Social Media and Collective Political Action

The closing (!) keynote of Web Science 2016 is presented by Helen Margetts from the Oxford Internet Institute. Her focus is on the use of social media for collective political action – that is, for activities undertaken by citizens with the aim of contributing to the public good. There is a strong feeling that such action is happening, but as yet not enough empirical evidence about how and why it is happening.

Even those who refuse to participate online are somehow caught up in the changes that the Internet has contributed to: our lives are intertwined with its technologies, platforms, and content. And these technosocial spaces also generate a substantial amount of transactional data about user participation that goes well beyond the sort of data – for instance about political attitudes and engagement – that were available in pre-Internet days.

Such data can be used to understand and model political participation, and the acts of participation that social media in particular allow can be understood as a kind of 'micro-donation' of time and effort to specific political causes. This is a political act that was impossible before, because the transaction costs had been just too great; the transaction costs of these acts are very small and tending towards zero.

There are also various forms of engagement that were not possible before: it is possible to connect with and 'friend' political leaders; to share with a wide range of others that you have engaged politically in certain ways; and even to share material of a political nature without actually having read it.

The British sense of politics is of 'politics as pain' ("it's not politics unless participating hurts"), and these new forms of participation are quite counter-intuitive in that context, therefore; this perpetuates the sense that what you do online is not politics proper. Yet most Britons now engage in politics through online rather than offline means, or are motivated into offline action through online campaigns.

Such tiny acts of participation can scale up to huge mobilisations, however, as movements such as those off Tahrir Square in Egypt or Podemos in Spain have demonstrated; and these developments have the power to directly affect individual lives, to provide public goods at the local level, and to achieve policy change. At the same time, however, the fact is nonetheless that most mobilisations fail, for reasons that aren't entirely clear or easy to predict. (This is true for most social media content overall, too, which is similarly unlikely to reach a large audience.)

Successful mobilisations online are those that develop fast. The same is true for general trends on social media: rapid, early virality is usually crucial. But on the flipside, collective attention also decays again very quickly; if there is no take-up within a few hours of the commencement of the campaign, it is highly unlikely to reach a larger audience at a later stage.

For these new forms of mobilisation, traditional predictors of engagement – such as conventional demographics – no longer seem to work particularly well. Mobilisation now takes place among some groups that have traditionally failed to become politically engaged, and social media often serve as drivers of social influence in such cases. This is because they facilitate social information (real-time information about the participation of others in our networks) and visibility (making our actions visible but potentially also anonymisable to others). Visibility in particular can act as a kind of shaming mechanism leading others to contribute.

There is also a strong relationship between personality and participation, unsurprisingly. Some personality types were more likely to initiate the process of engagement than others, even in the absence of social information from others encouraging them to do so; some will engage only at a later stage, when there is ample evidence about everyone else in their network already participating. These personality types also react differently to visibility effects: they can be shamed into action more or less easily.

When engagement platforms – such as the U.K.'s e-petition platform change their functionality, this can serve as a useful natural experiment about how such platform affordances direct user attention. When the U.K. platform introduced information on trending petitions, attention was concentrated much more strongly on those handful of trending petitions; this indicates the presence of petitioners on the platform who are coming to it to do 'a bit of politics' without necessarily having a specific cause in mind ahead of time.

Can such forms of collective action result in a kind of leadership without leaders, then? Can it generate political success without the support of conventional organisational mechanisms? These are real questions also about the sustainability of social media-fuelled causes: the transition from social media movement to political organisation or party may be quite difficult.

Such mobilisations are therefore unstable, unpredictable, and often unsustainable. This can cause considerable political turbulence, and the metaphor may be apt: just as turbulent and dynamic systems such as the weather are now predicted by high-powered computational systems that draw on data science methods, perhaps similar methods can be applied to the increasing volumes of digital data on public communication that can now be collected.

Is there a model for the kind of democracy that is now emerging? Helen suggests the term "chaotic pluralism": a pluralist pattern of competing interests and distributed power and influence, but much more disorganised, fast-moving, individualised, and unsettled than earlier models of pluralism have been.

To research these developments, we need multidisciplinarity, experimentation, mathematical and physical modelling, machine learning, new ethical frameworks, data integration across diverse sources, cross-platform approaches, and a reintegration of findings into the policy-making, service-delivery, and democracy-design processes.