"Foo over DNS" was innovative and amusing a decade ago. While a clever hack, it has no advantages other than the fact that the latency drops from 200 ms to 20 ms or whatever. I don't know any applications consuming wikipedia blurbs for which that's an issue.<p>As for parsing Wikipedia and extracting summaries, that's already been done by Freebase. They also let you search, so you don't even need the exact article title. I'm a happy user of their wikipedia blurb feature and a lot of the other data that their API offers. Freebase can be useful for enhancing practically any web app; go check it out if you aren't familiar with it.