Flickr's developer blog has a pretty interesting post about how they built a fast client-side search that worked equally well for data sets of 10 people and 10000 people. Ajax with XML & JSON was too slow and using dynamic script tags was too insecure so they rolled their own format and used split to quickly parse it.
...is a weblog about the liberal arts 2.0 edited by Jason Kottke since March 1998 (archives). You can read about me and kottke.org here. If you've got questions, concerns, or interesting links, send them along.