There's been some talk about "Personal Learning Environments" over the past few days. Scott McLeod (Dangerously Irrelevant) draws our attention to Ray Sims' mapping of his PLE, reminding me of a workshop I recently led for our grad students. My talk was far less ambitious, but if I were to draw a complete map, it would likely not look radically different from Ray's.
Greg Linden (Geeking with Greg) also writes about Gord Hotchkiss' essay on personalized search, the idea that search engines could be made more effective by "guessing intent" based on question context, search histories, and demographics. The idea is to "disambiguate" general language, leading users to "the answer they want," even if they are not aware of what that is exactly.
While no one wants a search engine with poor precision or recall rates, it's a very sticky question (and highly personal, as well) to anticipate user search satisfaction. Personally, I'd rather have some sense of the underlying logic of search tools, and develop instincts about what "works best," based on my own (somewhat random) interests. Serendipity is inevitable, and many users (myself included) believe it to be an important part of the discovery process.
Addendum: I suppose I shouldn't try to post while chewing gum. Thanks to Scott and Stephen for the corrections; my apologies for the errors.