Monday, 6 May 2013

Why your friends on Twitter are (probably) more interesting than you, and what to do about it

Statistical logic means that your lover has probably had more sexual partners than you. Similarly, at the gym, most of the other users train more frequently than you. And your friends have more friends than you do - this last observation was labelled the "friendship paradox" by sociologist Scott Feld. It's a fact because popular people get counted in more people's tallies of how many friends their friends have (here's more explanation).

Now thanks to a new paper by Nathan Hodas and his colleagues, we can add to this humbling stats lesson the fact that for most users of Twitter, our followers and followees (the people we choose to follow), are better connected, more active, and more interesting than we are.

Hodas' team analysed Twitter data from the second half of 2009 featuring 476 million tweets and 5.8 million users, with 193.3 million links between them. First off, they found that for most of us, the people we choose to follow are better connected - that is, they typically follow ten times as many people as we do. Our followers too are better connected, typically by a factor of twenty.

Seeing as we choose how many other people we want to follow, this first observation isn't such a blow to the ego. However, the researchers found a similar result for popularity. That is, the people who follow us are typically ten times more popular than we are (in terms of their own follower count). Less surprising, the people we choose to follow are also more popular than we are. Here there are two distinct groupings - in one, the people we follow are typically ten times as popular as us; in the other, they are typically 10,000 times as popular (this is thanks to celebrity accounts and such like).

Not only are our followers and the people we follow better connected and more popular than we are, the people we follow are also usually more active. Eighty-eight per cent of users were found to be less active than a typical person they followed; this rose to 99 per cent when omitting accounts that ceased activity during the period covered by the data. This is probably because we're more likely to follow accounts that are more visible by virtue of being more active.

A related observation was the link between activity and popularity - that is, more active users tended to be more popular, a correlation the researchers described as "especially strong". This suggests being more active on Twitter could be a simple way to gain more popularity, although we need to be cautious because there's no proof here for a causal link. "The detailed mechanism for this explanation is not yet clear," the researchers admitted.

To connectivity, popularity and activity, we can add interestingness. The researchers looked at the "virality" of links shared on Twitter - literally how many times they were re-tweeted. Here they found that 79 per cent of Twitter users posted less viral content than the people they follow.

Another issue the researchers looked at was what they called "information overload". Here they found that as the number of people we follow increases, the information that we're subjected to increases "super-linearly" - each new user that we follow typically equals hundreds of new items of information in our Twitter stream. In part this is because, as we heard, the people we follow are usually highly active (or at least more active than we are). The risk is that we end up subscribing to more information than we can possibly manage to consume.

This last point about overload is relevant to readers hoping to boost their popularity and interestingness on Twitter.  Comparing overloaded Twitter users (the third receiving the most amounts of info), and the underloaded (the bottom third), Hodas and his colleagues found that the overloaded tended to receive information that had gone massively viral, but tended to overlook "mini-cascades" - information that had viral potential. "It appears that overloaded users are only good detectors for information of mid-range interestingness," the researchers said. "Most likely the information that their friends already know." This suggests that if you want to be the kind of user who helps to break the next big story on Twitter, you need to be careful not to follow too many accounts in the pursuit of this aim. Pick and choose who you follow with care.

"If you have ever felt like your friends are more interesting or more active than you are," the researchers concluded, "it seems the statistics confirm this to be true for the vast majority of us."

_________________________________ ResearchBlogging.org

Nathan O. Hodas, Farshad Kooti, and Kristina Lerman (2013). Friendship Paradox Redux: Your Friends Are More Interesting Than You. arXiv arXiv: 1304.3480v1

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.

No comments:

Post a Comment

Note: only a member of this blog may post a comment.

Google+