Social Media

The next speaker at "Compromised Data" is Joanna Redden, whose interest is in government uses of 'big data', especially in Canada. There's a great deal of hype surrounding 'big data' in government at the moment, which needs to be explored from a critical perspective; the data rush has been compared to the gold rush, with similarly utopian claims - here especially around the ability for 'big data' to support decision-making and democratic engagement, and the contribution 'big data'-enabled industries can make to the GDP.

But how are 'big data' actually being used in government contexts? New tools and techniques for the analysis of 'big data' are of course being used in government, but how these affect policy decisions remains unclear. Social media analysis is similarly being used for public policy and service delivery; sentiment analysis is used for some decisions around law enforcement and service delivery, but adoption to date is slow.

The final presentation in this "Compromised Data" session is by Mary Francoli and Dan Paré, who focus on the question of engagement and mobilisation in a time of rapidly evolving social media use. One initial observation is that these terms lack definitional clarity - there are some very high-level definitions (e.g. building on UN definitions), but these remain vague; political and civic engagement are conflated, and specific forms of engagement are not necessarily defined in detail.

Simply voting is a form of engagement, for example, but is clearly different from other, more complex forms of political engagement. The literature increasingly links these types of activity with social media (and with the Net more broadly) - and the extent to such such forms of engagement occur, and how they interrelate with forms of offline political engagement, need to be studied in greater detail.

The next session at "Compromised Data" starts with Frauke Zeller, who begins by noting the multimodality of communication, including through social media: many texts are using more than one semiotic mode, combining text, images, audio and video. How can the existing methods for studying multimodality be transferred to online environments, and to research building on 'big data', however?

Some such work begins with exploring the networks between users, and between texts, but this is not enough - how do we move from the macro to the meso and micro levels of communication? How do we move to the manifest to more latent content, especially where non-textual content is involved?

The next speaker at "Compromised Data" today is Carolin Gerlitz, who begins by suggesting that social media data are both standardised and vague at the same time. She notes the German Twitter community which is focussed around favouriting on Twitter: the Favstar sphere sees favourites as a sign of importance and validation, and taking away favourites is therefore a serious affront.

This is an example of how the communicative affordances of social media platforms are being utilised by their users; these standardised activities mark the grammar of action on such platforms, and are specific to the particular platforms. Twitter's grammar has been comparatively stable, while Facebook has modified its available actions on a continuous basis, which destabilised the meaning of such activities.

From the always fabulous Association of Internet Researchers conference I've made it to the research colloquium "Compromised Data", organised by Greg Elmer and Ganaele Langlois in Toronto. We're starting with a presentation from Taina Bucher, on the enactive power of APIs. Beyond merely helping to collect data, APIs have a number of other functions. We must not get too seduced by the opportunities for data access and visualisation which they provide, but take APIs seriously as a mechanism for making meaning from data that privileges certain approaches over others.

This means a kind of "digital humanities" in reverse is required to understand these tools and their implications - a software studies perspective on APIs, from an expanded notion of software as a neighbourhood of relations that enables software to be studied in the same way as other cultural objects.

The final presenter in this AoIR 2013 session (and thus, the final presenter this year!) is Hazel Kwon, whose aim is to better understand the flows of communication on social media during protests. Her frame of research in this is Emergent Norm Theory, whose emphasis is on the rapid and transformative potential of word of mouth on collective behaviours. This is a process of diffusion for a collective identity.

Protests can be understood as collective behaviours. They may be prompted by the circulation of rumours, which are characterised by the informal and improvised circulation of situational information; gradually, key themes and issues are being identified and converted into key messages that define the protest action. They draw on a special type of crowd, the diffuse crowd. But existing theories largely consider such phenomena in the context of physically co-located crowds; translation to social media environments must necessarily develop somewhat different understandings.

The next AoIR 2013 paper is by Pieter Verdegem and Evelien D'Heer, who shift our focus to regional elections in Flanders. The role of Twitter in politics has been described from both optimistic and pessimistic perspectives; the Twittersphere has been seen by many to reflect existing social structures. Is there a move from formal and representative politics towards networked politics, though? From broadcasting to convergence logic?

Pieter and Evelien captured all @mentions in the #vk2012 debate, engaging in both content and network analysis. The hashtag was promoted by the public service broadcaster in Flanders, so it provides a useful point of entry into election-related discussions on Twitter and was frequented by politicians, journalists, and ordinary Twitter users. There was a significant spike in tweets on Election Day (14 Oct. 2012), with far less activity on other days - usually around 200 tweets per day. Activity picked up somewhat during the final week of the campaign.

The next presenter on this AoIR 2013 panel is Christian Christensen, whose interest is in the minority parties in the US presidential election. He examined the tweets of four minority parties, defined as parties which had enough ballot listings across the states to technically be able to win the election: the Libertarians, the Greens, the Constitution Party, and the Justice Party. This, then, is a study of third party politics - and such parties have traditionally adhered to a polarising and populist style of politics.

In combination, the four parties' candidates had some 129,000 Twitter followers, led by the Libertarians with 100,000. They tweeted only a limited amount of time during the campaign, mostly during the debates and on the day before Election Day. Retweets of their tweets were often centred around a small number of original tweets, and were more or less proportional to their total number of followers.

Well, with our Twitter and Society book officially launched, I'm now in a final AoIR 2013 session on politics and Twitter. First off, Kevin Driscoll is presenting on the role of Twitter in the US presidential election, noting how much "Twitter's opinion" was used as a yardstick for overall public opinion. There is some slippage here: "Twitter" as the Twitter community, "Twitter" as Twitter, Inc., and "Twitter" as a source of opinion data.

Kevin and his colleagues examined the Twitter activity around the three US presidential debates, following the live Twitter streams as the debates happened and dynamically adding more and more keywords to track on Twitter. They divided these tweets into retweets and original tweets. Some 0.01% of all users accounted for around 25% of all retweeted posts - and these users included politicians, pundits, journalists, comedians, and a variety of other accounts; 62 comic accounts were the source of 4% of all retweets.