Is Twitter a Two-Way Street?: To get a lot of followers on Twitter, do you need to follow a lot of other Tweeps? And if not, why not?
Freakonomics has more than 270,000 followers on Twitter. And we love every one of you. No, really! But we don’t follow anyone back. (When we first signed up with Twitter, some very smart media consultants said that reciprocity was the name of the game; but we didn’t want to get involved in some massive online tit-for-tat.) So, does that make us
That’s one of the questions we ask in our latest podcast, “Is Twitter a Two-Way Street?” (You can download/subscribe at iTunes, get the RSS feed, listen live via the link in box at right, or read the transcript here.) We try to find out just how social you need to be on a social-media site; if you want a lot of followers on Twitter, do you need to follow a lot of other Tweeps?
You’ll hear from the network-theory whiz Duncan Watts, a former Columbia sociologist who now works at Yahoo! Research. He’s the author of the very good book Six Degrees, as well as the brand-new Everything is Obvious (Once You Know the Answer). And he co-authored a recent paper called “Who Says What to Whom on Twitter.” From the podcast:
WATTS: It’s true that, you know, there are millions of users on Twitter who are listening to other users. But we also find that there is a remarkable concentration of attention. So about 50 percent of all tweets that a random person on Twitter receives on any given day come from just 20,000 users. So that’s about .05 percent of…so one half of a tenth of a percent of all users on Twitter.
WATTS: It’s worth emphasizing again that Twitter is not a social network. Now, social networks are characterized by very, very high levels of reciprocity. So if I say that I’m friends with you, it’s very likely that you will also say that you’re friends with me. It’s not always true, but it’s very often the case. And if not, I stop being a participant in that social network. It’s a funny kind of friendship if only one person thinks that it exists. Whereas, in communication networks it’s totally different.
You’ll also hear from Justin Halpern, who parleyed his Twitter feed “Sh*t My Dad Says” into a best-selling book and a TV show. Halpern has millions of followers but follows only one other person. (I don’t want to ruin the surprise by naming the person here.)
And you’ll also hear from Steve Levitt, who talks about a fake Steve Levitt Twitter feed and tries to help me come up with a Freakonomics following strategy. For now, we’ve latched onto one person.
Anyone who cares about Twitter should spend some time on Twitaholic, a site that scrapes follower/followed data from Twitter. You’ll see that the top Tweeters — Lady Gaga, Justin Bieber, Britney Spears, and Barack Obama — have a relatively high follower-to-follow rate. Lady Gaga has nearly 8.7 million followers, and follows more than 144,000 users; Obama has nearly 6.9 million follows, and follows more than 700,000 users. (Presumably someone other than the President himself does the following.)
All of which seems to argue that reciprocity matters a great deal. But once you look below the very top of the list, the ratio plunges. So: how important is reciprocity?
We asked Duncan Watts to look into the follower/followed numbers and see what trends he could identify. He came back with some results that he warns are “quick-and-dirty,” but worthwhile nonetheless:
Here are two sets of results (1) for the top 1000 most-followed users; and (2) the 1000 users who follow the most others (i.e. with the most “friends” in Twitter parlance). [These numbers were taken from Twitter a few weeks ago, so are slightly outdated]
In both cases, the x-axis is # friends (i.e., how many they follow) and the y-axis is the # followers. The results are binned, so the labels are a bit deceiving, but you get the drift. The boxes represent the interquartile range; the horizontal lines in the boxed represent the median; and the bars represent the 95% confidence interval.
1000 Most-Followed Users:
Bottom line is that there isn’t any trend: for the vast majority of most-followed users, there is no relationship between following more people and being followed by more people. The only exception is the far-right bin, which actually contains just a handful of users, listed here:
And for these users, it’s not clear what is going on. BarackObama may follow lots of people, but it’s unlikely that that’s why he has lots of followers. Same for Britney or Yoko Ono. Guy Kawasaki and StartupPro are more interesting: they have almost exactly as many followers as friends. My guess is that in these cases, the causality is the opposite of what you were told — that is, they automatically follow anyone who follows them, so it is followers leading to following, not the other way around.
1000 Users Who Follow the Most:
Here we do see a clear positive trend: on average, the more users they follow, the more followers they have. But again, it’s not clear what the explanation is. Some of them are probably spammers, and in those instances, it probably is the case that they are getting followers by following tons of people — as there wouldn’t be any other reason for people to know about them. But other instances may be a result of well-known people auto-following people who follow them. That would also generate a positive trend, but the causality would be the opposite. Without looking at the data over time, it would be impossible to differentiate these two explanations, and even then it would be hard.
Thanks especially to Duncan for all his input on this episode. Feel free to tweet it.