The Ethics of Algorithmic PR: Social Media, Materiality and Post-Hegemonic Power

This is a summary of an embryonic paper accepted as a poster for the EUPRERA Annual Conference 2014: Communication Ethics in a Connected World, LASCO Laboratory, Université Catholique de Louvain, Brussels, Belgium, September 11th-13th.

This year’s call for papers poses the question: what place does ethics have in today’s communication practice? Moreover, it asks what is the relation between ethical issues, power and rhetorical construction of communication and discourse? This paper will attempt to address these questions by proposing a new conceptual framework for PR: one rooted in today’s digitally mediated landscape but, significantly, challenging existing notions of power and the rhetorical nature of contemporary communications.

This post provides a short overview of this new framework, outline some examples observed through ethnographic study of contemporary practice and consider briefly the ethical implications of this.

First and foremost, this paper offers a critical interpretation of public relations, consistent with the emerging field of critical and cultural scholarship in PR characterised as the “socio-cultural ‘turn'” (Edwards and Hodges, 2011: 1). However, I suggest that to adequately examine questions of power and ethics in a contemporary communications environment it is necessary to adopt a neo-materialist position. For instance, Coole and Frost (2010), argue that from a neo-materialist perspective “the more textual approaches associated with the so-called cultural turn are increasingly being deemed inadequate for understanding contemporary society” (Coole and Frost 2010, 2-3).

As a theoretical framework neo-materialist perspectives build on earlier developments in science and philosophy to turn ontological attention to the material – that is physical elements – constituting the world around us as well as the purely phenomenological. Neo-materialism represents a resurgence in the centrality and validity of matter – understood as “a commitment to the mind-independent existence of reality” (DeLanda 2006,1) – in contemporary society and refocuses analysis on the material’s inter-relation with the hitherto dominant analyses of language and representation.

In such a context, society and its constituent parts can be scrutinized through theoretical lenses such as the ‘critical sociology’ of Bruno Latour (2005), the Foucauldian ‘dispositif’ (Foucault 1977) and Delueze and Guattari (1987) and Delanda’s (2006) semio-material ‘assemblages’. Applying such approaches to strategic communication and public relations we can begin to see how dominant analyses of rhetoric, representation and discourse must be expanded to incorporate and account for the hitherto unseen material components of communication, such as technological infrastructure, computer software and bodily physicality of contemporary media practice (Manovich 1999; Latour 2005; Gillespie, Boczkowski and Foot 2014). The paper proceeds by focusing attention on the increasingly computational nature of contemporary PR, in particular the role of algorithms in influencing and controlling strategic communications (Manovich 2012; Lash 2007).

Accepting such a conceptual framework of contemporary communication grounded in material-semiotic terms, I propose developing the notion of ‘algorithmic PR’ – that is a recognition that PR practice has computational software as an integral and central part. Arguably few – if any – analyses of contemporary PR have addressed this dimension of strategic communication in the digital media environment. Such a gap requires identifying and exploring to fully understand the critical and ethical implications.

 

Identifying Examples of Algorithmic PR

The first step in this process is to identify and substantiate empirical examples of what an be understood as ‘algorithmic PR’. The following are two tentative scenarios where algorithmic PR can be discerned. Data for these examples has been gathered through ethnographic observation and interviews conducted with a number of UK-based, international communication agencies.

Scenario 1 –  Global Brand Crisis Strategy

In this scenario, a global brand had been targeted by an activist group primarily via its Facebook page. The group had initiated its attack strategically on a Sunday evening when the brand’s social media managers were not actively monitoring the Page. By Monday morning the Page was filled with anti-brand messages and calls for the brand to intervene to stop a wider international situation.

At this point, the activists’ communication had only had an impact on the brand’s Facebook presence and a meeting with the brand’s corporate communications team and social media agency crisis team was convened to judge the most appropriate response.
According to best practice, it was agreed that a response to activists’ concerns would be developed and posted on the brand’s Facebook Page. As the crisis was largely contained within Facebook it was agreed that no further external crisis communications activity was required.

Before the statement could be issued, however, claims started to appear on Facebook that the brand was censoring comments made by activists. In an era of social media, where transparency and openness are paramount (and enshrined in the brand’s own organisational policies for participation in social media) this was a serious development. A rapid investigation by the agency and client was initiated to establish whether or who was deleting user comments.

While the investigation was underway a high-profile media blogger had picked up on the crisis and criticism of the brand’s alleged ‘censorship’ of the Facebook debate and published a scathing article criticising the brand for adopting an anti-democratic approach to online debate. This initial post was picked up by other bloggers and subsequently traditional media and shared widely. This directed much greater attention and scrutiny on to the issue, exacerbating the organisational crisis for the brand.

