odd time signatures

Tracking the Swarm

Back on the “track” bandwagon, folks, and this time in response to a series of discussions on Friendfeed and Identi.ca/Twitter about why it matters, and why tracking anything less than the full stream doesn’t work.

Brian Roy, who is working on a filtering tool for Friendfeed, wrote a post this morning claiming that “everything is myth“. My response: unicorns are mythical; track requires the full stream to be effective.

Start by looking at this. Note that the USGS official time for the China earthquake was 06:27:59.0. The first tweet came from scribeoflight, who I was not following (anyone in parenthesis is a track result and not from someone I follow) at the time at 10:35PM. Because GTalk doesn’t record seconds, it could have been as early as 10:35:01, or 6 minutes 2 seconds after the earthquake hit, reached scribeoflight, scribeoflightmessaged and it reached my IM. (note: time corrected — see comments.)

Every message that hit GTalk was sent by someone I didn’t follow. In the full transcript of that day, you can see that over time, I chose to follow those who were sending regular updates merely by sending twitter the command to follow them.

Another example:

Also in May, I tracked the term “tornado”, since they were hammering the Midwest. On May 22nd, messages started coming across about tornado warnings in Colorado, and messaged Amy Gahran, who I know lives in Boulder, to see if everything was all right, after receiving tracked messages indicating a tornado touchdown in Ft. Collins. Amy is a reporter, and while she was fine, she certainly had an interest in finding out what had happened and who was affected, since reports were also flooding in that a school had been damaged and one person killed. She wrote about our exchange here. What I found particularly compelling about her story was how difficult it was for her to get information in real time:

I went straight to the National Weather Service site — which is, I must say, a usability nightmare for someone seeking information about local severe weather in a hurry! I tried posting to Twitter links to specific NWS storm warnings, but apparently those pages are generated with session-specific URLs and so are useless if you send them as links to other Web users. (For future reference, it’s probably best to post links to NWS pages for state-specific current watches, warnings, and advisories — like this one for Colorado, so they can find the most up-to-date info for their part of the state.)

Next I jumped to the Ft. Collins Coloradoan site, where I found this report on the storm. That’s where I learned that the tornado actually touched down in Windsor, a town several miles southeast of Fort Collins (and much closer to the town of Greeley). Karoli sent me a links to video and photos of the large tornado, shot by a 9News team. I watched the video while looking out my window and noticing the sky above was growing darker. (UPDATE: Here’s more 9news video from during and after the tornado )

Here is the value of track, stated in 140 characters or less:

The ability to discover information, communicate, and act upon it across clouds in real-time whether mobile or stationary.

The notion of “limited track” is self-contradictory when seen in those terms. Track is not about ‘everything’, nor is it limited, nor does it guarantee “everything”. But what it does do is identify and follow the swarm around topics, people, events, or yes, disasters.

Speaking of disasters, there is a swarm around the word “Sylmar” right now. Looks like another wildfire has started, this time in Sylmar Hills. in the last five minutes, 35 messages have swarmed around that one term. It’s a bad fire, lots of wind, evacuations, and concern for the residents in that area. There’s a swarm around it, discovered through tracking the full stream, not my own Twitter stream. The value happens in the main stream, not my limited sphere of influence.

Photo Credit: Max Westby

Watch this space for the manifesto outlined on today’s NewsGang Live. Because Twitter needs to restore the ability to track, really track, in real time. Like we could back in May before they turned it off.

Twitter Track: Last seen 175 days, 11 hours, 22 minutes ago.

Comments are closed.