4
Nanotargeting and Automation of Political Discourse
This chapter seeks to detail what we call the “automation of political discourse”, a concept developed from three main elements: first, nanotargeting, which constitutes the action of automated political discourse1; second, algorithmic governmentality, which constitutes its method; and third, the public space of “communicative capitalism” (Dean 2005), which constitutes its emerging context.
4.1. On nanotargeting
4.1.1. Segmentation
The notion of target audiences has been used for a long time and is common in marketing. However, this segmentation is accentuated by communication technologies. In terms of supply, the public has access to an exponential choice of media channels and opportunities for interaction with them. This fragmentation makes the planning of advertising campaigns more complex, while making it possible to select the audiences targeted by these same campaigns: it is the transition from broadcasting (broad) to narrowcasting (reduced) (Du Plessis 2012).
These new technological possibilities facilitate an extreme segmentation of the market, which thus becomes an alternative to product differentiation (Baines et al. 2003). In short, it is no longer necessary to offer a product – or a message – that stands out from the others: it becomes necessary to offer the message to the right person, when they are in a good state of mind to receive it. Segmentation mainly operates, in this sense, in what could be called a “transfer” of production. The role of the advertiser is no longer primarily to sell a product to a consumer, but rather to produce a consumer for a product: we identify exactly when, where and in what state of mind this consumer will be receptive to a certain message. This production is adjusted to market needs, constantly creating new groups of consumers, according to new specific characteristics in time and space (Zwick and Knott 2009). However, electoral politics is no longer very different from the market in its purely economic sense. The political marketplace is not very distinct from the market for products and services. Politicians present themselves as brands (Guzmán et al. 2015), and most of the scientific literature on segmentation includes the political marketplace (Baines et al. 2003). The place of segmentation in this market was highlighted by the unified field in 1990 (O’Shaughnessy 1993) and concerns about this segmentation soon followed: Elihu Katz was already concerned, in 1996, about the consequences of electoral segmentation in the democratic public space (Katz 1996).
The segmentation marketing dynamics described are therefore at work in the political sphere, more specifically in the electoral sphere. Such techniques became popular as early as the 19th Century, when firms kept lists of personal information with the collaboration of postal services and newspapers (Couldry and Turow 2014). The same principle has been applied to large-scale data collection and processing: different forms already existed, since the end of the 19th Century (Semetko and Tworzecki 2017). The novelty lies rather in the refinement. Magin et al. (2017) classify the types of traditional political campaigns, in terms of communication, into three categories: directed at supporters, directed at the general electorate or directed at groups. In addition, there is now a fourth type made possible by recent technical capabilities, namely “individual campaigns based on personal data offered by commercial companies” (Magin et al. 2017, p. 1701). These campaigns are effectively established on political microtargeting.
4.1.2. Microtargeting
Microtargeting, the act of sending a particular message to a specific and chosen group of individuals without their necessarily being aware of it (Faizullabhoy and Korolova 2018), is not a new technique and has been generating enthusiasm in the field of political communication for at least two decades (Milbank 1999). Microtargeting has been used in commercial advertising for a long time and its method has been devised by marketing researchers for even longer (Sivadas et al. 1998). Its use in political communication, on the other hand, has more fundamental implications than in a consumerist sales context. Although the use is similar, persuasion techniques are an integral part of the democratic process and, for this reason, they must be observed more closely (Bay 2017). The ethics of a targeted message or algorithm that seeks to influence the political process are not the same as those of an algorithm that seeks to sell a product (Bay 2017).
The increasing integration of microtargeting into political communication has therefore raised several concerns in the academic community, particularly with regard to the impact of voter segmentation on democratic regimes (Couldry and Turow 2014; Guzmán et al. 2015; Shorey and Howard 2016). The Cambridge Analytica case, in the spring of 2018, particularly exacerbated these concerns.
