It doesn't matter if you're Justin Bieber or James Baker. On Twitter, the first few minutes tell the whole story.
In fact, MIT Sloan Professor Tauhid Zaman has figured out a way to predict how many retweets a Twitter message will receive, just based on what happens in the first few minutes of its life.
"If a Tweet goes up, I can actually forecast ahead of time how many people are going to retweet it," he says.
He can do this because pretty much all Twitter messages have the same lifecycle, and that can be mapped on a graph. Zaman wants to build a website that will analyze Tweets in real-time and make predictions about their popularity, but in the meantime, he's set up a website that graphs the number of retweets that followed messages from people like Will.i.am, Newt Gingrich, and Kim Kardashian. He calls it the Twouija, a nod to the old Ouija board game that purports to predict your future.
The predictions are far from perfect. For example, Zaman took a look at Kim Kardashian's April 15, 2012 message: "Happy Sunday tweeps! Have a blessed day!" After watching the retweets for five minutes, he predicted it would get 555 retweets within a few hours. In fact, it got 709. That's off by 30 percent, though it's not bad.
But you can see each of the graphs following the same pattern: there are a lot of retweets during the first few minutes, and then this gradually tapers off in a somewhat predictable way.
"It doesn't matter if you're a bigshot or a nobody," Zaman says. "If you're going to get anything you're going to get it in the first couple of hours, and this is the way you get it. It's going to evolve like the kind of curve you see in Twouija."
In the long run, Zaman thinks his models could be used by Twitter to build more interesting timelines. "The stuff in your timeline is just ranked by the age of the Tweets," he says. "And a lot of time, that isn't he best way to look at your timeline. What you want to see in your timeline are tweets that are new and young, and they're going to be huge."
