The Myth of Personalisation: how Netflix is shaping users' algorithmic identity

Abstract

In this thesis the Personalised Recommendation System (PRS) of the online streaming platform Netflix has been analysed. By doing so, this thesis has analysed the role that Netflix plays in the debate around online personalised content. Within this thesis, the following research question has been answered: How and to what extent does the PRS embedded within Netflix and the discursive and social practice surrounding it shape users' algorithmic identity? This question has been answered by proposing a modified framework of the narrative identity theory by Ricoeur, namely algorithmic identity. This research argues that a new framework is needed because the algorithms within platforms now provide the medium that shapes identity. Therefore, the construction of algorithmic identities emerges in three stages where: users use the platform to create data (algorithmic₁), this data is being used by the platform to set the PRS algorithms in motion and deliver personalised content (algorithmic₂), and finally users can reflexively read and understand themselves through the identity prescribed upon them within their personal profile page (algorithmic₃). The method of Critical Discourse Analysis by Fairclough enabled this research to analyse the following aspects of the PRS of Netflix: (1) the Netflix interface, (2) articles and social media posts around the production and consumption of the PRS and (3) the social practice of online personalisation. It has been concluded that all three stages of algorithmic identity can take place within the interface of Netflix. However, algorithmic identity construction does not fully take place since within the final stage, algorithmic₃, the users of this research do not affiliate themselves with the personal profile reflected back to them by the Netflix interface. As a result, while algorithmic₁ and ₂ take place, algorithmic₃ enables users to reflect upon the true personalisation that is being withheld from them because they are being translated into dividuals. Therefore, online personalisation as fully personalised is seen as a myth because there is a tension between the identity created by the algorithm and the identity of the users.