Anonymity and Online Discussion: A New Framework for Analysis

We present a way of disaggregating the concept of anonymity and linking it to particular democratic goods. We demonstrate its value in the particular context of studying online commenting by contrasting three distinctive commenting regimes used by the Huffington Post (HuffPo) in the period January 2013 - March 2015. But we think this may have wider relevance for scholars looking at online communication, as modes of identity disclosure are an unavoidable yet consequential design feature of online communication platforms. In this paper we disaggregate three aspects of anonymity: traceability, which we argue relates to inclusion; durability, which influences deliberative quality through varying levels of communicative accountability; and connectedness, which ties users to real-world relations.
Decisions about online architectures have a crucial influence on the quality of communication, but are often adopted by default or with regard to other factors. In the realm of online commenting, we observe a trend of commenting platforms being outsourced to the Facebook commenting API. These changes are often framed and justified in terms of a shift from an anonymous but easily abused environment, to one in which users operate under real-name identities. When we think in terms of a choice between anonymous and real-name architectures, it seems that there is a simple trade off between the goods and dangers associated with anonymity, such as trolling on the one hand and the freedom to express one’s views without fear of recrimination on the other, and the goods and dangers of real-name communication, which ties users more closely to discursive norms of civility but which also risks reproducing offline power relations.
We propose to analyse the concept of anonymity in three dimensions. We argue that cross-platform connectedness is associated with increasingly personalised communication as well as less communicative engagement between individual users; durability (continuity) of identity is associated with greater levels of civility, reduced trolling, and higher quality of deliberation (more reason-giving, etc); traceability is associated with exit - that is, people worried about traceability leads people to make a binary decision, to opt out, and the increasingly pervasive traceability promotes exit. Thinking of anonymity in these terms helps resolve some apparent trade-offs: it may be possible to capture the benefits of communicative accountability without the drawbacks of either the abusive space of easy anonymity or of the exposure to offline power relationships associated with real-name spaces.

Disaggregating anonymity
Anonymity exists on one end of a spectrum of degrees of disclosure of identity. In the context of commenting, anonymity means contributions to a discussion are not visibly linked to any particular commenter. Pseudonyms - names chosen by commenters - allow users to keep their real identity private if they wish, yet allows them to maintain a persistent alternate identity in the context of the forum in question. You might have an identity as a commenter on the
Guardian that you keep separate from your professional networks and from your social networks. Real name comments, obviously enough, appear under your real name, not a pseudonym.

While this is a useful starting point, many scholars analysing the different degrees of disclosure on online platforms have sought more fine-grained distinctions (Marx, 1999; Ruesch and Märker, 2012). We favour splitting online anonymity into three dimensions: traceability, durability, and connectedness.

Traceability refers to the extent to which your contributions can be traced to your real identity. Traceability is distinct from disclosure of identity to fellow commenters. You can make comments under a pseudonym and yet it may be possible (with some effort) for advertisers or the National Security Agency (NSA) to trace your real identity. Nissenbaum, for instance, argues that anonymity online, in the sense of ‘conducting one’s affairs, communicating, or engaging in transactions anonymously in the electronic sphere... without one’s name being known’, is undermined by technologies that have made it possible to track and / or piece together (‘infer identity from non-identifying signs and information’) the real identities of citizens online even when they are withholding their names or using pseudonyms (Nissenbaum, 1999). Because traceability is strictly distinct from the question of anonymity or pseudonymity
with respect to other commenters, it does not help us grasp the relation between disclosure of identity and discussion quality.

Durability refers to the ease or difficulty with which online identities can be acquired and changed. Where new pseudonyms are easy to create, online identities are disposable; if you acquire a reputation for abusive or untrustworthy behaviour you can just create a new pseudonym and start again. As Resnick and Friedman put it, cheap pseudonyms create ‘opportunities to misbehave without paying reputational consequences’ (Resnick and Friedman, 2001). Users are more likely to stick with a particular name, exposing them to the reputational consequences of their behaviour. The durability dimension is particularly important for the possibility of holding commenters accountable for claims they make, enabling challenges in terms of consistency, and exposing uncivil or abusive commenters to sanctions.

The third dimension has to do with
connectedness or bridging across different platforms and contexts. The most visible example of cross-platform connectedness is sites enabling users to login or register through a third party - typically Google or Facebook - rather than filling in a site-specific form. Connectedness also involves reputation, but is a global rather than local reputation. The durability or stickiness of an online identity is a necessary condition for building a local reputation, but it need not be connected to the wider reputation, a cross-referenced, cross-platform (including real life) reputation, of the sort you would want if you were renting out your apartment to someone you didn’t know (as in Airbnb).

