Understanding Algorithmic Journalism

Berlin.
The afternoon session at the Berlin Symposium, on intermediaries in public communication, begins with Chris W. Anderson’s presentation on data journalism (he’s not the ‘long tail’ guy, by the way). He begins by describing journalism as a media form that’s meant to bring the public together – to assemble the reading public. In a sense, Google, and data algorithms, similarly bring the public together – and intermediaries emerge in this process.

Algorithms are predetermined sets of instructions for solving a specific problem in a limited number of steps; one of the best known algorithms of recent years is Google’s PageRank algorithm, of course. They are hybrid entities, cyborgs, both human and machinic: they combine both human intentionality and social structure, and technological affordances. In other words, they’re part of the social world, not machines impacting on it from the outside – but they’re also not determined entirely by social and societal forces, but retain technological qualities.

In the context of journalism, they are part of the wider news network – of overall processes by which sociomaterial fragments of reality are interwoven to form news products. These fragments include documents, human sources, media platforms, audiovisual materials, digital data, newsrooms, audience metrics, eyewitnesses, etc. Algorithms help facilitate this process of network assemblage, but are also themselves part of the network itself, acting upon journalism as well as being acted upon.

The Google News algorithm is an obvious example for this; other data journalism tools which are capable of analysing very large datasets very quickly can also be included here. But how can these algorithms be studied? One approach is through Schudson’s lenses for examining the sociology of news, augmented with further approaches: these lenses include

• a political or regulatory lens (do different nations’ regulatory regimes affect how ‘open’ and available datasets are? does this affect the types of journalism which are able to emerge?);
• an economic lens (what are the costs of computational journalism? does this create an imbalance in the competition between journalism organisations?);
• an organisational lens (how are decisions of news coverage and news agendas made?)
• a cultural lens (does the idea of computational and algorithmic journalism change the audience’s or journalists’ views of what journalism is?);
• an institutional lens (how do different institutions work together to make computational journalism possible, or not?);
• and a computational lens (how do technologies enable computational journalism, and what forms of it?).

What research questions emerge from this, then? First, how are algorithms influencing journalistic work routines? Second, how do (internationally different) open data policies and transparency initiatives impact on journalism? Third, what are the sociotechnical affordances of algorithms – what do they enable us to do, what do they prevent us from doing? Fourth, what is the relationship between economic resources and the uptake of computational journalism? And finally, is there an emerging culture of algorithmic journalism – is a new form of news emerging, and through this, given the importance of news to the overall public sphere, are societal changes occurring?