Trying to Get a Peek at Facebook’s Gatekeeping Power over News Distribution

In their article “Publishing, sharing, and spreading online news: A case study of gatekeeping logics in the platform era”, the authors use a reverse engineering approach to study algorithmic gatekeeping in online news distribution.

News distribution has changed, and many people find their news stories on various online platforms such as Facebook. This means that the traditional media gatekeepers are conditioned by two additional gatekeepers to distribute the news. The first are members of the audience that chose what news stories to share. Their behaviour is largely driven by a logic to manage one’s social relations. The second gate is guarded by algorithms. They are driven by a logic that is aimed to help platforms capitalise on data.

Previous research has mostly understood the distribution of news stories on the platforms as aggregated user behaviour. However, there are good reasons to consider that the platforms’ algorithms have as much or even more influence on how news stories are distributed.

Getting some light into the black box

Proposing that platforms have a gatekeeping role and that they affect the circulation of information in society is one thing. Demonstrating that they have it and to what extent this impacts the distribution of news is something else, due to the black-boxed nature of their operations, for which there are no established and acknowledged methods to unpack.

This makes for an inspiring challenge for us researchers, first, to invent a method, and second – and perhaps more difficult – to convince our peers that made-up-on-the-go methods says something about the platforms’ gatekeeping power.

In our paper “Publishing, sharing and spreading online news” we tried to find a way to peek into the black box by using the approach of reverse engineering. To the best of our knowledge, this is the first attempt to study algorithmic gatekeeping in action systematically. Using an already existing and publicly available dataset from a quantitative content analysis by the Swedish Institutet för Mediestudier, we looked at online news stories.

First, we identified the news stories most associated with the three different gates (journalists, media users, and algorithms). News stories that were published online but never shared (we call them “duds”) reasonably only passed through the gates of news media and their publishing logic. The news stories that were shared only a few times (1–10 times in our study; we call them “squibs”) had passed through the gates of news media and some users but, we argued, they had not been tainted as much by the algorithmic gate. In contrast, the news stories that had been shared many times (100+; we call them “hits”) had also been subjected to algorithmic gatekeeping.

Second, by comparing these three groups of news items we hoped to see whether there were any differences in the content variables between the three different gates. Should there be a lot of differences between the groups, this might indicate that we have caught a glimpse of the differences in gatekeeping logic at work. A lack of differences would, on the other hand, suggest that the three gates have similar priorities, and that Facebook is a neutral conduit just forwarding tidbits that others – news media and their audiences – have deemed important.

While far from being certain, this approach could at least serve as an approximation and one way of understanding and studying the gatekeeping powers of platforms in relation to users and news media.

News media and users align in their gatekeeping – Facebook does not

If our methodological approach says something about what happens at the three different gates, we did uncover two important results. First, media and user preferences seem to align quite well, since there were few differences between them. Meaning, for instance, that the proportion of news that is about politics is statistically the same between the “duds” and the “squibs”. Second, Facebook’s algorithmic gatekeeping does not seem to align with either news media or users, since the proportion of news characteristics significantly differs between “hits” and “squibs”, as well as “duds”. In our view, then, these results point to Facebook’s role as a de facto active moderator of news distribution.

Where to go from here – methodological challenges ahead

The findings from our study indicate that algorithmic gatekeeping by platforms is a greater factor to news diffusion in digital networks than user preferences. In addition to shifting attention to the role of platforms, methodological innovation is direly needed. For this to happen, we as researchers must allow and encourage novel and challenging methods as we come across investigations into platforms when serving as reviewers and in graduation committees.

Read the full article

Publishing, sharing, and spreading online news: A case study of gatekeeping logics in the platform era

News dissemination online is driven by three gatekeeping logics: the gatekeeping logic of the news media (publishing), the gatekeeping logic of social actors (sharing), and the gatekeeping logic of platform algorithms (spreading), each guided by different values and with a different relationship to content. Using a reverse engineering approach, this study applies a 2015 dataset to empirically explore how a Facebook algorithm changed the overall composition of the news users saw, highlighting the ongoing issue of how the different gates and associated gatekeeping logics – especially that of platforms – influence news distribution. In contrast to previous studies, we find the relationship between news properties and the distribution of news online to be non-linear. Results point to Facebook’s role in the overall composition of online news, both directly and in interaction with other gatekeepers. As news stories become more widely spread online, algorithmic logics take precedence over user behaviour and preferences.
Read the article arrow_forward
Publishing, sharing, and spreading online news: A case study of gatekeeping logics in the platform era