To be honest, part of it is the knack of finding the right keywords. Some ideas I use: - Hit up Google Books / Amazon for SIP (Statistically improbable phrase

To be honest, part of it is the knack of finding the right keywords. Some ideas I use: - Hit up Google Books / Amazon for SIP (Statistically improbable phrases) related to your topic. These are useful. - Add a bunch of +[Author name] keywords into scholar in order to get papers that cite certain people. Eventually you build up a standard list of authors that you want to find papers citing (yes, I know, this can be overly narrow, but it's effective). - As mentioned elsewhere, use Citeseer. - Its much easier to navigate a reference graph than to search cold. So find papers you think are excellent, and then serach everyone who cites them, all other work by referenced and citing authors, and so on. - Use arXiv. You can set up RSS feeds for certain keywords, and search them. This is low-SNR but yields some gems. - Most importantly: Don't search! Ask other people what's worth reading and what's good, in your group/lab/friends.