How algorithms’ bias on Twitter’s timeline create the exclusion of creative contents

Algorithmic bias are excluding significant amounts of qualitative information at the expense of a creative culture’s production.

twitter bias

I write this post to underline how processes of social discrimination are actually at stake in encrypted codes used by Twitter developers and managers via their algorithms which are affecting the production of news on Twitter and impacting negatively the experience of non US citizens by creating streams for exclusion and segregation.

My interest for algorithms is ancient. Twitter is a platform I came to know  and enjoy through a work of research I have conducted from 2010 and 2013, during the PhD in ICTs I completed both in France and in Quebec (a french speakers’ part of Canada), about hashtag activism, digital interactions and online affiliations in times of social crisis.

There are many reasons that led me to choose Twitter versus Facebook. Twitter is the most purist platform in social media or at least the one that is looking more alike to IRCs or digital forums as it is dedicated to the production of real time information in shared streams based on some minimalist features that don’t involve much external interventions, i.e. anti-authoritative (which is also a reason to their big mess).

The move Twitter made after 2012, following its introduction in stock market and the call of investors to get it more lucrative as Twitter was not returning enough value, to complexify its kind of organic ecology and turn it into a business ROI, killed the democratic processes of producing and selecting the best information based on its intrisic value. It progressively led the staff to manipulate and to change the structure of the media with the use of filters and additive technology.

I am grateful I had to analyze Twitter production at that time because my work today would have been dismissed giving the many bias Twitter business team has introduced on the platform to become more profitable.

It’s the first PhD dedicated to Twitter based practices’ comprehension and mapping in the whole world (!) so better say that the body of literature existing on methods and concepts was more than partial or inexistent… I had to link together several pieces of incomplete expertise to design my own methodology of data visualization. As a result, I spent countless hours forging intuitive observations and doing research online to discern emerging patterns.

Still, because it was so new and exciting : some predicted the death of platform in two years, the group I was studying was just starting to use the platform… There was nothing better than to wake up in the morning and to connect online to see all the new babblings posted chronologically during the night as I was back in France at the time and Montreal is 6 hours earlier. You know this Christmas tree feeling? I had that or something resembling.

These times were nice because, while working on my research on digital coalitions in Toronto that opposed police brutality in summits, I had the chance to catch the emergence of the Arab Spring’s movement, converging with the Quebeckers’ Spring, and also the intensification of hashtag activism everywhere in the world like a blossoming in spring. It was the first ever social movement being watched in real time and I might say it was really emotioning to assist to all these hidden voices raising together with authenticity. That was magic.

At the time, Twitter was the place-to-be for researchers, librarians and talented social justice activists who had better access to primary sources and diversified and unfiltered contents so they took the lead to drive users where to go. It was a place for building solidarity networks, shared knowledge and information, and civic friendship around common issues and humanitarian causes, and it was also the place for personal narratives and cultural contents, the place of the voiceless.

There were the Arab Spring opposing dictature, the Quebecker student strike Red Square against the scholarship fees, the Occupy movement to reclaim affordable housing, the protests in Spain and South America against authoritarian policies and austerity, that would later give birth to the massive engagement for the American elections (Obama, Trump…), the movement for women against violence (#WhyIstayed, #BringBackOurGirls, #Metoo, #MosqueMetoo), the #BlackLivesMatter movement, #PreyforParis, #netneutrality, so on).

Basically Twitter INC. owns its design to the work of us.

There were critics too of clicktivism, hacktivism or slacktivism as the people never meet offline. There are limits of course but it is free, easy to access, be seen platform and it has become the privileged media from the oppressed.

For a researcher, the platform is a good site for observing the forms of participation of the audiences for news production and sharing. I wrote a consequent paper on how the information is processed through times and space and is remediated by Twitter users to enable minority groups to reach audience and visibility.

But I also noticed, following publication, how contents from the dominant culture (i.e. males in tech, white americans, digital natives, english speakers) were privileged at the expense of the ones produced by minority cultures. Let me make my point here. The dominant culture is determined by the composition of the main group in authority and is not a fixed identity, which means, for example, that in technology, the dominant group is the one of white males, when in education or health, it is the one of women. In a group of whites, it is the black person and in a group of blacks it is the white person. Numerical.

Even if there are historical groups of minority exclusion in labor or society (slavery, exploitation…), discrimination is also determined by the context. Working in the field of advanced technology, I am confronted as a woman to direct or indirect exclusion as a woman in my position only makes 0 to 20 % of the workforce.

On Twitter, producing a scientific account that doesn’t reflect a western-centered perspective or doesn’t play the big corps’ interests is an act of bravery and it takes money and time to counter the dominant narratives.