The Experiment
The two changes introduced by HuffPo provided us with a natural experiment. We collected more than 50 million comments on more than 50,000 articles featured on the HuffPo front page in the period January 2013 - March 2015.
The first of these phases allowed anonymous commenting. At this time, the platform experienced aggressive ‘trolling’ and the use of multiple accounts. In December 2013 HuffPo moved to regulate its forum by requiring users to authenticate their accounts through Facebook. On the face of it, little changed in this pseudonymous environment: user names and avatars remained, but behind the scenes Facebook’s database helped weed out fake accounts. Users did not have to create a new online identity, give up their old pseudonyms, or change the appearance of their HuffPo commenter profile. We read this first change as primarily about durability of identity.
In June 2014 HuffPo changed to commenting through Facebook, meaning that HuffPo user profiles were
replaced by Facebook profiles in a ‘real name’ environment. In this phase, comments appear below the line of the news article under the user’s real name, as well as - depending on a user’s privacy settings - appearing simultaneously on their Facebook page. This may have the effect that users comment on the HuffPo but speak to an audience located on Facebook. We read interpret this change as a move to more integrated and connected online identities.

While we found more durable online identities were associated with greater civility but less participation in online commenting, we also observed a shift in the sorts of issues on which users most frequently comment. This might be normatively encouraging: the relative rise in articles tagged Gay Voices, we speculate, may be a result of the inhibition of hostile and offensive commenters. Here greater civility seems to promote more participation and inclusive discussion. However, we also note a more general shift away from conflictual and politicized topics and towards ‘safer’ topics of celebrity, lifestyle and consumer issues. Turning to the second commenting change, we found that the Facebook phase exhibited markedly less argumentative engagement relative to the 'pseudonymous’ phase, as evidenced both by an overall reduction in the proportion of replies, and by measures of deliberative quality (Fredheim et al., 2015a, 2015b). This finding points in the same direction as a recent Pew survey that found people were less likely to discuss a controversial issue on social media than in face to face conversation (Hampton et al., 2014).
With these findings in mind, we argue that the pseudonymous phase harnessed the beneficial effects of increased durability, without introducing the costs of connectedness. More broadly, we can link these dimensions of identification to particular democratic goods. The dimension of traceability seems to be associated with inclusion and exclusion, as users make a binary choice to stay on the platform. The dimension of durability or continuity of identity is associated with deliberative quality in so far as (theoretically) it exposes users to communicative accountability and (empirically) is associated with a higher density of reason-giving in commenting. The dimension of connectedness ties users to real-world relations. This favours civility, but it also promotes what Schudson calls ‘sociable’ conversation rather than more adversarial ‘democratic’ conversation (Schudson, 1997). We argue that durable and disconnected identities are more conducive to productive, issue-based debate between heterogeneous users. This bears on discussions of polarization. It also bears on the drive towards integration of social media and news discussion.

Bibliography

Fredheim, R., Moore, A. and Naughton, J. (2015a). Anonymity and Online Commenting: An Empirical Study. http://papers.ssrn.com/abstract=2591299 (accessed 27 November 2015).

Fredheim, R., Moore, A. and Naughton, J. (2015b). Anonymity and Online Commenting: The Broken Windows Effect and the End of Drive-by Commenting.
ACM Web Science 2015. University of Oxford (accessed 27 November 2015).

Hampton, K., et al. (2014). Social Media and the ‘Spiral of Silence’
Pew Research Center: Internet, Science and Tech http://www.pewinternet.org/2014/08/26/social-media-and-the-spiral-of-silence/ (accessed 2 March 2016).

Marx, G. T. (1999). What’s in a Name? Some Reflections on the Sociology of Anonymity.
The Information Society,
15(2): 99–112.

Nissenbaum, H. (1999). The meaning of anonymity in an information age.
The Information Society,
15(2): 141–44.

Resnick, P. and Friedman, E. J. (2001). The social cost of cheap pseudonyms.
Journal of Economics and Management Strategy,
10(2): 173–99.

Ruesch, M. A. and Märker, O. (2012). Real Name Policy in E-Participation.
CeDEM 12 Conference for E-Democracy and Open Government 3-4 May 2012 Danube-University Krems, Austria. Edition-Donau-Univ. Krems, pp. 109 (accessed 23 March 2015).