group 3.2





It has been one week since we created a fake Facebook account with the made up persona Karen Susan Johnson, who is a single mom, fangirl of David Cameron, but above all a flat earther. We have been actively using this fake account and we have followed and joined some flat earth groups, befriended some fellow flat earthers as well as reposted and liked other posts that cohered to Karen’s interests. For this research, our goal was to see what kind of information and data was gathered, and furthermore we also wanted to see where most of the posts originated from, if they had more of a political character or not, and also how many men and women Karen had interacted with. With the help of the FBtrex tool we have gathered some interesting information, which will be elaborated on in this blog post.

Information gathered with FBtrex

In the data that we have gathered we could see how many post, videos and more was tracked with the FBtrex tool. This information we have implemented into a pie chart to give you a better overview about the number.

Graph 1: chart on what/how many information was tracked on Facebook from march 7 – march 13

In this chart there can be clearly seen that there is a lot of information gathered of regular posts of Karen’s Facebook. 58% of the information were from post, 31% from photos, 7% from videos, 1% from events, and 3% from groups. From this chart we could conclude that Karen had more interactions with regular Facebook post than with post from groups or events.

Graph 2: pages where most post were from

In the graph displayed above can be seen which pages Karen has interacted with the most. The pages that she had the most interaction with and which she liked and reposted are Hugues Bison, BlancNoir, Ligue des droits de l’Homme, Sea shepherd Globe, The Sun, Fakir, and Gael Ranger. Most of them are very popular sources.


After extracting the csv-file of information that fbTREX had gathered during the week Karen had been an active facebook account there was an immediate conclusion that, even though it was only a week, there has been a lot of data gathered in that time period.

Picture 1: part of the excel-file with the information gathered about Karen

When converting the cvs-file to a excel spreadsheet, the informations spanned over 1000 rows of cells which surprised us a little that this much information had been gathered over a course of a week. Also, the number of men/women Karen had interacted with from 7 march-13 march were:


Men 21

Women 11


As we have mentioned in a earlier blog post, our main focus point with Karen is the flat earth conspiracy. This because of the rising flat earth movement, where according to YouGov poll one in three millennials in the US believe the earth to be flat (Foster & Branch 2018). Over the time since the account started we have joined several flat earth groups, for example “The Flat Earth Revolution”, and written bold statements through Karen’s account proclaiming her opinion about the earth’s possible lack of roundness.

An interesting project, according to us, would be to combine the data gathering method of Blue Feed/Red Feed posted by The Wall Street Journal but make its focus on gathering information on pro vs. con flat-earth conspiracy. Because of the uprising in believers of the flat-earth movement, which started on Youtube in 2015 (Weill 2018), we believe there would be enough data to be gathered from both sides to give a clear overview much like the Blue Feed/Red Feed gives right now over the conservative vs. liberal opinions.


However, all the information that was gathered with the Fbtrex tool was not very clear and did not really told us a lot about filter bubbles. It was very obvious for us in the beginning when we made this fake page that the post on Karen’s page would be different than ours. We did not really needed the tool to tell us this, but it was helpful to see with which sources Karen had interacted with.



Foster, Craig A. Branch, Glenn. 21.08.2018. Do People Really Think Earth Might Be Flat?

Weill, Kelly. 17.11.2018. Inside the Flat Earth Conference, Where the World’s Oldest Conspiracy Theory Is Hot Again.




For this weeks project we had to create a fake Facebook account, nurture that person and look into a specific current topic. To do this we had to create the account and then use the software fbTREX to track and see the filter bubbles that might occur and also how much of this we would see on our own accounts.


Karen with little Steve Jr.


For our fake Facebook profile, meet Karen Susan Johnson. She is a 36 year old single mother of two, and pregnant with the third child, from Blackpool in the UK. Karen is very much an entirely different person from how any of us in our group are. Karen has various interests including Minion memes, David Cameron, posts about ‘mom life’, the Sun articles and the Flat Earth society. Karen has a conservative political leaning, even though she does not exactly knows what that entails as the main reason she categorizes herself as being conservative is because her husband was conservative as well. Karen is very much invested in the Flat Earth society and uses Facebook to explore this interest. Therefore, our main focal point of research is the current flat-earth conspiracy that has been on the uprising for the past years starting with Youtube videos published in 2015 and onward (Weill 2018). The flat-earth conspiracy is one of the world’s oldest conspiracy-theories and according to the pollster YouGov, on in three millennials (people born early 1980’s to mid-1990’s) is the US believe the world to be flat (Foster & Branch 2018).


“I am a flat earther”


Karen is subscribed to various Facebook-groups about flat-earth and also shares posts about how the earth being a globe is a lie. The profile picture used for Karen is a stock photo with a made up photographers name in watermark to make it look more authentic.

Various groups Karen likes displaying her interests


The tool used to track Facebook, both our own accounts and our fake account is called fbTREAX. The project behind the software started in the summer of 2016. It started out as a hacker tool which later became a researcher tool that has been validated by academic achievements. In the future, fbTREX wants to become an empowering tool that will lead a cultural shift. The project of fbTREX is to look deeper into the personalization of filter bubbles and algorithms. Since the software is open-source anyone can use it and contribute to the data collection as well as see how it works and help make it better. To be able to look at the algorithms and the personalized filter bubbled of facebook fbTREX needs to collect a lot of data. According to their website, they are aware that some people might find the issue of privacy in this disturbing but they have, according to us, taken very good precautions to minimize the data collected while still achieving the goal. The tools for protecting user integrity are opt-in, public posts and opt-out. Opt-in means that users have to manually install the web browser addon for data to be collected, public posts means that fbTREX only look at posts that are posted as public on the users feed and therefore never collects data from private groups, opt-out allows the user to manually opt out of showing the posts to fbTREX.

fbTREX logo


Even though we do believe this research to be a very good and interesting thing, we do find some problemizations with it, mainly from a ethical point of view. Since we created our fake profile to be very much unlike what we are, due to filter bubbles we actually saw very little of the character in our own feeds. This indicates that the filter bubbles do exist but it took some of the edge away from the project for us. We think that if we would not have made a equally extreme person, or perhaps had the person living in the Netherlands we would have seen more of her.