4.1.3. Nanotargeting
The term nanotargeting was first used in the field of political communication in 2009 by Josh Koster, campaign consultant and “new media” specialist (Harvey 2014). It is put forward with strong enthusiasm in marketing studies where we speak, with some reservation, of a “promised land” (Jovanovic 2014, p. 415). However, the concept of nanotargeting is still evolving. While Koster applied it to the segmentation of “niches”, which concerned groups of individuals, it is now associated with microtargeting directed at a single individual or selected individuals rather than a group (Barbu 2014). Kerpen (2011) was the first to give it this meaning – at least in marketing – when he applied the term to his successful attempt to create a Facebook advert that only his spouse could see. Alexander Nix, CEO of Cambridge Analytica, also boasted that he was able to target specific individuals using the techniques developed by his company (Mavriki and Karyda 2017). In our opinion, nanotargeting differs from microtargeting in three main ways.
4.1.3.1. Affects and psychometric analysis
The notion of affects has been particularly prominent and developed in research in communication with the development of digital social networks. The concepts such as affective networks (Dean 2010), affective economics (Andrejevic 2011), affective news (Young and Soroka 2012) and affective public (Papacharissi 2014) have been designed to try to capture new communication dynamics. Dean and Paparachissi explain that digital communication networks are mainly built around the circulation of affects: affective networks that do not constitute real communities, but feelings of community (Dean 2010, p. 22), and public affective networks organized around structures of feelings (Papacharissi 2016, p. 320).
While microtargeting is mainly built on data that relate to purchasing habits, income, social class, location, employment or community membership, nanotargeting adds emotions, feelings and behaviors to this list. Massive data mining on the Internet introduces techniques of “sentiment analysis” and “opinion mining”, terms that appeared in the academic field at the very beginning of the 21st Century (Pang and Lee 2008). These techniques are used to quickly identify a person’s psychological state in order to offer the most appropriate message. It is the understanding of an individual’s state of mind that allows advertisers to determine exactly what media should be used, what time and technique should be chosen and what message should be offered to not only receive the attention of the targeted individual, but, most importantly, be able to predict their reaction to the message (Du Plessis 2012). Psychometric analysis, based on data collected through sentiment analysis and mood analysis, is, in our opinion, a central feature of the automation of political discourse. First, because it marks a break with the types of data that could be collected before digital data mining; second, because it is part of this quest for immediacy that defines automation; and third, because it significantly changes the “classic” segmentation of markets. In short, psychometric analysis techniques offer added value to traditional market segmentation by integrating data that were previously difficult to obtain, particularly in real time, through sentiment analysis and mood analysis. It is the attempt by some digital communication actors to build electoral markets not only through the collection of demographic information, but also through “comprehensive, real-time, trustworthy databases of Internet users’ behavior and conversations” (Andrejevic 2011, p. 604).
4.1.3.2. Instantaneity
The second main distinction of nanotargeting, in our view, is instantaneity: firstly, psychometric data, unlike demographic data, changes rapidly; secondly, generated message from collected data has to be sent at the precise moment when the targeted individual is in the best state of mind to receive the message. More so than ever, we are trying to anticipate a behavior that would be triggered by the reception of a certain message and, in fact, produce it.
4.1.3.3. Individuality
The last distinction, individuality, stems from the first two. Microtargeting was aimed at groups of people, sometimes quite small, but always groups. Nanotargeting, by adding sentiment building and real-time action possibilities carried by the Internet, reduces the message to the individual level. The targeted individual receives a specific message, built in part by the data they have produced themself, and sent at the moment when they are in a specific state of mind: in this sense, each message is unique.
4.2. On algorithmic governance
The concept of algorithmic governance broadly refers to “a certain type of (a)normative or (a)political rationality founded on the automated collection, aggregation and analysis of Big Data so as to model, anticipate and pre-emptively affect possible behaviors” (Rouvroy and Berns 2013).
Algorithmic governmentality takes place in three stages: first, the collection by data monitoring of huge amounts of data, otherwise known as Big Data. Big Data can be defined by the construction of large datasets from information that is collected on several people, by several devices, and that can be linked to each other to derive correlations and allow research (Shorey and Howard 2016). These data are generally anonymized and decontextualized, i.e. cut off from any relationship with the subject from whom they were produced. Second, these data are processed and knowledge is produced from them in an automated way: algorithms perform this task. Finally, thanks to this knowledge, we can take action on behavior (Rouvroy and Berns 2013).
