Susan Herring and Jannis Androutsopoulos’s chapter on Computer Mediated Discourse 2.0 outlines approaches that consider the temporal, multi-modal and situational contexts to consider in CMD. The examples in the chapter treat the computer platform as a neutral conduit of information where variation in content and timing of interactions are determined by the user. While I agree most formats of computer mediated discourse closely follow non-digital precedents, there are factors in the design and algorithmic makeup of platforms like Facebook and Twitter that present a new layer of influence and context that is difficult to document.
The most controversial study of the influence of platforms over discourse was by Facebook themselves. In 2012, their “Emotional Contagion” study selected 700,000 random, non-consenting users and divided them into four groups: a group whose newsfeed blocked posts with negative words (i.e. “sad”), a group whose newsfeed blocked positive words (i.e. “happy”), and two control groups. The researchers found that users in each group responded by using more negative or positive words to match their group. In other words, the Facebook platform itself influenced hundreds of thousands of people to write in a negative sentiment when they might not have otherwise. Outside of the published results, this study presents something more eye opening: Facebook is manipulating data all the time with goals and effects we can not see or know. How might we account for this flux when collecting discourse data? Can we measure how an online environment varies among subjects at the time of their posts or interactions?
In addition to modifications tailored at an individual level, Facebook and Twitter manipulate context and temporal norms. These social platforms create a dominant discourse by promoting trends that may or may not be reflective of a community’s interest or values. Because of advertising revenue incentives, controversial (click-able) content is more favorable than nuanced posts. In this way, the trending algorithm steers the subject of discourse. A post feed is also curated by the platform so that some posts rise to the top while others fade or are never seen. In this way, trending shuffles the temporal assumptions of communication. Is there a way to factor in the rise of a subject matter in discourse that may have gained traction from an outside party (i.e. advertiser, trolls, bots)? Does the constant reactiveness of the feed distinguish it from the ways other mass media influence discourse?
Algorithms are created by humans and in many ways their influence on the discourse environment can be compared to a non-digital predecessors: mass media influence, propaganda, or an administrator like a switchboard operator. However, there is speed of transmission, constant calibration, replicability of messages and audience reach that is unique. As a result, the computer is not a neutral transmission agent. Curated platforms with users who are unknown, anonymous or automated (bots) can steer and amplify discourse into radically polarized directions without being detected. Any research with Facebook or Twitter discourse data must account for these factors.
Sources:
Susan Herring and Jannis Androutsopoulos. Computer-Mediated Discourse 2.0. Handbook of Discourse Analysis, Second Edition. 2015.
Adam D. I. Kramer, Jamie E. Guillory and Jeffrey T. Hancock,
Experimental evidence of massive-scale emotional contagion through social networks, Proceedings of the National Academy of Sciences, 2014.
[http://www.pnas.org/content/111/24/8788]
Tarleton Gillespie, #trendingistrending: when algorithms become culture. Algorithmic Cultures: Essays on Meaning, Performance and New Technologies, Robert Seyfert and Jonathan Roberge, eds. Routledge, 2016.
[http://www.tarletongillespie.org/essays/Gillespie%20-%20trendingistrending%20PREPRINT.pdf]
Quick links to examples of bots influencing discourse:
[https://respectfulinsolence.com/2017/09/28/antivaxers-on-twitter-fake-news-and-twitter-bots/]
[https://www.nytimes.com/2018/02/19/technology/russian-bots-school-shooting.html]
