Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

Ben Collier; Gemma Flynn; James Stewart; Daniel Thomas

doi:10.1177/20539517221078756

Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

Collier, Ben; Flynn, Gemma; Stewart, James; Thomas, Daniel 2022-02-24 00:00:00 We have identiﬁed an emerging tool being used by the UK government across a range of public bodies in the service of public policy - the online targeted advertising infrastructure and the practices, consultancy ﬁrms, and forms of expertise which have grown up around it. This reﬂects an intensiﬁcation and adaptation of a broader ‘behavioural turn’ in the gov- ernmentality of the UK state and the increasing sophistication of everyday government communications. Contemporary UK public policy is fusing with the powerful tools for behaviour change created by the platform economy. Operational data and associated systems of classiﬁcation and proﬁling from public bodies are being hybridised with traditional con- sumer marketing proﬁles and then ‘projected’ onto the classiﬁcation systems of the targeted advertising infrastructures. This is not simply a case of algorithms being used for sorting, surveilling, and scoring; rather this suggests that targeted interventions in the cultural and behavioural life of communities are now a core part of governmental power which is being algorithmically-driven, in combination with inﬂuencer networks, traditional forms of messaging, and frontline opera- tional practices. We map these uses and practices of what we describe as the ‘Surveillance Inﬂuence Infrastructure’, iden- tifying key ethical issues and implications which we believe have yet to be fully investigated or considered. What we ﬁnd particularly striking is the coming-together of two separate structures of power - the governmental turn to behaviourism and prevention on one hand, and the infrastructures of targeting and inﬂuence (and their complex tertiary markets) on the other. We theorise this as a move beyond ‘nudge’ or ‘behavioural science’ approaches, towards a programme which we term ‘inﬂuence government’. Keywords Targeted advertising, government, power, marketing, criminal justice, public policy This article serves to set out our initial, exploratory Introduction ﬁndings about the use of these techniques in the UK The practices of private sector advertising and marketing public sector, discussing some of the emerging ways have long existed in a mutual relationship with government in which public bodies are using what we term the - from wartime propaganda to public health messaging. In Surveillance Inﬂuence Infrastructure (SII), developed their contemporary forms, marketing practices have for targeted advertising, to facilitate public policy out- evolved substantially beyond postcode-based demographic comes through ‘behaviour change’ strategies. We ﬁrst targeting, supported by the proliferation of online advertis- set out relevant context, explaining the thinking ing infrastructures which allow continually-updated target- behind these approaches to public policy - approaches ing based on behaviour and online activity. Our empirical which apply behavioural science and ‘public health’ research shows that these advanced marketing techniques are now being incorporated into the business of government and law enforcement. Although the ‘algorithmic turn’ and University of Edinburgh, Edinburgh, Scotland University of Strathclyde, Glasgow, Scotland ‘behaviourist turn’ are both well-established within UK governance, their combination in an emerging set of prac- Corresponding author: tices represents a novel, powerful, and in some cases poten- Ben Collier, University of Edinburgh, Edinburgh, Scotland. tially concerning frontier of government policy. Email: collier@ed.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https:// creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as speciﬁed on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Big Data & Society logics to a wide range of social issues. We then discuss housing, and health at poor communities and communities the complex network of online services and infrastruc- of colour. State engagement with ‘traditional’ media (parti- tures which generate databases and algorithmic models cularly newspapers) was crucial to manufacturing broad that facilitate the targeting of adverts. We then identify social consent for these approaches through concerted (drawing from an analysis of publicly available docu- media campaigns and moral panics which transformed the ments) a range of examples of how contemporary tar- cultural environment - around muggings, ‘hoodies’, anti- geted advertising though the SII is being used in social behaviour, asylum seekers, or the ‘beneﬁt cheat’ practice in the UK public sector. We provide some (Hall et al., 1978). initial explorations of what this means for public The rise of a ‘risk’ model of social groups and issues was policy and the character of state governance more gen- reﬂected in a broader managerialisation of the business of erally. Our paper concludes with a critical reﬂection government across the 1980s, becoming known as New on emergent ethical issues and some of the areas Public Management (Barberis, 1998). This managerialisa- which might conceivably beneﬁt from these techniques tion entailed the increasing collection of data and categor- -the ‘potential futures’ of surveillance-targeted behav- isation of publics (through patient databases, offender ioural messaging. matrices, and other systems) with a view to managing the public through metrics, outcomes, and measurable pro- cesses. Many of these metrics relate as much to the Governmental modernity: From social targets and outcomes used to monitor the intermediary marketing to responsibilisation ﬁrms granted public contracts under the neoliberal model To understand these developments it is useful to situate as to the public themselves. This also included the appropri- these practices within the history of government in the ation of the tools, standardised techniques, and models of