To Nudge or To Remain Neutral: Should Public Media Organisations Use Recommender Systems?

Abstract

Recommender systems (RS) are increasingly taking over the selection, ranking and presentation of information online. Besides the potential of RS technology to increase access to information, enhance user experience, and increase engagement, profit-driven RS may also cause addiction, limit autonomy, and harm the public debate. Meanwhile, public service media (PSM) must adapt to the changing market dynamics and preferences of audiences, while simultaneously protect their democratic and societal role, as well as uphold editorial standards. This raises the following research question: Should public service media use algorithmic recommender systems? In this thesis, commercially incentivised, personalisation-based RS are rejected, because they have autonomy limiting features and undermine the democratic role of PSM. Instead, value-driven design, the idea that values (like diversity) can be programmed into algorithmic design, offers the possibility of operationalising core PSM values. This thesis argues that public service media, constrained by three conditions regarding transparency, manipulation, and governance, should use public value-driven recommender system technology in their digital platforms, because it could protect their societal commitments.