The Impact of Algorithms on Public Opinion Formation

The next speaker in is ECREA 2014 session is Arjen van Dalen, whose interest is in the impact of algorithms on public opinion formation at the micro (individual), meso (discussion) and macro (social networks) level; his focus here is on the latter.

Algorithms transform such public opinion formation: some 30% of users read news on social media, and that number is likely to increase. The business strategies of news media are increasingly adjusted to this trend, and the number of social media engagements with news (likes, shares, etc.) are increasingly being used by journalists as an indicator of public opinion, too.

Models for public opinion formation have always stressed the importance of individuals' social networks. This extends to social media, of course, and spread and contagion of information are increasingly being studied. Algorithms are increasingly taking over journalistic gatekeeping roles, and this is increasing the speed with which information spreads through social networks – but they constitute a black box compared to journalistic selection strategies, however flawed those may also be considered to be.

These new ways of algorithmic information selection can lead to the emergence of infostorms, where specific information spreads virally through the network. Changes to algorithms also change the dynamics of such infostorms considerably, and affect the relative value of likes and shares as a proxy of public opinion.

But there is also a growing awareness of the impact of such algorithms, amongst audiences, journalists, news organisations, and other stakeholders; research must further examine these effects, and due to the black box nature of current algorithms may need to draw on agent-based modelling to explore these effects fully.