group 2.5

final research report (group 2.5)

Fake Person Data

For this research project, we created a fake persona two weeks ago by the name of Amy Stanson, a 38-year-old woman who works as a zookeeper at Artis Zoo. Amy is a Pro-Brexiter and lives together with 2 snakes. She is originally from Carnforth, went to university in Lancaster and now lives in Amstelveen. In this research blog post, we are going to further elaborate on our previous blog post since the account. In the first blog post, we tracked Amy’s timeline for a couple of days and compared this to our own timeline using the Facebook Tracking Exposed tool. The results of this research were that Amy’s Facebook was overly biased because it had a too narrow focus on certain topics due to the lack of Facebook friends. Furthermore, Amy got friend recommendations to people that are in our year of study, while she has no link with these people based on her profile whatsoever. We did not really have an answer to this.

We noticed that we were missing a lot of posts on Amy’s timeline and that English news sources seem to appear more often than other sources. Amy’s timeline is also filled with pro-Christian posts and posts that portray Islam in a bad daylight.

There were not a lot of ads on Amy’s timeline compared to our personal Facebook timeline. The lack of ads could be caused by a small number of friends. This posed a problem since we then do not know how Facebook views us. We should be able to get ads based on the demographic information we entered but the ads seem to be lacking.

We decided to unfollow some of the reptile fan pages and to start liking only the posts from news outlets and snake memes. Furthermore, we added Amy on all of our personal Facebook accounts to see if this would change her feed and if it would lead to different advertisements on her feed. Unfortunately, our attempt to break Amy’s filter bubble did not work and there were no significant changes to her feed or advertisements.

Before  unfollowing snake pages and only liking political posts and snake memes.

After unfollowing snake pages and only liking political posts and snake memes

In the end, it was harder to break out of a filter bubble than we originally expected. One of the reasons could be that we only created the profile two weeks ago. Our personal Facebook accounts have been around for more than 5 years so there is much more depth in our profiles. If a person were to use Facebook as their only source of news, this could be problematic since coming across opinions on your timeline outside of your filter bubble is highly unlikely. However, Facebook shows what they think their user would like and providing audi alteram partem is not one of the duties of Facebook.

The percentages of the form of the posts from Amy’s timeline.

The percentages of the content of the posts from Amy’s timeline.

The number of posts filtered by the top 5 sources of Amy’s timeline.


We produced these statistics:

Sources: (amount of posts / percentage; from high to low) – Total Number of posts: 1023

PVV: 158 / 15.44%

Daily Mail: 114 / 11.14%

Brexit News: 96 / 9.4%

The Telegraph: 89 / 8.7%

The Sun: 75 /7 .33%

Breitbart: 67 / 6.55%

Leave Means Leave: 62 / 6%

Fox News: 61 / 5.9%

De Telegraaf: 57 / 5.57%

Daily Telegraph: 51 / 5%

UK Independence Party: 37 / 3.6%

9GAG: 30 / 3%

Infowars oz: 29 / 2.8%

Leave.EU: 27 / 2.6%

Nigel Farage: 15 / 1.4%

Forum voor Democratie: 11 / 1%

The Prodigy: 10 / 1%

VT: 4 / 0.4%

Epoch Times Nederlands: 4 / o.4%

LADBible: 3 / 0.3%

The Legend: 2 / 0.2%

Wonderful Earth: 2 / 0.2%

Amy Stanson: 2 / 0.2%

UNILAD: 1 / 0.1%




Research Fake Persona

For this research, we created a fake persona by the name of Amy Stanson, a 38-year-old woman who works as a zookeeper at Artis Zoo. Amy is a Pro-Brexiter and lives together with 2 snakes. She is originally from Carnforth, went to university in Lancaster and now lives in Amstelveen. In this research, we are going to compare Amy’s timeline to our own timeline for a couple of days. The tool we will be using is Facebook Tracking Exposed. It tracks all posts in your Facebook timeline which are public. This phenomenon is semi-problematic since people who have their personal posts set on public may not know that the tool will collect these posts too.


Day 1 – Alec

On the first day the posts on Amy’s and Alec’s Facebook feeds could not have been more different. Alec’s personal feed mainly consisted of videos showing animals doing nothing and looking amazingly cute, memes, especially socialist memes and finally stuff that is offered in the region of Amsterdam through Facebook Marketplace. Whereas the most interesting posts in Amy’s feed were about a 1 billion pound Brexit bribe for “left-behind” towns, a teacher who forced his student to wet himself, so he soaked his pants and an article about a transgender man who gave birth to a child and the reactions of the haters. The Daily Mail is informing about white supremacist propaganda whose effort raised by 182 percent, while it is not clear if they are happy about it or not.

Day 2 – Julian

The second day Facebook was pretty quiet for Amy and Julian. The most significant news from major networks Julian follows was the unfortunate deadly crash of Ethiopian flight with 156 passengers on board. American news outlets are mostly covering the CNN’s post about Manaford being sentenced to 47 months while Mueller’s recommendation was 19 to 25 years. Liberal Europe seems to have taken a break from it’s Brexit mess as BBC informs about the biggest causes of death across the world instead. The picture of a demonstrating woman holding a board  “Always a slut for equal rights” seems to have taken over Julian’s feed. Amy’s feed still appears to be a little empty or pages she has liked seem to lack content. It is just a lot of nonsense. The Pro-Brexiters seem occupied by some kind of Cold war spy games, exposing the Liberal subversive elements within the group and kicking them out and asking other members to report “suspicious” behavior. The most extreme news claim that the EU is trying to control the United Kingdom borders even after Brexit and that Britain refuses Barnier’s bid . The tragically crashed plane is not mentioned on Amy’s feed or Julian’s feed.

