I just added a new recipe on IFTTT that sends me a message whenever someone sends a tweet within a specified radius. They bill it as a neighborhood tweet watch but it could obviously be used to watch for tweets within any area or place. Right now it used a simple point-and-radius geofence but presumably that could be extended into any sort of bounding box.
I mentioned that I’d added this and got a number of responses including a comment from Arjun Ram that suggested that IFTTT’s geo strategy is better than Twitter’s. It raises the question of whether Twitter does in fact have a geo strategy and why they’ve been so slow in doing anything in that area. After an early acquisition of the geo-team at Mixer Labs in 2009, they’ve not really done much. Not only has Twitter been slow to encourage location tagging of tweets that could benefit (something like 2-3 % of all tweets have a location tag), they’ve not even deployed simple tools to allow location filtering of those tweets that are geotagged. Now IFTTT has done it.
- Monitor tweets about a sporting event or concert only from people at the venue.
- Monitor tweets along a route to detect traffic incidents. With a little creative UX design, you could make a driver friendly crowd-sourced traffic and road condition similar to Waze.
- Local businesses could monitor for people who were close by and sent out a hashtag about what they wanted (#lunchdeals) and respond with an offer.
- News, public safety or marketing companies could monitor public places for increases in twitter activity that might give early warning of a spontaneous event or unusual occurrence happening there. This would take some serious data munging to establish baselines activity from which anomalies could be detected.
Those may not all be good ideas but they’re just a few I though of in 10 minutes. The point is that it seems like the Twitter Firehose has a vast amount of information some of which is usefully tied to location. I’m surprised that Twitter has not done more to make something useful out of it.