After some research it was confirmed that no individual employee of the brand or agency was responsible for deleting activists’ posts and comments. In fact, the content was being ‘censored’ by Facebook’s built-in ‘auto-moderation’ functionality without knowledge of the agency or brand. This algorithmic tool detects profanity and other pre-determined ‘keywords’ appearing on the page and automatically ‘holds’ the comments for approval or deletion. While it can be argued that such a function is beneficial in helping brands and organisations from publishing offensive content, it is notable here that Facebook’s algorithm was responsible for exerting non-human agency to censor online discourse. This ‘unseen’ and material aspect of the communication process at work on the brand’s Facebook Page subsequently damaged the brand’s reputation to a greater extent that the original went. As a result it catalysed the spread of awareness of a critical issue and triggered a much wider crisis for the organisation.

Scenario 2 – Non-representational Communications Strategy

Discussion with a ‘digital reputation manager’ from an international PR agency revealed that a common strategy to help improve the public perception of an organisation with a poor reputation or public record would involve the targeting of Google results page. This approach was chosen due to the central importance of search engine results pages (SERPS) (and Google’s market dominance in particular) in shaping public awareness and perceptions of an organisations. Studies indicate, for example, that the first page of Google results generate 94% of clicks and the top result responsible for a third of all clicks.

Taking advantage of this situation, the digital reputation management activity of the agency would focus on developing a strategy that aimed to push negative, damaging or undesirable content off Google’s initial SERPs. If possible, this content would be replaced with positive – or more usually ‘non-negative’ content. This was achieved by studying (and, to an extent, second guessing) Google’s PageRank algorithm. Anecdotally if you can’t something on Google, it doesn’t exist.

The PageRank algorithm is Google’s the proprietary and commercially sensitive algorithm that determines where websites and content are displayed in Google’s results based on a given search enquiry. While the PageRank algorithm is a tightly guarded secret, a number of tactics can be deployed to ‘game’ or ‘optimise’ the results (see Philips and Young, 2009: 24).

These tactics and – more broadly – the strategic approach I would term ‘non-representational communication’. That is, it is a communications strategy that privileges as its outcome, the material effect of influencing an algorithmic, computational response, rather than exert a representational or phenomenological response by a human. The strategy creates and disseminates content that is designed to interact with and generate a positive outcome in Google search results solely as the desired outcome. This is in opposition to representational communication content which is designed to establish a mutual or communicative understanding based on a textual or visual interaction.

To illustrate this point, a representational approach to communication might be premised on producing information that represents the organisation’s position and adopted phenomena to elicit an emotional or informational response by the individual receiver. In a non-representational approach the individual receiver is only a secondary consideration. Rather the ‘message’ is created purely to trigger a positive (material) response by Google’s material algorithm.

Other examples
There are other examples of Algorithmic PR at work which are currently being gathered and analysed as part of this project. In many instances, the same non-representational strategies are being adopted by PR practitioners and communicators either intentionally or by proxy through the increased adoption of digital technology, such as Facebook, Twitter and other communications activities and processes requiring computational interaction – for example social media monitoring, social media measurement and brand or issue analysis using big data all rely on largely unseen – or at best – overlooked algorithmic or computational processes.

Algorithmic PR, Post-Hegemonic Power and Ethics
Having provided some examples of Algorithmic PR it is now important to explore some of its theoretical and practical implications. In keeping with the primary concerns of the conference this analysis will focus on assessing the usefulness of existing notions of ethics and power within PR.

Existing analyses of PR’s communicative power have tended to focus on the hegemonic potential of strategic communications’ rhetorical and discursive dimensions. That is, the ways in which discourse (as imagery and text) are created to represent specific ideologies and then seek to normalize them through repeated circulation or co-option or rejection of opposing ideologies.
However, algorithmic PR functions at a material level within the algorithmic software embedded in the technological infrastructure of communications tools. As a result power operates prior to and within the formation of conventional hegemonic representations. This is a notion Lash (2007) terms ‘post-hegemonic power’.

Interpreted as such, post-hegemonic power isn’t constructed from the outside and imposed on people through representative communication created and crafted by human agency, but rather generated from within through non-representative, material and non-human elements present in digital platforms (Beer 2009). This raises potent questions for scholarly understandings of PR, ethics and power.

For example, the notion of ethics and ethical values are traditionally understood as socially constructed and thus rooted in the individual agency – and processes – of practitioners. Read from  the ontological perspective of neo-materialism and post-hegemonic reading of power, ethical concerns need rethinking as they immediately become entangled in complex semio-material assemblages constituted through human and non-human agency – that is, through socially constructed practitioner decisions (i.e. doing the right thing’) and the augmentation by computational behaviours embedded in algorithms.

