During a conversation at work, @mulls wanted a quick way to see who are the most influential people tweeting about a specific topic. @chadauld and I came up with the simple metric of ranking users by their follower count. And this is the result.
Users we consider influential are colored bright yellow. Mid-range users are pink, gradiating down to users that no one cares about in white, and ordered by date — most recent on top.
For the math nerds in the house, we deem important users to be those where
followers > μ + σ / 2
and un-important users are where
followers < μ - σ / 4
That said, given more time, we should probably develop a better metric. Take into account their updates per day, or how many followers *their* followers have. Something like that.
Anyway, because Twitter's API ocasionally doesn't respond quickly enough, we load the follower counts on the client side by making callbacks to Yahoo!'s YQL service, which pulls the data for us and also serves as a proxy for when Twitter goes down.
This was written in a couple hours, so don't hate too much. (We know it doesn't currently work in Internet Explorer.) For this demo we've limited the results to two pages of data. Feel free to download and run it on your own box to get a complete set of results.
Update 4/18: We've added an experimental new feature which attempts to highlight (er...ignore) users that we detect are spammy. Our reason for doing this it to try and filter out users that are merely contributing to Twitter's noise. You'll see that these users are greyed out. And here's the formula we're using to detect them:
(followers > μ + σ / 4) && (following > followers * 0.75)
Do you find this hack useful? We'd love to get your feedback as we're thinking of incorporating this functionality directly into Sideline. Let us know! Feedback is welcome either at @tylerhall, @chadauld, or @ysideline.