My researches conducted me early to worry about the limits of the micro-media platform that already introduced filters by the end of 2012 to highlight in the timeline the most popular contents among users. In a democratic situation, people vote for these contents, but in an oligarchy, the army of bots is used for the amplification of the dominant narratives, dismissing the potential of technologies for democracy and social change.

Algorithms are each day excluding more valuable and qualitative information due to bias generated from their mere creators.

End 2015-2016, Twitter INC. implemented an algorithm that allows users to see more tweets from the people they interact with and the most popular tweets from the followees – the people you follow (this, fine), but the algorithm also makes you see more tweets of what Twitter decides to be popular contents for you based on the number of sees, likes and retweets, just as Facebook. These decisions are based on personal assumptions and normative contents. And it’s where things get wrong. Algorithms adding bias where there aren’t!

Plus, it is based on numbers only and not on contextual evaluation thus whose buying bots to make their narratives more visible instead of producing valuable contents get extended visibility in the timeline. This contributes to a poor design.

It’s understandable that Twitter has tried to improve its literacy giving the increasing amount of information being produced daily by users (or bots!) but there is an issue about the way the algorithm used has been coded at the expense of non US citizens by people who are serving their own personal interests in a competitive market. It’s much more broader than the firm strategy.

And THIS is really bothering me a lot because as a social network researcher having interests for alternative contents and creative practices, I know that views that are not the most popular, i.e. emerging ways of seeing the world and doing information, are not necessarily majoritarian and often gets more significance than others.

Filters completely messed with my experience on Twitter as a scientist, and also in my interactions with others when I faced limits because I was not having thousand of followers or bots to make me visible. Talking about unconsensual subjects or taboos (like rape abuse back in 2012, war tactics or corruptions in liberal politics, back in 2013), I faced cyber-harassment.

As a coincidence, the highlighted tweets in the timeline be it randomized or by hashtag research are from American people located on the West Coast, i.e. from the same location than the very guys who are programming the machine, when they don’t represent a significant portion of the people I follow. And frankly, I don’t care about their views. My scientific and personal interests are not about tech insiders’ views. Basically I am super bored now.

This peculiar features are making of Twitter a media for celebrity where the most seen get the most authority biasing readers into the choice of source and information, confronting him with the test of authenticity and with information overload.Even scientists are at stake with the process of curation of news so there are making things hard for people lacking literacy.

The algorithm is rendering the processes opaque giving premium access to sponsors and partners in the timeline, enacting democracy and science.

Manipulation also makes technology vulnerable to scams as some class of people can afford buying followers and then plaguing already existing contents, usurpating online identities, hiding the merits of whose producing the original works or fighting for public goods, the authentic people keeping Twitter alive and human.

The ones that we see today in the #netneutrality protest, the ones who come from the margins and from the minorities, not the beautifully marketized tribes portrayed in some demagogic campaigns, the ones of the ugly, the difform, the hurt, the betrayed, the disfigured, the ones who fight to reach ends every day, the ones who never found their way in polls, the ones of the invisible.

To protest against that increasing ABUSES from the funders, so ingrateful, I have asked Twitter : « remove filters for 2018 » in order to reinitialize the platform the way I used to know, cherish and use and not in the best interests of its investors. Now with the end of net neutrality in the U.S, we can fear that this move for profits accelerates its path with even more censure, even less symetry, which would permanently damage the fragile ecosystem of news and contents that users have been contributing to build through the years, for some, at their own perils.

I am talking about all the social justice activists that used Twitter for criticizing abuses and ended up in jail. I am talking about all the women who bravely confronted public stigma to share their experiences about rape to end violence against women. I am talking about the people who got murdered in countries of war for exposing truths. And I am talking about the little Twitter itself that has lost his soul and his crying his mamma.

This is really saddening me too much. I hope Twitter got that without the users that made it grow, it is nothing else than a ridiculous gadget in the hands of capitalists of disaster.

This is consterning because Twitter, Facebook and Google are pro-american imperialism tools that aim at benefiting the US oligarchy ONLY, by appropriating the work of local and foreign communities to make high profits. This is consterning because they play the game of the sophists, sustaining authoritative governments.

Maybe we could encourage Twitter dream team to read a bit more about the concept of friendship in the terms of Aristotle, as a civic friendship bridging people together in times of scarcity, and not in the terms of their 3 best friends.


Licence Creative Commons
Ce(tte) œuvre est mise à disposition selon les termes de la Licence Creative Commons Attribution – Pas d’Utilisation Commerciale – Partage dans les Mêmes Conditions 4.0 International.