| Popis: |
Young adults’ ability to construct themselves as informed citizens depends on the information they access online on social media platforms on which recommendation algorithms play an important gatekeeping role (Claes et al., 2021). Yet, popular narratives related to recommendation systems tend to overestimate the impact of algorithmic news curation on users while downplaying their agency over the system (see the “filter bubble” theory [Pariser, 2012]). Consequently, these narratives conceal users’ strategies to act upon algorithms and restrict our understanding of the algorithmic public opinion. However, current research on the subject provides us with a more nuanced image of how users interact with recommendation systems (Bruns, 2019; Dahlgren, 2021; Moeller & Helberger, 2018). Other contributions have also highlighted how users adapt their behaviour according to complex sets of representations and imaginaries (Bucher, 2017; Eslami et al., 2015). These tactics and knowledge around algorithms are evidence of an “algorithmic literacy” (Bruns, 2019) that should be studied as it influences users’ exposure to information (Swart, 2021). For the first part of our communication, we will report the findings of a survey involving 13 young adults from Brussels (Belgium) about how they perceive and interact with algorithms on social media platforms. We will illustrate the tactics they have developed to retrieve some form of authority over the system. However, most of these tactics are limited by the affordances of the system (Davis, 2020). This raises the question of whether it is possible to develop a system that upholds users’ autonomy (Milano et al., 2020) rather than trapping users (Seaver, 2019). In the second part of our communication, we will present an experimental website (ALVEHO) developed in collaboration with the French-speaking public broadcaster in Belgium (RTBF). Our platform aims at enhancing algorithmic literacy by allowing users to tinker with the algorithm thanks to transparency design and control functionalities. In this section, we will report the first experimental results made with ALVEHO and reflect on the features that could potentially encourage users to scrutinise algorithmic recommendations and question their effects. |
