornell Law Library's Legal Research Search Engine, and was pretty impressed. It seems to get at the complaint that some in library studies have that there needs to be some curation of Web content for instructional purposes (ie, using Wikipedia for everything and only Wikipedia, I mean, is just not ok), and it also would make a very good assignment for grad students to create such a Custom Search Engine in a subject area. The idea of how you would go about casting your net, winnowing out what you get, thinking very carefully about the decision to bring something in, put something out, all of that would make for a very interesting experience and one the discussion of which would be interesting in its own right. You could put a regular Google Search box right next to your Custom Search Box and people could see for themselves what you had done in pointing to things you think are of the highest quality. We do this with Blogs already (our blogrolls are recommendations) and the selection of other sites we link to. We do it with "resource" pages like mine for Open Access Resources for Authors. In fact, I think it would make a great addition to any resource page to add a "search these resources" custom search box. There are people who like to browse a set of sites; there are people who know what they are looking for and want to search. You can speak to both. Cool.