UK. Governmental modernity, at the heart of much of private industry, including market segmentation and Foucault’s scholarship, is generally linked to the ascend- postcode-based marketing using the Mosaic or Acorn clas- ancy of a distinct rationality of power. This rationality con- siﬁcations. Foucauldian scholarship generally makes sense tends that, as the empirical study of human societies of this through the lens of biopolitics - the protective power progresses, their essential dynamics will be understood to and subjectifying force which the state exerts by gathering the extent that they can eventually be designed to function data about its citizens, with its companion, necropolitics - in more beneﬁcial and harmonious ways (Foucault, 1982; where the state represses through dramatically overextend- Garland, 1997). ing or tactically withdrawing its gaze (Mbembe, 2008). In the 1970s and 1980s, the adoption of the ideals of Modernist ideas of transforming society had given way to modern government - that the state could shape important forms of intervention and shaping based around surveil- aspects of experience and personhood - continued to lance and the management of ever more diffuse and globa- develop. Within public health, the rise of a ‘social market- lised forms of risk - maintaining the existing social order ing’ approach in this period saw commercial marketing and mitigating its worst effects. techniques deployed using the limited targeting available As the modes of governance in the UK and US shifted in through the mass media to transmit public health messages the neoliberal era, the state’s role became less the centra- (Atkin and Wallack, 1990; Lupton, 1995, p110; Grier and lised design of secure societies and prosocial citizens, and Bryant, 2005). These were designed by partnerships of ad more concerned with the responsibilisation (Garland, agencies, community groups, government agencies and 2001) of private citizens and businesses, who could pur- academics (Atkin and Wallack, 1990). While health lit- chase services from private sector providers. This was eracy, empowerment, social norm adaptation and collective coupled with the continuing ascendancy of repressive action may have been the aim of some of these pro- force targeted against communities deemed a risk to main- grammes, they were often designed with a model of the stream society - in health, crime, welfare, and other areas of ‘heroic’ individual able to change their own behaviour social policy. The role of the state in social design in this while ignoring more structural features (Katz et al., 2000; ‘marketised’ mode, though distanced and softened, was Wallack, 2002). Despite notable successes, these pro- not necessarily diminished; in idealised neoliberal societies grammes suffered from problems such as poor design, inad- the state takes the form of a ‘steering’, not a ‘rowing’ force, equate research and poor stakeholder coordination (Cook in which the delivery of government policy is devolved to et al., 2021). the private sector and civil society, but the state still ulti- Another approach sought to build public consensus mately sets the goals and agenda (Crawford, 2006). This around legitimacy for policy action. This was accompanied privatisation coincided with an increasing perception throughout the second half of the 20th Century by an expan- within government of the apparent intractability of social sion of the reach, severity, and targeting of the repressive issues, leaving only individuals, able to protect themselves, force of the state, particularly through the targeting of but with no sense of a wider possibility of collective social violent policing and punitive policies within education, change (Loader, 2006). Collier et al. 3 reﬂect capitalist, entrepreneurial, and ‘resilient’ models of Behaviourism and government the good citizen. With the election of New Labour in 1997 and the ascend- There is now a well-developed research literature, ancy of ‘third way’ politics came a reinvigoration of inter- including a number of Nobel prizes, on the potential and ventionist social policy. A key feature was application of use of BPP in a government context (Baggio et al., 2021; scientiﬁc evidence, expertise, technocratic methods and Gofen et al., 2021; Lepenies and Malecka, 2018) - this ‘e-government’ to the business of public policy and itself is far from new. The literature focuses not only on public services (Giddens, 2013). Within this approach, the fundamentals of behavioural psychology, but also on communications was generally seen as an important but analysis of the policy actors who undertake these interven- separate aspect of government - gaining consent and aware- tions (Gofen et al., 2021). A whole range of behavioural ness for government policy and judging the public mood, levers have become well-established within government, rather than constituting a policy ‘lever’ in its own right. not only at the level of policy, but also in formulation of However, under David Cameron’s Coalition government law (Thaler and Sunstein, 2009). In part this follows the (post-2010), preventative policy was re-imagined and turn to evidence based/informed policy and is associated brought together with communications practices in the with a drive across government of experts and policymakers form of the Behavioural Insights Team, also known as the promoting these methods (Feitsma, 2019; Jones and ‘nudge’ unit. Whitehead, 2018). BPP approaches in government integrate Nudge, a term coined by Thaler and Sunstein (2009), is a heterogeneous set of theories - incorporating ideas from one of the better-known parts of a preventative turn in gov- cognitive psychology, social psychology, experimental ernment social policy, and involves reshaping the ‘choice and evidence-based government techniques such as the architecture’ in which individuals make decisions. The pro- RCT, and behavioural economics. This has seen it emerge vision of information by government is complemented with as a dominant policy paradigm in the UK in recent