I thought it would be interesting to incorporate earthquake information into a real-time world tweetmap I have been playing with. Would you be able to see nuances in tweets after small earthquakes? Nuances caused by the small earthquakes that the tweeters aren't even aware of, but are visible when viewed across a large population? Maybe tease out interesting patterns with a kind of Fourier analysis-inspired sentiment analysis. I find this kind of concept interesting - there are events going on we don't consciously notice, but affect out behavior in subtle ways.
Getting real-time earthquake data is easy - the USGS has an api with a jsonp endpoint, so you can get earthquake data updated every minute. And earthquakes are happening all the time throughout the day, so there would always be some kind of data.
I started down the road of incorporating real-time earthquake info into the tweetmap. I got to the point where there were going to be some tricky user interface things to work out, and I noticed something: most of the little earthquakes were along the California coast or in Alaska. Yes, there were some scattered around the world, but they weren't as close to populated areas as often as I'd like. It was going to be a good amount of work to work this in.
And I decided that the analysis of earthquake+twitter information is something best suited with historical data, not a real-time thing you watch for.
And so I am abandoning this side-effort for now. I think the overarching idea is interesting, just that this is not a particularly useful way to approach it.
In hindsight, any analyses of twitter+earthquake data are best done with historical data, not realtime. A map would help there - replaying tweets relative to when the quakes occurred, for example - but the real interesting work would be in extracting any of the "hidden" impacts the small shaking of our planet, which seems best suited to a different framework for analysis than what I have available here.