Algorithmic governmentality introduces several notions. First of all, there is the capture of attention. Algorithmic governmentality is based on efficiency, instantaneity and real time. This capture takes place in an “attention economy”, where we seek to capture affects rather than reasoning (Harsin 2015). Emanating in part from marketing and economic logic, we are witnessing a “systematic capture of any fragment of human attention available for the benefit of private interests (the economy of attention), rather than for the benefit of democratic debate and the general interest” (Rouvroy and Berns 2013, p. 167).
Then comes the action on the behavior. A call to deliberation and reason forms reflexivity; a call to affect causes a reaction. Algorithmic governmentality does not promote dialog, debate or even reflection: the injunction to efficiency rather imposes an action on the behavior. Actions are predicted, according to data, and a signal is sent to provoke a certain reaction. We want to act on what has not yet been done, but on what has been predicted, i.e. to build the real: the “target of algorithmic governmentality”, explains Rouvroy, is “this unrealized part of the future” (Rouvroy and Stiegler 2015, p. 119).
These trends result in the disappearance of the subject. The algorithmic governmentality, in fact, “produces no subjectification, it bypasses and avoids reflexive human subjects” (Rouvroy and Berns 2013, p. 174). The capture of affects and then action on behaviors shapes models that are “supra-individual”, i.e. there is never a call to the subject himself and to rationality. This leads to a depoliticization that “does not give rise to or take any active, consisting, reflexive statistical subject” (Rouvroy and Berns 2013, p. 180)2.
Finally, there is a “disappearance of the common experience” (Rouvroy and Berns 2013, p. 167). A social space automated by algorithms, optimized, without reflexivity, simply without a subject, leads to a public space atrophied by its potentially subversive meaning. The emphasis on affects leads to a potential radicalization of opinions and the loss of common narratives and norms. Paradoxically, it is the constitution of a state of citizenship that is hyperindividualized, but deprived of its political capacity.
4.3. Public space and communicative capitalism
Political discourse, of course, is an integral part of the public space. In Habermas’ work, Cossette (1987) explains that public space is above all a normative proposition: the theorist shares with us a “fiction” made possible by the conditions of the modern state, an unfinished reality. Habermas’ model of public space is thus conceived “as the bearer of a utopian possibility […] an ideal set in a critical context” (Cossette 1987, p. 11).
Habermas’ public space is built on an institutional and liberal conception as well as on the fundamental role of free and rational deliberation (Goodsell 2003). For him, institutions must build a deliberative space that respects certain standards, including the equality of the people who take part and universal accessibility. For Castoriadis and Arendt, public space also deals with the issues of collective interest: this space is in fact the condition for the possibility of politics (politique)3 (Straume 2012). The public space is not maintained by institutions, but by political action: for Arendt, it is “active life” and for Castoriadis, it is the “ontology of being” (Straume 2012). The model of representative democracies, in this sense, is rejected by both authors, because it is based on a procedural vision of democracy in which there is no real participation, and therefore neither equality nor effective freedom (Poirier 2009). In this sense, the public space, according to Castoriadis (1997), is institutional in the sense of society and political action.
For the three authors, the notion of public space is also based on a tension with the private sphere. Different areas of life are classified, either in the public sphere, in the private sphere or somewhere in between. The economic sphere, in particular, is classified by Habermas outside the public sphere (Hohendahl 1992). Castoriadis and Arendt go in the same direction: Castoriadis (1997) separates society into three main spheres: the oikos, the family, the private sphere; the agora, the market place, the private/public sphere; and the ekklēsia, the public sphere, the place of political power. Arendt (1958), on the contrary, argues its distinction, particularly through the concept of active living, and presents the private sphere as the space of necessity and the public sphere as the space of the construction of freedom.