years, direct attempts to leverage existing social capital, repurpos- commanding signiﬁcant rhetorical and practical power - ing of ‘deviant’ social norms, and interventions in the built what Mols et al., describe as a ‘ﬁfth’ mode of governance environment and in consumer choice (Halpern, 2015). This alongside ‘hierarchy, markets, networks, and persuasion’ includes economic levers, such as changing the price of (Mols et al., 2015). tobacco, architectural levers, including design elements of In part, BPP has succeeded in establishing itself follow- the built environment, and, in addition to these older ‘situa- ing the government-wide shift to evidence-based tional’ approaches, the targeting of messaging at particular approaches to policy. These generally favour targeted inter- groups to inﬂuence the psychological and behavioural pro- ventions that are different for different groups, which can be cesses involved in making decisions. This turn to a psycho- studied semi-empirically. A skeptical critique might logical or ‘Behavioural Public Policy’ (BPP) draws on suggest that nudge in particular packages up simple, expertise from behavioural economics, psychology and easy-to-grasp ‘scientiﬁc’ theories, non-radical ‘tweaking’ neuroscience, often attempting to exploit hypothecated forms of policy change, and self-referential measures of unconscious biases in the brain to shape attitudes and performance which make them easy to ﬁnance, justify, behaviour (Halpern, 2015). In broader preventative enact, and evaluate. This increasingly established policy policy, this ‘in the moment’ behavioural shaping is some- paradigm is accompanied by a series of more critical times supported by more abstract attempts to shape ‘risky debates: around the legitimacy and rigour of behavioural cultures’,where the ‘culture’ (loosely deﬁned) of particu- science as an intellectual project, the practical efﬁcacy of lar social groups is seen (in a problematic and often nudge interventions, and what this development means implicitly racist or classist sense) to contribute to social for broader frameworks of political economy (Schubert, problems. 2017) and the evolving rationalities of government This can be seen across a range of policy areas, perhaps (Leggett, 2014). BPP’s status as an intellectual project has none more controversial than the UK’s approach to the come in for substantial critique, particularly from within domestic `War on Terror’ which has leaned heavily on sur- psychology itself, reﬂecting a tension between two distinct veillance and communications in addition to the more direct bodies of psychological knowledge and practice (Mols exercise of disruptive force, typiﬁed as the ‘inﬂuence opera- et al., 2015). One emphasises an individualist psychology, tion’. In this account, radicalisation (and other social issues) often based in neuroscientiﬁc research about individual can be tackled through the logics of public health, through a decision-making, which contrasts a countervailing combination of surveillance, individual behaviour change, approach rooted in social psychology that often informed cultural, and structural interventions (Heath-Kelly, 2017). older social marketing approaches. Individual-focused Where they are rolled out from the centre, rather than devel- understandings of nudges have generally predominated in oped locally, cultural programmes in the UK have often government. Social psychologists, on the other hand, promoted ‘state sanctioned’ versions of the cultures of com- advancing the ‘social identity approach’, generally critique munities deemed by the state to be risky, which tend to individual-focused forms of nudge, instead arguing that 4 Big Data & Society individuals should be understood within broader social con- Digital communications, tracking and texts and communities, whereby human decisions are advertising infrastructures viewed as shaped in large measure by prevailing social Thus, communications forms an important (and, crucially, norms, and to a lesser extent by the choice architecture often the cheapest) part of BPP. Targeted marketing has the individual is presented with when being nudged its roots in the 1920s (Grier and Kumanyika, 2010) with (Mols et al., 2015). Thus, they argue, successful lasting the aim of increasing the relevance of the messaging by behaviour change depends instead on shifting norms and market segment, and has been exploited by commerce broader changes to social cultural context. Ultimately, using a whole range of techniques and models of behaviour however, the responsibility for policy success in a behav- change, attitude change and reinforcement. However, com- iourist frame still relies on the citizen, with the role of the munication in the Internet age need not be the one-to-many state being to shape their behaviour (Gandy and Nemorin, style of the billboard, cigarette packet, or television advert 2019). (though these also use rudimentary forms of targeting); The stealth-based aspects of traditional nudges have the increasingly personalised and fragmented online also come in for critique (Jones and Whitehead, 2018). media has transformed how commerce addresses its BPP has been framed as ‘Liberal Paternalism’ markets, and as we argue, how governments communicate (Lepenies and Malecka, 2018; Sunstein, 2016; with their citizens. Hausman and Welch, 2010), with ﬁerce debates over As consumption of media shifts online, legacy commu- both its legitimacy and the scope for consent for nications channels increasingly fail to reach many groups in nudges that are in some tellings supposed to work only society (Ofcom, 2020), so advertisers have turned to online when the citizen is not actually aware of them (Schmidt channels - with over 3/5ths of UK ad spending pre- and Engelen, 2020). The secrecy of some nudge practices pandemic being spent on online channels, spending that can cause severe negative reactions when revealed, shifted even more to online markets over 2020-21 which can undermine the broader institutional legitimacy (WARC, 2021). ‘Top down’ advertising