Data Journalism: Group 1.5

data journalism

  1. ‘Mark Hooker’: Our Fake Facebook Personality to Tackle Filter Bubbles
  2. Filter Bubbles on Facebook: Using FbTrex
  3. Blue Feed, Red Feed: Challenging Political Filter Bubbles

Written by: Amber Kouwen, Nanda Mohamed, Lucy Holaskova, Ivana Sramkova, Desi Slavova and Aidan Fahle


‘Mark Hooker’: Our Fake Facebook Personality to Tackle Filter Bubbles

Algorithms – the basis of all computing technology. Without them, Web 2.0 as we know it would cease to exist. They tell the computer how to execute a certain task(s) and then begin to accomplish the desired goal. However, even though algorithms are indeed a revolutionary concept, they have evolved into something more corrupt than emancipatory. One of the most famous examples of this targeted use of algorithms is Facebook. Facebook is designed in a way that it functions to display posts on a user’s news feed according to their likes and interests, thus triggering a positive reaction. For example, if I am liberal, the algorithm will not show me posts from conservative news sites. This creates ‘filter bubbles’ or ‘echo chambers,’ which could be classified as some sort of safe haven of information catered to individual groups of users.

Due to these problems, Mark Hooker was born; a dog-loving American man, who enjoys American football, going to church and technology. We created this fake persona in order to analyse the Facebook algorithm, focusing specifically on the other side of the political spectrum – therefore making him like things that are different from our own personal viewpoints. For that reason, we made him into a flat earth believer, anti-abortion advocate, and an overtly religious Trump supporter. We worked on his profile for more than one week, constantly sharing and liking new content which was in line with his beliefs and values. To do this, we used the tool FbTrex, which collected all public data from Mark’s profile, and stored them in its server. Thanks to this browser extension, we were able to keep track of all the sponsored posts Mark was receiving, and therefore, we could create statistics which served as an argument in favour of the presence of a ‘filter bubble.’

Due to these personalised information bubbles, a user’s ability to critically evaluate information is damaged, which is a major threat to the political sphere as people generally do not seek information once they have already obtained it. Therefore, if users receive their daily news from Facebook, which shows them information catered to their viewpoint, they will not go looking for it elsewhere, thus leading to a gap in information.

Posts that appeared on Mark Hooker's Facebook timeline, sort by subjects

Posts that appeared on Mark Hooker’s Facebook timeline, sort by subjects

Posts that appeared on Mark Hooker's Facebook timeline, sort by form

Posts that appeared on Mark Hooker’s Facebook timeline, sort by form

There was strong evidence concerning the existence of filter bubbles on Mark’s profile. Keeping in mind all the personality traits we assigned to him, his values and political opinion, we found out that the biggest number of posts present on his profile were about dogs. This included mostly pictures and some articles, coming up to 194 in total. Indeed, this goes in line with his interests, as he himself owns a St. Bernard rescue dog, showcasing strong love for man’s best friend. The second most prominent category was made up of pro-Trump posts, of which there were 150, with articles being the most frequent medium. Mark liked news pages supporting Trump and his policies such as Breitbart and Politico, so it only makes sense that these kinds of posts would rank high statistically. The third category was news about technology and inventions, mostly from the news source Wired. We conceived Mark to be a construction engineer, therefore it was only fitting that he would receive related posts on his news feed. The rest of the content was mostly concerned with Christianity, sports, and anti-abortion, thus showcasing a low variety overall. As predicted, Mark did not receive any liberal news or any other posts that contrasted to his own worldview.

Posts that appeared on Mark Hooker's Facebook timeline, sort by source

Posts that appeared on Mark Hooker’s Facebook timeline, sort by source

However, Facebook’s algorithm seems to be so smart to find a way around the fake persona Mark. From giving friend suggestions of the creators themselves to posts on Mark’s timeline that go in line with the creators’ interests, Facebook knows everything and can easily identify if a profile is indeed real or fake.

In conclusion, a lot of valuable information can be learned from this experiment. For example, we are now able to say with confidence that we are indeed constrained into our own filter bubbles. These filter bubbles have a strong outer layer, which can be compared to a cell membrane. This ‘information membrane’ doesn’t let anything in or out. In order to leave, we need to be able to actively diffuse ourselves through this membrane and deliberately look for information that would counter our own viewpoints. Only in this way are we able to falsify information, thus separating fake news from reality.

References n.d. ‘Facebook.Tracking.Exposed’. Accessed 16 March 2019.


