Twitter’s new geolocation support was supposed to launch for developers at today’s Twitter Conference in LA (which I’m attending), but it wasn’t quite ready yet. Still, Twitter’s platform lead Ryan Sarver announced several details about how it will work, at least initially, in a developer session. In quickly-jotted bullet points:
- Twitter will soon be able to store location data–that is, latitude and longitude coordinates–on a per-tweet basis, and for your user profile.
- Including location information in your tweets will be opt-in only. You will have to visit your Twitter account’s settings page on the web site to allow Twitter to store that data. It will not be enabled by default. Even if your Twitter client sends lat/log points along with your status update, if you didn’t explicitly opt into including that information, Twitter will drop it at the point of entry and it will not be stored or published.
- Users won’t see any new features on the Twitter web site when geo launches except for the settings page where you opt in. Twitter is giving API developers a head start to display and transmit geo data in tweets in their apps first.
- In practice, expect to see your Twitter client include a checkbox below the posting area labeled something like “include my location with this tweet.” If you check the box when you send a tweet but you haven’t given Twitter permission to store your location data, you’ll have to visit your settings page on the web site to do so.
- Interesting: Twitter will scrub geo-data stored in tweets more than 14 days old to avoid subpoenas about a user’s location. They will outright delete the location information from their database, not just anonymize it.
- While Twitter usually encourages application developers to cache data, in the case of geo, they recommend apps don’t keep historical location data so that developers don’t become a subpoena target, either. They also recommend “fuzzing” location and time data, so that instead of knowing that Joe Smith was at 8th avenue and 15th street at 2:11PM Eastern time on March 7, 2008, you only show that Joe was in Brooklyn on that day.
- The geodata-scrubbing isn’t a permanent solution. Twitter is looking into ways to store this data in a “safe” way in the future, so Twitter won’t always scrub +14-day-old data, just at first.
- Besides just using the free-form text field in the location field already available in your profile, there will be no way to tell Twitter you’re in a broad area, say, a city or neighborhood like San Diego. They will only take and store lat/long coordinates. On the front end, they may only display a broad area name, like a city or a neighborhood instead of a specific point, but they will store the specific lat/long coordinates.
- Right now, Twitter does support some light geolocation functionality based on the “anything goes” location field in profiles. Try a search for happy hour near:11215 -RT to see tweets (minus retweets) about happy hour in Park Slope, Brooklyn.
- Two interesting location-based Twitter apps available now: Happn.in, and Trendsmap.
This is pretty exciting stuff, if still nascent. Questions? I’ll answer what I can in the comments.