In past eras of limited publicity and professional multipliers, journalists dominated the procurement of viewpoint diversity, lending relevant contenders’ ideas visibility alongside hegemonic voices. Today, the algorithm-amplified activities of countless voices and audiences govern the visibility of competing viewpoints in the digital public sphere. However, as I shall argue, the enormous expansion and fragmentation of publicity has merely shifted, but not diminished the importance of journalism. As a point of departure, I submit that viewpoint diversity is suspended between two poles: the hegemony of one viewpoint, and the entropy of all available viewpoints – most of which are unsubstantiated or redundant. Accordingly, exposing audiences to relevant contestation is equally critical as is organizing the cacophony of voices and discriminating relevant from irrelevant diversity. In this effort, journalists navigate an information environment that is heavily pre-structured by algorithms.
In this paper, I review the existing literatures on viewpoint diversity and digital journalism to identify key contributions and conflicts that algorithms create for the management of relevant diversity. I propose that algorithms play a highly ambivalent role depending on which criterion of relevant diversity is considered. With regard to identifying diverse experiences, algorithms may be capable of identifying viewpoints shared by many in society, but remain limited at exposing other groups to these experiences. Herein thus lies a key journalistic contribution. At the same time, these algorithms also wash singular or contrived experiences to the fore where these appeal to common fears and stereotypes. With regard to identifying diverse knowledge, next, algorithms are largely useless: They detect only what is believed or contested, not what is important to know, and their proxies for truth – the beliefs of authorities and crowds – are dubious at best. Professional journalism remains indispensable for determining what claims valuably complement existing knowledge. Still, algorithms can flag widely held beliefs that require checking and, if necessary, correcting. With regard to procuring political-normative choices, finally, algorithms are generally useful for crystallizing alternatives, but also easily rigged. However, both their capacities and limitations are rooted in the strategic advocacy of political and social groups, and thus make little difference compared to ‘old-fashioned’ journalistic work. While journalists are no longer responsible for rendering contention public, their contribution by far exceeds the pointing-out of relevant contributions. By bridging distinct ‘filter bubbles’, verifying pertinent claims and exposing baseless viewpoints, journalism is at least as critical for managing excessive diversity as the amplification of suppressed contention.