杏吧原创

Is Facebook more than just an online ‘echo chamber’?

Research by the social network suggests its site encourages the spread of diverse viewpoints, despite accusations that it does just the opposite
A majority of influence in Facebook, represented in orange, can be generated by a person's many weak ties - their outer circle of Facebook friends. That's because, while a person's strong ties are individually more influential, there are fewer of them
A majority of influence in Facebook, represented in orange, can be generated by a person鈥檚 many weak ties 鈥 their outer circle of Facebook friends. That鈥檚 because, while a person鈥檚 strong ties are individually more influential, there are fewer of them
(Image: Facebook)

Facebook is more than just an online echo chamber in which users just repeat views that match their own, according to from the social-media giant鈥檚 own data team.

鈥淭he research suggests that鈥 online social networks actually increase the spread of novel information and diverse viewpoints,鈥 says study leader , discussing the findings .

Clearly, Facebook has a vested interest in distancing itself from the charge that it is that some commentators contend is poisoning political debate. But while the new study does question common assumptions about the flow of information through social networks, it seems unlikely to lay this accusation to rest.

The analysis comes from an experiment that ran over seven weeks in 2010, while Bakshy was working towards his PhD at the University of Michigan in Ann Arbor. It involved more than 253 million users of Facebook and almost 76 million web addresses (URLs) shared by these people on the site, giving more than a billion unique user-URL pairs.

Friendly ties

One common problem with studies of social media is that it is difficult to tell whether people share material because they have seen their friends share it, or whether they just happen to be highlighting the same content because they tend to look at the same websites. To control for this, Bakshy and his collaborators at Facebook randomly assigned the user-URL pairs to 鈥渇eed鈥 or 鈥渘o feed鈥 conditions. For the latter, any content which shared the URL in question was blocked from that user鈥檚 Facebook 鈥渘ews feed鈥 for the duration of the experiment.

The researchers also divided the users鈥 friends into 鈥渟trong鈥 or 鈥渨eak鈥 ties, depending on measures including the number of times they sent each other private messages, or how often they were both tagged in the same photo.

Not surprisingly, whichever method the researchers used, users were much more likely to share URLs flagged up in individual posts by friends with whom they had stronger ties.

However, Facebook users have many more weak ties than strong ones. Cumulatively, the researchers found, most of the information shared on Facebook comes predominately through a user鈥檚 weak ties. That鈥檚 important, argues Bakshy, because friends with weaker ties are more likely to read and share material that you would not otherwise encounter: 鈥淭he information they are sharing is more novel.鈥

Selective view

Maybe so, but without knowing more about the content of what was shared, it is hard to dismiss the charge that Facebook acts as an echo chamber in which people propagate material that chimes with their own views. 鈥淚 would think it鈥檚 more likely that you would selectively share the things from your weak ties that coincide with your view of the world,鈥 says of Purdue University in West Lafayette, Indiana, who works on data mining of social media.

While the new study certainly isn鈥檛 the last word on Facebook鈥檚 status as an online echo chamber, it highlights a dilemma facing the site鈥檚 developers as they strive to develop algorithms to improve users鈥 experiences. Filtering material to highlight content posted by close friends should increase the average interest of each user鈥檚 personal news feed, but it would also block the flow of novel views and information coming from weaker ties.

鈥淚t鈥檚 definitely a balancing act,鈥 says Bakshy, who joined Facebook鈥檚 data team in June 2011. 鈥淭hat鈥檚 why we have data scientists doing research on the impact of how our algorithms work.鈥