Take for example, the crisis case study discussed above. Practitioners were attempting to operate ethically by engaging with critical activists whereas Facebook’s algorithm had other “intentions’.’ How can practitioners, tasked with applying and adhering to ethical standards, such as consciously “protect[ing] and advance[ing] the free flow of accurate and truthful information” and “foster[ing] informed decision making through open communication” (PRSA n.d.), ensure that this occurs when algorithms can now be seen to play such a central role.

The Facebook brand crisis case study cited above illustrates this point neatly. Practitioners, believing they are operating as openly and accurately as possible, find their actions undermined by unseen software embedded within the platform. In turn, this type of challenge raises additional questions about social media and communication ethics. Some practitioners and scholars superficially suggest that social media is fostering a corporate and social environment requiring increased transparency and openness – values broadly supportive of ethical communication (Wright and Hinson 2008; Bertot, Jaeger and Grimes 2010) . Recognising algorithmic PR’s potential to exert hidden post-hegemonic power challenges this ‘ethical turn’ of social media.

More significantly from a macro-perspective, given the increased significance of non-human agency in shaping communicative power and PR practice, where can power be located? Where does it operate and who or what can influence or shape it?

The ‘flat’ ontological status of neo-materialism prevents ascribing a permanent reading of post-hegemonic power as either rooted a priori in human or non-human agency. Rather, its presence lies dispersed within complex and continual interactions of the material and semiotic. For example, when humans interact with algorithms, both the algorithmic software and human ‘software’ (that is, the brain’s computational capacity) both respond to each other and adapt accordingly (Manovich 1999).

We can see such a scenario in the context of algorithmic PR when, for example, a communications manager will craft content designed to respond optimally with Facebook’s Edgerank algorithm and in return the Edgerank algorithm will respond in a situationally specific way to make the communicator’s content or message more or less visible to the Facebook user network. This is a concern faced and addressed by many PR practitioners on a daily basis when planning messages and content for distribution through Facebook.

In such material-semiotic assemblages (Deleuze and Guattari 1987; DeLanda 2006) tracing agency becomes a much more complex and multifaceted task that PR scholarship must recognise and take steps to address.

References:

Beer, D. (2009) ‘Power Through the Algorithm?’ New Media Society. 11: 985.

Bertot, J. C., Jaeger, P. T. and Grimes, J. M. (2010) ‘Using ICTs to create a culture of transparency: E-government and social media as openness and anti-corruption tools for societies’ Government Information Quarterly. 27, 3: 264-271.

DeLanda, M. (2006) A New Philosophy of Society: Assemblage Theory and Social Complexity. London and New York: Continuum.

Deleuze, G. and Guattari, F. (1987) A Thousand Plateaus. Minneapolis: University of Minnesota Press.

Edwards, L. and Hodges, C. (2011) Public Relations, Society & Culture: Theoretical and Empirical Explorations. London: Routledge.

Foucualt, M. (1977) ‘The Confession of the Flesh’ in Gordon, C. (1980) (ed.) Power/Knowledge Selected Interviews and Other Writings. 194-228.

Gillespie, T., Boczkowski, P. J. and Foot, K. A. (2014) Media Technologies: Essays on Communication, Materiality and Society. Cambridge, MA. And London: MIT Press.

Lash, S. (2007) ‘Power after Hegemony: Cultural Studies in Mutation?’. Theory, Culture, Society. 24: 55.

Latour, B. (2005). Reassembling the Social: An Introduction to Actor Network Theory. Oxford: Oxford University Press.

Manovich, L. (1999) ‘Database as Symbolic Form’. Convergence. 5: 80.

Manovich, L. (2012) Data Stream, Database, Timeline. Software Studies Initiative blog. Online. Available at: http://lab.softwarestudies.com/2012/10/data-stream-database-timeline-new.html [Accessed 23rd February 2014)

Phillips, D. and Young, P. (2009) Online Public Relations. London: Kogan Page.

PRSA. (n.d.) ‘Ethical Guidance for Public Relations Practitioners’. PRSA website. Online. Available at: http://www.prsa.org/aboutprsa/ethics/#.UxBj615Rndh [Accessed 23rd February 2014)

Wright, D. K. and Hinson, M. (2008) Examining the Increasing Impact of Social Media on the Public Relations Practice. Institute for Public Relations. Online. Available at: http://www.clayton.k12.mo.us/cms/lib/MO01000419/Centricity/Domain/2/NSPRA/SM_ImpactOf.pdf [Accessed 7th September 2014]