Day 3 – Ewout

On the third day Ewout’s personal feed consists of a mix of his hobbies, interests and some political messages. Political messages from parties like the D66 party about their new leader and the direction they are now taking. Besides the political messages, quite some news outlets that report on Formula 1, Football and Techno music seem to reach Ewout. The sources from these reports are either pages/groups he has already liked in the past or new ones that offer their sponsored content to him.

Amy has a rather monotone newsfeed at this point. Her main news sources are English newspapers and one Dutch newspaper. The feed is dominated by British, American and mainly Dutch alt-right and populist political news. This ranges from the Brexit groups that Amy has joined to the PVV, to FvD and Breitbart. In an attempt to break the rather strange monotone newsfeed Ewout has allowed Amy to like various new pages and to join some new groups. These groups are related to her love for reptiles, England and Brexit. Besides that, she has liked pages that reinforce her political beliefs like Donald Trump, FvD and In an attempt to lighten up the newsfeed Amy has also liked 9gag to receive a daily dose of humor. Ewout has shared content on Amy’s timeline that reinforces her believes and also shared a picture with a personal text that showcases her reptile passion.

Day 4 – Oriol

Oriol’s personal feed on the fourth day mainly consisted of memes from 9gag and other meme pages, as well as pictures and information from concerts and festivals, since that is the only content he follows on Facebook. I got some news regarding the political situation in Catalonia, but those were few and quite scattered.

Amy’s newsfeed continues to be flooded by pro-Brexit news from sources such as, however, some news with a different tone regarding Brexit are scattered here and there, mainly shared in pro-Brexit groups by people disagreeing with the subject. Amy’s feed, however, is now dominated by reptile lover pagers and pictures of snakes. Since these pages seem to have a bit more of activity during the day, and they are liked and shared more often, we see a lot more reptile posts than political posts at this stage. Some funny content now is also present on her feed, coming from 9gag, after Ewout liked the page the day before. It breaks the monotony of politics and snakes that is Amy’s newsfeed.


It was not easy to search for appropriate content that can nurture the account, however, we tried searching through all the sites that we would normally never set a foot on. There is a strange feeling about sharing and liking the content of sites we normally do not want to go viral. We came across a strange fact during our experiment: Amy got friend recommendations to people that are in our year of study, while she has no link with these people based on her profile whatsoever. We were wondering how this could happen.

It was difficult collecting information from Amy’s profile since we could not befriend real people due to ethical reasons. Our opinion was that we could not be interfering with people’s personal lives just for our gain as journalists. The profiles were also overly biased from the start and therefore has a too narrow focus on certain topics while a normal profile is far broader.


Blue Feed/Red Feed, how does it work?

A fairly popular project surrounding the American elections in 2016 was the Blue Feed/Red Feed project by Jon Keegan of the Wall Street Journal. The project allowed anyone interested to see extremely liberal and extremely conservative views on debatable topics juxtaposed next to each other. The journalists used data to create and filter the feed. In this blog post, we aim to find out who is behind the project and to closely examine their data and methodology.

Who worked on the project?

The statistics used in the Blue Feed Red Feed were created by Facebook scientists Eytan Bakshy, Solomon Messing and Lada Adamic. Eytan Bakshy is a senior scientist on the Facebook Core Data Science and was the leader behind this study. Adamic contributed as a network scientist and Messing is a staff research scientist for Facebook. These numbers were compiled under the study “Replication Data for Exposure to Ideologically Diverse News and Opinion on Facebook”, and aimed to look at how a set of users from the platform Facebook reacted to news appearing in their feeds. The figures from this case study are being promoted by the Wall Street Journal, and their ongoing experiment, Blue Feed Red Feed.

What does the data tell us?

The method section shows that the Wall Street Journal used 129 entries of the original 500 which contained outlets that were considered extremely liberal or extremely conservative. The liberal part consists of 61 extreme sources and the conservative part consists of 68 extreme sources. When one takes the mean of the political alignment scores of both sides, the liberal sources end up with a score of 0,83 and the conservative sources with 0,9. This figure could suggest that fewer liberals are exposed and engage with far-right news than the other way around. Some of the data points in the dataset, however, seemed faulty because they did not meet the limitations that he used in his research. As we dug into the data that was used for the Blue Feed / Red Feed project, we tested it for shortcomings. We checked the data for broken links, which resulted in a number of 16 broken links, 9 conservative and 7 liberal. This means that the pool of sources that makes up the feed becomes smaller every day, potentially offsetting the representativeness of the feeds.

What can we conclude from this project?

To start off, restrictions and limitations can be found in every study. What is striking in the Blue/Red Feed project is that as a viewer of the feed you have to be aware that the data shown is not biased by the creators in any form. The study is merely comparing liberal and conservative content on Facebook to demonstrate the power of filter bubbles. This means that there is no explicit conclusion by the editors themselves, but rather the viewer himself is supposed to decide what the results mean. The goal of the study is to make people aware of their position in the filter bubble by making a bold comparison. One downside to the study is that it only looks at posts that are shared more than 100 times and have a background of 100,000 followers. Therefore it might exclude other possible relevant news outlets or viral content which do not derive from sources with many followers.

What alternative statistics would support the project?

The choice in data and methodology behind the Blue/Red Feed was chosen carefully and fit the characteristics of the project very well. However, there might be room for the development of the actuality and the broadness of the study. By broadening the limitations, a wider range of sources could be used for this study. Broader limits of inclusion would allow a larger number of sources to be included. The statistics should be updated and moderated regularly because of the increased pace of today’s world and its constantly changing media landscape. Ideally, the data would be produced by a real-time algorithm that filters users and sources to produce input for the Blue/Red feed project. The arguments presented in the Blue/Red feed study may be further supported by conducting large field research. This could provide a possible extension to the offline environment.