麻豆传媒

In the wake of the US election, concerns are surfacing over the filter bubbles that mediate the information people see in their social media feeds.

Filter bubbles are formed by the algorithms social media sites like Facebook use to decide which information to show you, based largely on your own tastes. The idea is to keep you engaged, but the result may be a worldview skewed to fit your own preferences and biases. With 62 per cent of Americans at least occasionally, the fear is that filter bubbles could affect how you make decisions in real life.

鈥淚f this window is filled with highly partisan and, in some cases, false news, then many people will be assessing political candidates and information on the basis of distorted and misleading information,鈥� says at King鈥檚 College London.

So what can social media sites 鈥� and their users 鈥� do to burst these bubbles?

You can take control of your own feed. You don鈥檛 need to go to extremes, says at the Oxford Internet Institute 鈥� there鈥檚 no value to wading into the fever swamps of white nationalism or climate denialism, for example. But equally don鈥檛 unfriend people just because they didn鈥檛 vote like you.

Friend or foe?

鈥淵our context is shaped by who you friend and unfriend,鈥� he says. 鈥淓very time you tailor your feed to get rid of people whose opinions you don鈥檛 like, you add more boundaries to your reality.鈥�

People who support losing political parties, he says, are particularly vulnerable and their bubbles may shrink further if they retreat into like-minded safety.

How to break out? It isn鈥檛 easy. Network effects reward extremists and drown out moderate views, says at the Massachusetts Institute of Technology. The problem is not so much the extremists on the other side of the political spectrum, but those on your own. 鈥淚f you say something moderate, you might get chewed out,鈥� he says. 鈥淪ocial network filter bubbles create an environment in which moderates have incentives not to speak. There鈥檚 a lot of enforcing going on.鈥�

The filter bubble phenomenon is largely a design problem, says Hidalgo. Facebook and Google have already moved to on fake news sites, but this does not stop users sharing false information, and bubbles are made of more than just falsified news reports.

Hidalgo has other suggestions. Facebook, he says, ought to design a 鈥渇lip feed鈥� button. Its algorithms could identify your bias, then show you stories selected from the other end of the political spectrum. An alert could pop up when an algorithm detects that your feed is getting too closed off, and it could suggest people to friend or pages to follow with a view to widening your perspective.

Confirmation bias

Social networks might also introduce random, high-quality news stories into feeds, he says, or a Rotten Tomatoes-style reputation meter 鈥� an icon placed near the like button that could crowdsource people鈥檚 ratings on a story鈥檚 trustworthiness. Even if people manipulated it by scoring stories highly if they fit their confirmation bias, the button may serve as a reminder that not everything you read on the internet is true 鈥� something that鈥檚 easily overlooked in the rush to confirm your beliefs. 鈥淯nfortunately, we still have the problem of over-believing things,鈥� says Hidalgo.

Howard says part of the problem is that all sources are presented in the same way, whether you鈥檙e looking at The New York Times or your neighbour鈥檚 blog. He thinks Facebook should divide and label content into news and other posts. That way, opinions and false information are not misinterpreted as legitimate news.

鈥淚t鈥檚 time for us to treat Facebook as a media firm,鈥� says Howard. 鈥淭hey need to get a public editor and curate their content.鈥�

If social media sites are unwilling to do that, he suggests that regulation could hold them to similar standards as journalists and broadcasters.

But at MIT鈥檚 Center for Civic Media thinks we should avoid giving social networks even more authority over our reality. 鈥淒o we really want them to have even more control over our online existence?鈥� he says. 鈥淒o we actually want to give Facebook the power to decide what鈥檚 true and what鈥檚 false?鈥�

There鈥檚 another caveat. 鈥淲e are still so early in our understanding of how social networks really work,鈥� Matias says. We don鈥檛 know how much impact filter bubbles have on our views 鈥� in fact, 鈥渢he idea of the filter bubble has never been empirically proven鈥�.

And regardless of the effect of filter bubbles, Matias is sceptical about how useful 鈥渂ursting鈥� them would be. Filter bubbles don鈥檛 only exist in social media, after all, and how far outside your own worldview would you have to look in order to gain an unfiltered perspective 鈥� presuming such a thing is even possible?

In fact, maybe this new obsession with the filter bubble is itself viewed through a filter bubble. 鈥淚n 2008, we celebrated when Obama used Facebook in creative ways to win the election,鈥� says Howard.

听