However, in the observation of each individual, there is a disappearance of these borders, a “colonization” of the public sphere by the private sphere. The capitalist dynamic aimed at always favoring the process of commodification, as Castoriadis explains (in Poirier 2009), leads to a process of privatization of the public space. For Arendt (in Goodsell 2003), the distinction between the public and private sphere is becoming blurred. This colonization of the public sphere, according to Arendt, has transformed it into a “pseudo-space” of interactions where individuals no longer act but behave, recalling the action of algorithmic governmentality described by Rouvroy and Berns.
The notion of public space allows us to highlight the criticism developed by Dean (2003) with the concept of communicational capitalism, which in several respects parallels that of algorithmic governmentality. These two concepts are, according to our understanding, closely linked within speech automation. These three elements mainly intersect.
First, the function of the message and its relationship to the individual. In Habermas’ model, explains Dean (2005), the message has a function based on its understanding, i.e. its “value” is measured according to whether it has been understood in the right way: the content of the message is essential for its use. This is not the case in communicative capitalism. The content of the message does not matter: its circulation, rather, verifies its success. “A contribution [message] need not be understood; it need only be repeated, reproduced, forwarded” (Dean 2005, p. 59). At first glance, this definition may seem contradictory to the idea of microtargeting, where a message has very specific characteristics and does not necessarily have to circulate widely. The essential element is the action of the message. The content of the message, from a reflexive point of view, is also not important in a microtargeting campaign. Its success is confirmed by its action on behavior, as Rouvroy (Rouvroy and Berns 2013) explains, by the effectiveness of these signals. In this sense, the individual targeted is no more important either, than the importance of certain correlations previously plotted by the algorithm – such as his potentiality to react to a certain signal – and his ability, perhaps, to circulate the message in a certain network. We want to influence our behavior or even our place (Dean 2005).
Second, the central notion of communicational capitalism of technological fetishism refers to Rouvroy’s thinking in several ways. For Dean, technological fetishism is the idea that complex political elements could be condensed into a single problem, and therefore a single solution, technological of course. This is what the algorithm actually claims to do: in a neutral and objective way, it absorbs huge amounts of data on complex subjects and derives correlations that make it possible to analyze them, but not to understand them. This is also a loss of reflexivity.
Third, the notion of depoliticization, essential to algorithmic governmentality, is also essential in the concept of communicative capitalism. Political fetishism, through its condensing action and technical “solution”, depoliticizes problems. Algorithmic governmentality operates the same process, depoliticizing decisions: political reflection is useless in a context where data provide us with an accurate understanding of reality, where figures speak for themselves.
4.4. On the automation of political discourse
It is at the intersection of the theories presented that we find the global portrait of the automation of political discourse. Bernard Stiegler (in Rouvroy and Stiegler 2015) states that, unlike the process of reason, which is synthetic, automation is an analytical process. Automated speech occurs through data analysis, algorithmically processed and detached from any human reflexivity and arrangement. First, it creates elective “markets” (Harsin 2015), in which it instigates individuals not to appeal to their reason and deliberative capacities, but to capture their attention and try to influence their behavior, in real time: the technique operates both hyperindividualization and “desubjectivation” (Dufour 2008; Mondoux 2011; Rouvroy and Stiegler 2015). This entire process of capturing data and returning signals from the same data is circular and operates in fundamentally depoliticizing market logics, where the message, as long as it has the desired effect and becomes sufficiently large in a certain network, ultimately does not matter (Dean 2005). This same process, therefore, contributes to the formation of a reality that is not constructed, real, neutral and objective, but which in fact carries ideology and is constitutive: first of all, a political discourse that becomes apolitical, without reflexivity, personalized, but at the same time without subject; then a public space colonized by “an hypertrophied private sphere” (Rouvroy and Berns 2013, p. 167) where the common and democratic experience is disappearing.
4.5. References
Andrejevic, M. (2011). The work that affective economics does. Cultural Studies, 25(4–5), 604–620.
Arendt, H. (1958). The Human Condition. The University of Chicago Press, Chicago.
