How I Learned to Stop Worrying and Love the Bots

Abstract

This articles inquires about the consequences of social bots for the study and understanding of social media.

Media scholars, whether ones who study old or new media, are predisposed to believe that the texts we study are ones produced and consumed by humans. For the most part, that has been unquestionably true prior to the advent of the Internet as we know it, and even during the late 20th century at the time of the Internet’s most rapid rise in use and development, it was largely true. Yet, some 10 years later, it is also the case that we are increasingly finding that machines communicate with humans, interpellating us “as always-already situated within the learned behaviors of an individual user and also the aggregate bloc of users whose communication has been mined and algorithmically processed to present a seemingly autonomous and coherent interlocutor” (Jones, 2014, p. 253).

Indeed, the last couple of years in particular have seen an upswing in the preponderance of machines communicating with humans and in some cases seeming human: The uncanny valley is less uncanny when it is textual and not visual. Wired magazine’s late 2014 headline, “Bots Now Outnumber Humans on the Web” (McMillan, 2014) is perhaps the most forthright among the many that have been appearing the last few years that acknowledge the degree to which scripts, algorithms, bots are woven throughout computer-mediated communication, particularly social media. Wagner, Mitter, Korner, and Strohmaier (2012) provided a succinct definition of bots found on online social networks: “Social bots are automatic or semi-automatic computer programs that mimic humans and/or human behavior” (p. 41).

There are recent studies that seek to understand the role bots play in online social networks (Boshmaf, Muslukhov, Beznosov, & Ripeanu, 2011; Chu, Gianvecchio, Wang, & Jajodia, 2010; Edwards, Edwards, Spence, & Shelton, 2013; Zhao, 2006, among others), and most of them focus on the functionality of bots, that is, on their credibility as interlocutors, or their ability to “pass” as human. Fundamental and interesting as that focus is, it skirts around the question of the social and remains rooted in the realm of the Turing test; we are still haunted by ELIZA (Weizenbaum, 1966). The question is no longer whether bots can pass, but how social interaction with them may be meaningful.

But how to address the social question? The automation of posts that seem human may seem to pose some questions regarding method and analysis. Some might question whether they ought to be at all included in data that are collected and analyzed, or that they should be excluded or separated during analysis. Bots do pose some theoretical challenges. They represent, literally and figuratively, what is being posted, distilling it, washing it, processing it, and finally posting it, waiting for a response and all the while indefatigably taking in more of what is posted, infinitely repeating the cycle. They are altering the border between human and machine as surely as that border is being shifted by increased use of intelligent agents like Siri or Cortana. How shall we account for social structures that include social machines?

Of course, they are also, at least at this time, human creations, scripts, code written by people, for particular purposes. They both elicit and operate within particular contexts and constraints that rely on the symbolic construction of reality. As Carey (1992), borrowing from Cassirer, wrote, “one must examine communication, even scientific communication, even mathematical expression, as the primary phenomena of experience and not as something ‘softer’ and derivative from a ‘realer’ existent nature” (p. 26). The expressions and exchanges visible on social media ought not therefore be analyzed and interpreted solely without context or without consideration of their symbolic and affective dimensions, whether originating in humans or machines. To put it another way, we can neither ignore the prevalence of human–machine communication nor can we excise it from our analyses and theories of social media. Particularly for those of us committed to the idea “that reality is continually being made through human action” (Grossberg, 1993, pp. 331–332), it may be difficult to reconcile how much a role machines may have in that process. It may be easier to see that social media can connect people in myriad, fascinating, ways but more difficult to see that it can connect people and machines, although both connections are pertinent, for both contribute threads to the tapestry of social media and, ultimately, to the reality in which we live.