The age of the filter bubble – a small world in a large network

Everyone agrees with me, no-one is arguing against my beliefs, every news article supports my stance – is this what reality looks like? For an in­creas­ing number of internet users, it does. Society rarely agrees and varying opinions are the pre­requis­ites for a demo­crat­ic debate culture. However, since the result of the US Pres­id­en­tial Elections in 2016 we’ve realised that the image that social media gives us isn’t complete and some opinions are even omitted. Whether we like it or not, we live in a filter bubble.

When Donald Trump won the election and became the 45th President of the United States in 2016, many people around the globe were shocked: there was nothing to indicate that this political outsider would be backed by the majority of people. At least that’s how Trump’s opponents saw it. For his sup­port­ers on the other hand, it was clear that they were all against Clinton. Both sides had made excellent arguments on the net and expressed their own opinions, but the other side hadn’t heard them. All of them had only read and commented within their filter bubble.

Our society has shifted a large part of everyday life to the internet. For many people, com­mu­nic­a­tion and acquiring in­form­a­tion happens ex­clus­ively online. Facebook’s newsfeed acts like a news magazine, Google as a lexicon, and mes­sen­gers like WhatsApp or Skype serve as forums for sharing in­form­a­tion with friends, col­leagues, and family. We now find almost everything we want to know online. Internet providers know that too: Google, Facebook, Netflix, and Instagram know how important their part is in society. They are therefore con­stantly refining their al­gorithms when it comes to user-friend­li­ness: they only show us the in­form­a­tion that is sup­posedly relevant to us.

This is nothing new: the popular service providers on the internet collect data about user behaviour on their platforms and promise to adapt the user ex­per­i­ence even better to the needs of users – often without them having to do anything. In the past, data col­lec­tion has been cri­ti­cised by many experts, but primarily under the (very important) aspect of data pro­tec­tion. The term 'data leech' describes how com­pre­hens­ive Google, Facebook, etc. collect and analyse users’ personal data: how much time does someone spend online? Where do they live? What are their hobbies?

Of course, all this in­form­a­tion is also used by these companies for their own purposes: Google and Facebook earn a large part of their revenue from per­son­al­ised ad­vert­ising, for example. But the in­form­a­tion should also help to better tailor offers to the re­spect­ive user. This means not only is the ad­vert­ising per­son­al­ised, but also the offers.

As a result, these services only show us the news, in­form­a­tion, and opinions that match our user profile. This may seem a positive thing at first: feeds are no longer stuffed with articles that don’t interest you in any way, popular posts are no longer cluttered with comments that you don’t want to read, you don’t have to wade through arguments that don’t lead anywhere, etc. But in the long term, this creates problems that only come to light when you question the filtering mech­an­isms of social media.

The concept of the filter bubble goes back to the activist Eli Pariser, who in his book, The Filter Bubble: What the Internet is Hiding from You, cri­ti­cises the extent to which in­form­a­tion is per­son­al­ised on the internet. He notices that different users – depending on their political attitudes, for example – get different results even after using the same search terms. However, the Google bubble is not an isolated case: other services on the web also use al­gorithms for per­son­al­isa­tion. Bubbles like these also occur on Facebook. The problems that result from this, are not only in­di­vidu­al in nature, but also have an impact on society as a whole.

Discourse is important in a func­tion­ing democracy: ex­chan­ging different per­spect­ives is not only important between politi­cians from different parties, but should take place through­out the whole of society. Only then it is possible to see things from different per­spect­ives and to broaden your own horizons. However, those who live in a filter bubble will rarely find arguments that go against their own views, but are more likely to find a lot of support. Since many internet users do not yet have a suf­fi­ciently critical awareness of how to deal with the new media (known as 'media com­pet­ence'), their own per­cep­tion within the bubble is projected onto the entire world outside.

Instead of seeing one’s own opinion as just one of many, the filter bubble makes it seem like there is only this one opinion and that no others exist. This explains phenomena such as Trump’s sur­pris­ing victory. Within the filter bubble of liberal in­di­vidu­als, there were no signs that enough people would share the Re­pub­lic­an candidate’s thoughts. Journ­al­ists, who are also in a filter bubble like this, act as mul­ti­pli­ers and spread this pre­con­ceived opinion in other media.

The formation of filter bubbles con­tra­dicts two basic ideas that correlate with the spread of the internet as a mass medium: On the one hand, it stands for the net­work­ing of the most diverse people across the globe, but anyone living within a ho­mo­gen­eous group will no longer benefit from this advantage. Secondly, the internet has been praised as a virtual place where in­form­a­tion is freely ac­cess­ible and cannot be censored. This enabled the internet to form an an­ti­theses to tra­di­tion­al media, as content is filtered by the re­spect­ive editorial offices. Now this filtering can also be found on the internet, but instead of an editorial team doing the filtering, it is now an algorithm that selects what users should know.

It is quite easy to presume it’s the fault of large cor­por­a­tions and their al­gorithms: Facebook, Yahoo!, and Google do not, or only in­suf­fi­ciently, educate their users about how and why they filter certain in­form­a­tion, and don’t give them the op­por­tun­ity to change or disable the filtering. In general, however, all users are jointly re­spons­ible for the content they receive. Facebook, for example, shows less news from users whose links we don’t click on. This means that we are already signaling a dis­in­terest in reports that do not agree with our opinion. The algorithm continues to do this and presents only the in­form­a­tion that seems to be of interest to us.