The ethical stand points are our biggest concern with the project, we all felt very insecure about having our data tracked through Facebook. Another big issue for us was the creation of the fake account since Facebook nowadays wants a phone number when creating a new profile as a security aspect against fake profiles. We all felt a bit contempt to provide our own phone number.

In conclusion, we found the project to be very interesting in the sense that it do indicate that filter bubbles exist and that due to the Facebook algorithms we didn’t see much of our fake accounts posts and updates due to her being so very different in her interests from us. We also find the project fbTREX to be a good way to gather this information about the Facebook algorithms and how they affect every user. However, we had some, mainly ethical, issues since we all do value our privacy and by submitting our data, and by creating the fake account, we had to give data such as a phone number that we didn’t feel very contempt with. It would be interesting for us to take this experiment a step further and see how far we could go with Karen’s extreme persona before we would be called out for catfishing or simply have the account deleted by Facebook.


Foster, Craig A. Branch, Glenn. 21.08.2018. Do People Really Think Earth Might Be Flat?


Weill, Kelly. 17.11.2018. Inside the Flat Earth Conference, Where the World’s Oldest Conspiracy Theory Is Hot Again.





On May 18, 2016, at 8:00 a.m. Eastern Time, American newspaper The Wall Street Journal published Blue Feed, Red Feed (BFRF). The concept shows two feeds, red being “very conservatively aligned” and blue being “very liberal” (Keegan 2016). The Wall Street Journal created these feeds to show how reality can contradict Facebook’s enthusiasm about sharing and connecting globally (Keegan 2016). The feed speaks to the growing relevance of “echo chambers”, where Facebook feeds becomes insulated mediums for like-minded groups of people. The topics covered by BFRF data gathering are president Trump, health care, guns, abortion, ISIS, budget, executive order and immigration Each category leads to a side-by-side glimpse of the feeds of liberal and conservative Facebook users (Keegan 2016).

One of the most obvious and prominent aspects of BFRF is its interactive and continuous nature. Unlike most ‘traditional’ articles consisting solely of text written at one point in time, BFRF is constantly updated in real-time. Another non-traditional note is the interactive aspect that is introduced by allowing readers to choose the different topics they want to see in the opposing feeds. Each displayed post comes with a date tag, indicating when it was originally published on Facebook (Keegan 2016). The feeds are updated regularly: on most topics, the most recent posts are merely a few hours old. In fact, the article itself states “updated hourly”, right next to the date it was originally published: May 19, 2016. Seeing as this was nearly three years ago, BFRF contains a considerable amount of data (Keegan 2016). It is possible for readers to look at older posts by loading more posts.


The creators of BFRF uses information collected by Facebook, extracting information needed for the feed through tracking content shared by users (Keegan 2016). Facebook calculates the sites that appear on the feed by examining the “self-described political leanings” of the people that have shared the links, and then calculate a “political alignment score”; an indication of position along the political spectrum (Keegan 2016). By analyzing these users’ political labels with the help of machine learning (Graff 2016), Facebook uses these scores to categorize five demographics: liberal, very liberal, neutral, conservative and very conservative (Keegan 2016). To appear on this feed the post must have a minimum of 100 shares and come from sources that have at least 100,000 followers (Keegan 2016). To get this information together, the Journal uses several data-mining programs. Some of the programs they used are Nod.js to decode the JavaScript of the Facebook Graph Application Programming Interface, before saving it to a MySQL database, a method of managing data (Graff 2016).


The structure of the feed is unlike traditional journalistic structures and serves a more rhetorical conclusion on filter bubbles than a concrete one – one that users can interpret themselves. The creators of the concept have given no indication of putting an end to the feed, and so the data that is gathered may continue to serve the same purpose. The data which is produced through BFRF is in a state of continuous change, but it is not fickle: it serves a concrete reminder to all users of social media but is responsive to ongoing political issues.

Besides this, sometimes it gets obvious that BFRF is currently more curated by machines rather than humans. For example, when checking the feeds for the topic ‘guns’, the blue feed displays a post about popular rock band Guns N’ Roses (Keegan 2016). While including the word ‘guns’, as the article’s algorithm undoubtedly recognized, it does not cover what the creators likely intended it to. Other articles within the feed talks about actual guns, and issues surrounding their legislation in the US. On the other hand, it might make a point about the nature of the blue feed being less inflamed than the red feed on this topic, and that for some Democrats, music can sometimes be more relevant than firearms.

Within the context of data in BFRF, we can summarize it as a web application utilizing the online data collection of Facebook posts across the political spectrum. The gathering and selection processes of these liberal-leaning and conservative-leaning Facebook posts encompasses the use of various tools, mostly based on numbers. As the application is a ‘live’ one, meaning it is updated hourly with additions of new Facebook posts, it is not clear whether BFRF will end any time soon, and thus the amounts of data gathered will continue to proliferate. In a more interpretive sense, one cannot help but wonder if the message behind BFRF will always stay the same, whether we will always need this data peep show to tell us that the grass is not always greener on the other side.



Graff, Ryan. 2016. “How WSJ Used Data and Design to Show Americans Their Polarized Politics and Media”. Northwestern University Knight Lab. 2016.

Keegan, Jon. 2016. “Blue Feed, Red Feed”. WSJ. 2016.