Baines, P.R., Worcester, R.M., Jarrett, D., and Mortimore, R. (2003). Market segmentation and product differentiation in political campaigns: A technical feature perspective. Journal of Marketing Management, 19(1–2), 225–249.
Barbu, O. (2014). Advertising, microtargeting and social media. Procedia, Social and Behavioral Sciences, 163, 44–49.
Bay, M. (2017). The Ethics of Psychometrics in Social Media: A Rawlsian Approach. Research Report. University of California, Los Angeles.
Bennett, W.L. and Manheim, J.B. (2006). The one-step flow of communication. The Annals of the American Academy of Political and Social Science, 608(1), 213–232.
Castoriadis, C. (1997). Democracy as procedure and democracy as regime. Constellations, 4(1), 1–18.
Cossette, J.-L. (1987). L’espace public chez Habermas la légitimité à l’aune des raisons. McGill-Queen’s University Press, Montreal.
Couldry, N. and Turow, J. (2014). Advertising, big data and the clearance of the public realm: Marketers’ new approaches to the content subsidy. International Journal of Communication, 8, 1710–1726.
Dean, J. (2005). Communicative capitalism: Circulation and the foreclosure of politics. Cultural Politics, 1(1), 51–74.
Dean, J. (2010). Affective Networks. MediaTropes, 2(2), 19–44.
Du Plessis, E. (2012). 2020, The Brand Feeling. Wharton University of Pennsylvania, Philadelphia.
Dufour, D.-R. (2008). Vivre en troupeau en se pensant libres. Le Monde Diplomatique, 646, 20–21.
Faizullabhoy, I. and Korolova, A. (2018). Facebook’s advertising platform: New attack vectors and the need for interventions. Workshop on Technology and Consumer Protection, 6.
Goodsell, C.T. (2003). The concept of public space and its democratic manifestations. The American Review of Public Administration, 33(4), 361–383.
Guzmán, F., Paswan, A.K., and Van Steenburg, E. (2015). Self-referencing and political candidate brands: A congruency perspective. Journal of Political Marketing, 14(1–2), 175–199.
Harsin, J. (2015). Regimes of posttruth, postpolitics, and attention economies. Communication, Culture & Critique, 8(2), 327–333.
Harvey, K. (2014), Encyclopedia of Social Media and Politics. Sage Publications, Thousand Oaks.
Hohendahl, P.U. (1992). The public sphere: Models and boundaries. In Habermas and the Public Sphere, Calhoun, C.J. (ed.). MIT Press, Cambridge, 73–98.
Jovanovic, D. (2014). Age of hyper, micro and nanotargeting. In 5th International Scientific Conference, Economic and Social Development. Varaždin Development and Entrepreneurship Agency, Varaždin, 408–416.
Katz, E. (1996). And deliver us from segmentation. The Annals of the American Academy of Political and Social Science, 546(1), 22–33.
Kerpen, D. (2011). Likeable Social Media: How to Delight Your Customers, Create an Irresistible Brand, and Be Generally Amazing on Facebook (& Other Social Networks). McGraw-Hill Professional, New York.
Magin, M., Podschuweit, N., Haßler, J., and Russmann, U. (2017). Campaigning in the fourth age of political communication. A multi-method study on the use of Facebook by German and Austrian parties in the 2013 national election campaigns. Information, Communication & Society, 20(11), 1698–1719.
Mavriki, P. and Karyda, M. (2017). Using personalization technologies for political purposes: Privacy implications. In 7th International Conference E-Democracy: Privacy-Preserving, Secure, Intelligent E-Government Services, Katsikas, S. and Zorkadis, V. (eds). Springer, 33–46.
Milbank, D. (1999). Virtual politics. The New Republic. Available at: https://newrepublic.com/article/90746/virtual-politics.
Mondoux, A. (2011). Identité numérique et surveillance. Les Cahiers du Numérique, 7(1), 49–59.
Neuman, W.R. and Guggenheim, L. (2011). The evolution of media effects theory: A six-stage model of cumulative research. Communication Theory, 21(2), 169–196.
O’Shaughnessy, N.J. (1993). America’s political market. European Journal of Marketing, 21(4), 60–66.