Eli Pariser assumes that con­tra­dic­tion is rife in everyone and then compares this with healthy and unhealthy food: we know that we should eat food that is good for us, but we are happy to choose products that will satisfy us at that exact moment. Pariser argues that a mixture of the two should be the solution: in­form­a­tion that matches our profile, but also in­form­a­tion that chal­lenges us. Al­gorithms should also be struc­tured in this way.

There’s no party that can take full re­spons­ib­il­ity for filter bubbles: it seems to be a mix of social and technical phenomena. On the one hand, every person tends to seek con­firm­a­tion of their opinion. On the other hand, technical de­vel­op­ments used in the web are designed to make browsing as pleasant as possible – not to create in­tel­lec­tu­al chal­lenges. The fact is that one single person can’t possibly see all messages that appear on the internet every day. Therefore, there is basically nothing wrong with a filter based on technical al­gorithms, but the resulting excess must be examined crit­ic­ally.

Two other terms often appear in con­nec­tion with the filter bubble, which are 'echo chambers' and 'fake news'. An echo chamber is a room in which you can create a strong echo. In the fig­ur­at­ive sense, this refers to a virtual space where opinions only intensify and there are no longer any mit­ig­at­ing in­flu­ences. Echo chambers like these are created within a filter bubble because a fed-in opinion (e.g. in the form of a Facebook post) is only amplified by the echo of the other members within the bubble and is no longer put into per­spect­ive by a different point of view.

Among other things, this explains the success of so-called fake news. The alleged factual reports either depict distorted facts or are even totally made up. Agitators feed these fictional stories into a filter bubble where they can spread these alleged facts without being chal­lenged. This creates a per­cep­tion of the world that is de­term­ined more by opinions than by facts, which leads to conflicts instead of dis­cus­sions.

There are some voices that criticise the theory of the filter bubble when it comes to diversity of opinions. It is ques­tion­able how high the influence of a filter bubble really is and whether the internet or the cor­res­pond­ing al­gorithms amplify it. According to research by Oxford Uni­ver­sity, TV was the most popular source of news in 2013 (80% of people got their news from this source), but by 2016, this figure had dropped to 70% and had been overtaken by online sources (72%). The research also showed that 84% of 18-24 year olds get their news online, whereas it’s only 21% of those aged 55+. This age group prefers TV (54%) whereas only 9% of ad­oles­cents use TV to keep up-to-date with world events. Since so many people are using the internet for in­form­a­tion, the Google bubble could therefore have a sig­ni­fic­ant influence on what in­form­a­tion users receive. In addition, journ­al­ists also have social media accounts and use Google to research so in this respect, this also in­flu­ences the media beyond the internet. But there are very con­tra­dict­ory opinions when it comes to the Google bubble: Eli Pariser provides clear evidence of how per­son­al­isa­tion in­flu­ences Google’s search results. However, these ob­ser­va­tions date back to 2011 and Google continues to make regular changes to its search engine. However, you shouldn’t forget that filter bubbles existed long before the internet: before the de­vel­op­ment of the world wide web, numerous people in or­gan­isa­tions, within circles of friends, and at get-togethers argued their opinions and tried to get others on their side without needing Google or Facebook bubbles. The internet has enabled different view­points to be heard and therefore more plur­al­isa­tion. But to have a free internet in the future, where diverse people with differing points of view and opinions are on the same level, you shouldn’t un­der­es­tim­ate the danger of filter bubbles.

If you want to free yourself from your filter bubble, you have several options: the first step should be to question your own surfing behaviour. If you’re looking for con­flict­ing views, you will find them despite the Facebook bubble and Google’s (supposed) per­son­al­isa­tion. This is how social media al­gorithms can be con­sciously in­flu­enced and trained: for example, if you 'like' the pages of several political parties, you should receive a wider range of in­form­a­tion from the political spectrum in the future. This way, everyone can create their own diversity. In addition, the network offers tools that at least free the search of per­son­al­ised results. Every internet user is free to search using search engines other than Google. The German search engine, Unbubble, for example, reveals that no in­form­a­tion about its users’ behaviour is collected or evaluated. This means that no per­son­al­ised search takes place and therefore no filter bubble can form. In addition, there are add-ons for some internet browsers, which help to prevent surfer behaviour from being tracked. What’s good for privacy also helps to prevent filter bubbles. If companies are unable to collect personal data, they can’t per­son­al­ise results or ad­vert­ise­ments. You should also be careful when revealing personal in­form­a­tion on social media. If you don’t want to give up Facebook, you can at least be more re­strict­ive when it comes to the data you enter. Assuming that the filter bubble is also expanding into tra­di­tion­al media, it makes sense to use as many different kinds of media and sources as possible in order to obtain in­form­a­tion and news. For example, this can be done online by using balanced news ag­greg­at­ors such as Feedly or News360. These enable you to perceive other per­spect­ives and broaden your horizons, despite the menacing filter bubble.