I would tend to agree that the STL can be quite useful for games programming. I’ve just writen an implementation of the A* algorithm using the STL priority_queue and map templates. It outperforms Matt Bucklands versions using a variety of different heuristics and IMHO is much easier to understand than the one from his book. It also adds nodes on the fly so unlike Matts it does not require you to pre-generate the graph. Its so good I’m thinking of writing a paper on it :D. Id be happy to share it if anyone is interested in seeing it.
think if you look you should find some papers published in relation to pathfinding and STL.