Can The Retweet Speak? Agency in viral video diffusion

I thought I’d share my abstract for Theorising The Web ’14, the conference run by the Cyborgology blog team.

I wrote it over the weekend, mostly spent hanging out with Jay Springett at Critical Exploits and elsewhere. He’s posted his TTW pitch, “Infrastructure Territory and The Geopolitics of The Stacks”, saying wisely that “even if i am unsuccessful, i think the piece is a useful early 2014 ‘flag in the ground’ to orient my self around this year. as this is the direction a lot of my thinking and speaking will headed”.

Good plan. Let’s do the same.

As you may be able to see, this proposal is built on all the viral media research I’ve done at FACE with Twitter (and others) in 2013 – seeking now to connect this with theoretical debates, yet in a grounded, bottom up way. What do the actual messages sharing viral videos say towards this theoretical split between agency and ‘interpassivity’?


Much analysis of viral media uses a ‘transmission’ model of communication where the people sharing the content are constituted as essentially passive or mute. This is variously achieved:, from over-reliance on biological metaphors of virality and infection stemming from Dawkins & ‘The Selfish Gene’; to arguments of ‘clicktivism’ or digital ‘interpassivity’ (Zizek 2009, Jodi Dean 2009, Paul A. Taylor) where the affordances of social media systems undermine the scope for individual agency. Network analyses (e.g. those of the Kony 2012 phenomenon) primarily explain viral outcomes from the structural network properties of the initial “seeders”. In contrast, other work (e.g. Berger & Milkman 2012) ascribes viral potential to the content itself, particularly its emotional payload. The decision-making of actual people receiving and re-sharing the content – the behaviour defining ‘virality’ – is strangely absent.

There are counter-currents: ‘Spreadable Media’ (Jenkins et al. 2013) argues that people sharing media content engage in active choices about what to share and how to frame it, continually re-authoring the content’s meaning. boyd, Golder & Lotan’s 2010 paper, ‘Tweet tweet retweet’ is also crucial for framing retweets as communicative acts in their own right, not mere transmission.

Can the retweet speak? This paper addresses this question of agency in digital media sharing through empirical examination of the actual active communicative act involved in sharing videos on Twitter. Media-sharing on Twitter is interesting precisely because of the ‘thinness’ of the 140-character message that is available to ‘speak’, and complicated further by retweets and automated sharing. It thus offers a fascinating limit-case for the assertion of agency and user-generated meaning.

Data is drawn from the Twitter Firehose API, gathered through work at London research agency FACE using our Pulsar research platform. The dataset contains every tweet sharing the URLs of the Commander Hadfield and Dove Real Beauty Sketches YouTube videos.

This paper will first analyse how people share videos, quantifying the ratio of retweets vs. original tweets as an initial proxy for passive vs. agential modes of sharing. Coding the content of people’s messages when they share a video URL (and how others interact with these tweets) allows us to build a framework for the types of communication taking place. However, message content also provides a route for quantifying the impact of  ‘automated participation’ such as default tweets from ‘Share This’ buttons. Finally, social network analysis will visualise the impact of ‘influencers’ and marketing strategies in shaping how viral media travels through Twitter, and identify the messages & framings of the content that attain the greatest visibility.

To conclude, I will ask who wants to understand viral video diffusion and why, through critically reflecting on the business objectives of commercial projects I’ve carried out for Twitter, Tumblr and  media companies.  Virality can appear a mob-like threat to traditional media control – but also a resource to be harnessed. I discuss commercial social network analyses as creating legibility and power/knowledge, allowing two ways to profit from virality: as ‘influencer’ or as ‘platform’.

Comments are closed.