Filter Bubbles on Facebook: Using FbTrex

Most people view their Facebook feeds as a window to the world, painting a bigger picture of what is happening around them. However, a “bigger” picture does not mean a full one. Many do not realize that they have a very limited set of information sources due to filter bubbles, implying that the content you receive is related only to the interests of you or your friends. Furthermore, echo chambers, which refer to the reinforcement and repetition of the same views and beliefs, further prevent the consideration of other ideas. It gets more problematic when we include the numerous advertisers who are trying to influence and change public perception.

In order to closely analyze this problem, our team decided to compare two completely different Facebook profiles using FbTrex. One of the profiles belongs to a real person and is regularly used, whereas the second one is a fake profile created specifically for this research. Both profiles are separated by different interests, beliefs and political views. With this research we want to outline the effects of filter bubbles, echo chambers and advertisers, in order to raise awareness about the biased realities created on Facebook feeds.

To combat personalization algorithms found on many social networking sites including Facebook, FbTrex was created to highlight the faults and problems associated with automated decision making on these social network sites. Moreover, it wants to explore the benefits of publicly owned datasets, collect forensically valid evidence to exert their rights, as well as educate people on algorithm influence. Furthermore, FbTrex wants to expose how Facebook affects people’s interpretation of information. Some examples include targeted political advertising, misinformation, echo chambers and algorithmic censorship. FbTrex enables citizens, authorities and communities to keep Facebook accountable for their decisions. This feature is best utilized by ordinary people, as those are the ones most affected due to their minimum freedom and power on the platform. FbTrex wants all users to have the ability to decide how Facebook shapes their informative experiences.

Example of Data from FbTrex

Example of fake account’s data from FbTrex

FbTrex works by having users share some data found on their Facebook account, not their personal information, but rather what Facebook generates for them. FbTrex first creates a copy of the user’s timeline with the user’s consent, and then reuses the data collected to perform analytics, comparisons and statistics. This may concern some people, as it requires sensitive information from all users. However, FbTrex has self-imposed limitations to their own data ethics such as only observing timelines and not individual profiles or pages. Furthermore, they only store public posts on their server and users who install the extension have full control over their data, since they can delete what they submit whenever they want. Lastly, no one has access to an individual’s data unless the owner grants them access. With all of this in mind, FbTrex hopes to raise criticism of Facebook’s current data exploitation model and to empower more people in this age of information.

Mark Hooker's Facebook Account

Our fake Facebook account: Mark Hooker

Our fake account was “Mark Hooker”, a 32-year-old male born in Massachusetts, USA. He is currently living in The Netherlands. Based on predetermined characteristics, Mark follows and engages with Facebook content connected to NFL football, Wired magazine, anti-abortion and religious publications, together with Trump-positive content.

On Thursday, March 7th, 2019, Mark’s public Facebook profile was created. In the course of one week, his online persona was strengthened daily by liking and sharing content based on his aforementioned likes and dislikes. On Thursday, the majority of news stories recorded on Mark’s timeline were from right-wing sources Politico and Breitbart. On Friday, Mark’s feed contained a very limited number of topics which revolved around abortion and Trump. On Saturday, the majority of posts on Mark’s feed were connected to anti-abortion and dog-related videos. In order to switch the Facebook algorithms a little bit, Mark started liking, sharing and commenting on more diverse content, such as political memes and Trump-supportive articles. Moreover, on Sunday, which is church day for Mark, what was noticeable on his news feed were the variety of suggested religious (catholic) groups. This was impressive, because it proved that even a four-day account has already provided enough data for Facebook’s algorithms to create a news feed without any articles that would even in the slightest oppose Mark’s views. In order to switch the account a little bit again, on Sunday Mark focused on tech-related articles as well. Thus, on Monday, Wired’s articles were prioritized on Mark’s feed instead of anti-abortion and political content. Additionally, Mark started receiving targeted ads on Monday. However, Facebook probably managed to track the VPN because the ads were more connected to the likes of the students that were using Mark’s account on Monday instead what Mark himself likes.

Mark Hooker's timeline, including FbTrex Tool

Mark Hooker’s timeline, including FbTrex tool

In conclusion, by using FbTrex we were able to recognise the filter bubbles each Facebook account lives in. Our two accounts had two different personas, and Facebook very precisely presents posts in their feeds according to their distinct interests.

References n.d. ‘Facebook.Tracking.Exposed’. Accessed 12 March 2019.