Pang, B. and Lee, L. (2008). Opinion mining and sentiment analysis. Foundations and Trends in Information Retrieval, 2(1–2), 1–135.
Papacharissi, Z. (2014). Affective Publics: Sentiment, Technology, and Politics. Oxford University Press, New York.
Papacharissi, Z. (2016). Affective publics and structures of storytelling: Sentiment, events and mediality. Information Communication & Society, 19(3), 307–324.
Poirier, N. (2009). Espace public et émancipation chez Castoriadis. Revue du MAUSS, 34, 368–384.
Rouvroy, A. and Berns, T. (2013). Gouvernementalité algorithmique et perspectives d’émancipation. Réseaux, 177, 163–196.
Rouvroy, A. and Stiegler, B. (2015). Le régime de vérité numérique. Socio, 4, 113–140.
Semetko, H.A. and Tworzecki, H. (2017). Campaign strategies, media, and voters: The fourth era of political communication. In The Routledge Handbook of Elections, Voting Behavior and Public Opinion, Fisher, J. et al. (eds). Routledge, New York.
Shorey, S. and Howard, P.N. (2016). Automation, algorithms, and politics. Automation, Big Data and politics: A research review. International Journal of Communication, 10, 5032–5055.
Sivadas, E., Grewal, R., and Kellaris, J. (1998). The internet as a micro marketing tool: Targeting consumers through preferences revealed in music newsgroup usage. Journal of Business Research, 41(3), 179–186.
Straume, I.S. (2012). A common world? Arendt, Castoriadis and political creation. European Journal of Social Theory, 15(3), 367–383.
Wilson, J. (2008). Political discourse. In The Handbook of Discourse Analysis, 2nd edition, Schiffrin, D., Tannen, D., and Hamilton, H.E. (eds). John Wiley & Sons, New York, 398–415.
Young, L. and Soroka, S. (2012). Affective news: the automated coding of sentiment in political texts. Political Communication, 29(2), 205–231.
Zwick, D. and Knott, J.D. (2009). Manufacturing customers: The database as new means of production. Journal of Consumer Culture, 9(2), 221–247.
Chapter written by Samuel COSSETTE.
- 1 The notion of political discourse is rather a large one: the two terms are polysemic. To discern what constitutes or does not constitute political discourse is an arduous task. In a literature review, Wilson (2008) notes that the notion sometimes refers to discourse that is explicitly political, and other times to any discourse related to a certain political context. It is important, however, Wilson notes, to avoid over-generalizing the notion as this could undermine the clarity of the analysis. In this sense, he proposes to offer, each time, a definition adapted to the objectives of the analysis (Wilson 2008, 398). In this chapter, much of which deals with the notion of nanotargeting, we will limit our understanding of the notion of political discourse to digital messages, although other elements of political discourse are affected, to varying degrees, by automation processes.
- 2 Analyses of new communication capabilities offered by objects such as algorithms or microtargeting methods often tend to limit the arrangement of subjects in their relationships with these objects. Some even put forward the idea of a return to a one-step flow of communication (Bennett and Manheim 2006; Neuman and Guggenheim 2011), where targeting and prediction methods would be so precise that they could go beyond the “decoding” step performed by the subject receiving the message. These hasty analyses risk recreating theoretical shifts from the past, the most famous being the “hypodermic syringe” theory, making the same epistemological error, that of considering an all-powerful device and ignoring the subject. Here, we analyze the tools of political discourse automation, the theoretical considerations it proposes and some of the risks we perceive in terms of political communication, but without denying its central place in these analyses: this will, however, need to be the subject of further research.
- 3 For Castoriadis and Arendt, the concept of politics is essential to the understanding of public space and its role. The two contrast le politique, which concerns rules, institutions, procedures and their applications, as well as the existence of a central power, and la politique, whose ultimate objective would be the infringement of freedom and the questioning of that power (Poirier 2009). It is “true politics” and “politics”, for Castoriadis; genuine politics, authentic politics and politics proper, for Arendt (Straume 2012).