Personalization in Social Media: Challenges and Opportunities for Democratic Societies

Personalization algorithms perform a fundamental role of knowledge management in order to restrain information overload, reduce complexity and satisfy individuals. Personalization of media content in mainstream social media, however, can be used for micro-target political messages, and can also create filter bubbles and strengthen echo chambers that restrain the exposure to diverse, challenging and serendipitous information. These represent fundamental issues for media law and ethics both seeking to preserve autonomy of choice and media pluralism in democratic societies. As a result, informational empowerment may be reduced and group polarization, audience fragmentation, conspiratorial thinking and other democratically negative consequences could arise. Even though research about the detrimental effects of personalization is more often inconsistent, there is no doubt that in the long run the algorithmic capacity to steer our lives in increasingly sophisticated ways will dramatically expand. Key questions need to be further discussed; for instance, to what extent can profiling account for the complexity of individual identity? To what extent are users, media and algorithms responsible in such practices? What are the main values and trade-offs that inform designers in such a fundamental societal algorithmic arbitrage? How is social media’s personalization directly or indirectly regulated in the European Union? The thesis firstly presents a critical overview of information societies, analyzing social media content personalization practices, dynamics and unintended consequences. Secondly, it explores the role of serendipity as a design and ethical principle for social media. Thirdly, the European legal landscape with regard to personalization is analyzed from a regulatory, governance and ethical perspective. Finally, it is introduced the concept of ‘algorithmic sovereignty’ as a valuable abstraction to begin to frame technical, legal and political preconditions and standards to preserve users’ autonomy, and to minimize the risks arising in the context of personalization.