Blue Feed, Red Feed: Challenging Political Filter Bubbles

Many US citizens were shocked when Trump was elected as the 45th President of the US in 2016. They were led astray by their own Facebook news feeds, seeing only negative posts about Trump and positive posts about Hillary. These Facebook users were living in their own politically divided filter bubbles, leaving no room for the acknowledgement of information that opposes their personal beliefs. This not only causes the American nation to be even more politically polarised, but is also harmful for our democracy. Therefore, the Wall Street Journal created the application Blue Feed, Red Feed, which juxtaposes both liberal and conservative posts, thus challenging the damaging political filter bubble.

Blue Feed, Red Feed was developed in 2016 by Jon Keegan, showing blue (very liberal) and red (very conservative) feeds side-by-side. By selecting different topics, users can discover two distinct political views on the same subject. Whether the sources are classified as liberal or conservative is based on a 2015 study by Facebook scientists Eytan Bakshy, Solomon Messing and Lada Adamic published in the journal Science.

Blue Feed, Red Feed timeline

Example of a Blue Feed, Red Feed timeline

The ways in which the scientists classified the articles into liberal or conservative is fairly straightforward. First, a data set of 226,310 “hard news” stories was created by classifying news stories and opinion pieces from 81 of the most shared news sites on Facebook. Certain English keywords were used as indicators of what should be classified as “hard news.” Secondly, each piece of news in this data set was given an alignment score, which is based on the average of the self-described political affiliations of people who shared that particular news piece. For example, if the proportion of conservative Facebook users that shared a particular news piece was greater than 50%, then the news piece would be considered conservative in this framework. Furthermore, based on this alignment score, each piece was further classified into one of the 5 categories: very liberal, liberal, neutral, conservative and very conservative. To appear in the actual blue/red feed itself, articles needed at least 100 shares and come from sources with no less than 100,000 followers.

What they found out by collecting, filtering and analyzing the content shared by 10.1 million Facebook users is that depending on their political preferences people receive drastically different information regarding the same topics. Facebook users are separated into distinct filter bubbles depending on whether they have conservative or liberal leanings. This is why Blue Feed, Red Feed is so significant. It provides a relevant, easy to use tool which displays in real time how opposing news sources depict the same topics such as guns, abortion, immigration, etc. With this tool, people are able to escape filter bubbles and gain a better understanding of opposing viewpoints, which otherwise would have remained invisible.

One potential, though controversial interpretation of Blue Feed Red Feed concerns the manipulation of people through Facebook, which has a strong and undeniable power to create filter bubbles. One of the main points of the Facebook business model emphasises the creation of a personalised online “habitat” in which one can calmly dwell, seeing only posts from like-minded friends. However, this proves to be problematic. For example, when taking voting into account, Facebook aids by connecting individual behaviour to voter files, consequently targeting ads to those voters. It is clear that most people today get their news mainly from Facebook, therefore they cannot ever be free from this bias, never really reaching any form of an “objective” outlook on the political spectre. Moreover, if this concern is looked at from the perspective of Facebook as a business, it only amplifies the fact that Facebook may be using our data to manipulate us. Their ultimate goal to make their users stay on the website as long as possible and help foster “safe” environment in which people are not triggered by the posts of their extremist friends.

In conclusion, Blue Feed, Red Feed, illustrates the disparity in political news on Facebook, by providing a side-by-side look at the variety of liberal and conservative perspectives of similar topics. Even though the Internet is rapidly changing, and the data is constantly transforming, Blue Feed, Red Feed posts are still being updated every hour. Moreover, each source used for the project was first carefully examined, chosen out of millions, and classified in one of five categories. Overall, Blue Feed, Red Feed bursts Facebook’s filter bubble by providing a unique lens to see how current news stories are being manipulated on Facebook to support a specific political agenda.

Bakshy, E., S. Messing, and L. A. Adamic. 2015. ‘Exposure to Ideologically Diverse News and Opinion on Facebook’. Science 348 (6239): 1130–32.
Bakshy, Eytan, Solomon Messing, and Lada Adamic. 2015a. ‘Replication Data for: Exposure to Ideologically Diverse News and Opinion on Facebook’.
Bakshy, Eytan, Solomon Messing, and Lada A. Adamic. 2015b. ‘Exposure to Ideologically Diverse News and Opinion on Facebook’. Science 348 (6239): 1130–32.
Bilton, Ricardo. 2016. ‘The Wall Street Journal’s New Tool Gives a Side-by-Side Look at the Facebook Political News Filter Bubble’. Nieman Lab (blog). 18 May 2016.
Graff, Ryan. 2016. ‘How WSJ Used Data and Design to Show Americans Their Polarized Politics and Media’. Northwestern University Knight Lab. 21 June 2016.
Keegan, Jon. 2016. ‘Blue Feed, Red Feed’. WSJ. 18 May 2016.