communications on which such interventions depend and stigmatise tar- practices have developed further in three key ways with geted groups (Mols et al., 2015). This phenomenon is the rise of digital platforms. First is the reﬁnement of part of a wider set of unexpected consequences which detailed real-time metrics about the communications avail- can result from this approach, collectively termed ‘blow- able to those who use the services - from simple views and back’. The complex and harmful side-effects of these likes, to rafts of data related to location, time and many approaches contest the idea of the ‘biddable’ citizen other characteristics of the individuals engaging with each who can be nudged or messaged in a ‘hypodermic’ communication, including successful sales, or ‘conversion’; model, as a passive recipient of government power, The second is the creation of tools, dashboards and ‘analy- who makes choices within a decision environment but tics’, to interpret and visualise this data and shape ongoing is powerless to change, re-imagine, or re-interpret that communications programmes. The third, since many of the environment (Hausman and Welch, 2010). Instead, the platform businesses work on an advertising model, is the citizen often proves far more active and critical a offer of paid channels to reach audiences, targeted and per- subject of messaging than nudge generally assumes. sonalised from second-to-second using the data and analy- The apparent successes and failures of the ‘nudge tics tools that continually collect information on unit’ perhaps obscure the broader movement of BPP individual’s behaviours, interests, and personal network - ideas throughout the UK Civil Service, including the “surveillance advertising” model (Crain, 2019). These within communications work. These forms of knowl- developments are part of a broader evolution of the business edge have long-standing roots in the civil service and models of the large international companies which provide other major government institutions, such as the NHS. most Internet services. Whether this is viewed through the While communications includes announcing govern- lens of surveillance capitalism (Zuboff, 2015), platform ment policy, perennial public information campaigns, capitalism (Srnicek, 2017), or data capitalism (West, attempts to shaping public opinion, or the broader 2019), this represents a change in some of the core ethics shaping of a sense of nationhood (Rose, 2000), commu- of marketing, with users not only segmented by sociodemo- nications units have developed over the past several graphic characteristics but also by the emergent properties decades a series of their own policy ‘levers’, wielding of enormous datasets of collected behaviours - clusters sur- the power to achieve policy aims in their own right by faced automatically and at scale through ‘big data’ and attempting to shape behaviour. In developing sets of ‘algorithmic’ techniques. professional standards around these practices, behav- To reach the online ‘eyeballs’ of those targeted and to ioural psychology and behavioural economics have inﬂuence their attitudes and behaviour, there is a rich and become a core body of professional expertise on which public policy and communications can draw to diverse ecosystem of channels, now dominated by practices enact BPP. known as “programmatic digital display advertising” .In Collier et al. 5 this model, advertising ‘space’ is sold in complex secondary towards a programme of control which we term ‘inﬂuence and tertiary markets and these proﬁles are not only collected government’. We now discuss the empirical case for this directly based on behaviour (as well as more traditional in depth. demographics), but also using data gathered from other data brokers to infer characteristics and behaviours where Methods this data is missing. Connecting these individual proﬁles to the connections in a person’s online social network Although considering the ways in which control technolo- allows messages not only to be targeted at the individual, gies such as these might potentially be abused provides a but to those around them (their family, friends, and collea- useful hook for critique, it is vital also to understand the gues) in order to shape their behaviour indirectly (Crain, reality of how they are being used in practice. Focusing 2019). Wider context for this targeting can be provided on the UK (though we also found evidence of these by search terms (for example, searching for a particular approaches elsewhere), we draw on publicly available product or service), visiting sites and services, ‘social’ documents to map out the evidence on how governments engagement, geographical location, characteristics of are already using these technologies to address contempor- people in an individual’s close network, and characteristics ary challenges of governance and control. In doing so, we of other people in a location in order to tailor messaging seek to build an empirical case for our argument that two even more effectively (Crain, 2019). pre-existing trends, the algorithmic turn and behavioural The targeting of adverts is only one part of what has turn in government, are fusing in the practices of become established as an infrastructure of inﬂuence government. methods facilitated by digital platforms - a whole set of The source documents were obtained through Internet standardised tools, processes, business services and search, initially using the keywords ‘behaviour change’, metrics can be called on by anyone willing to pay. ‘targeted advertising’ and ‘digital marketing’ in .gov Particularly inﬂuential individuals who sit as opinion space, and subsequent snowball sampling. They are all leaders or tastemakers at the centre of local networks (and accessible on gov.uk or other UK public sector domains, more successful national and international digital celebri- including the subdomains local.gov.uk and nationalcrimea- ties) can be identiﬁed using advanced analytics, and their gency.gov.uk. We also conducted a comprehensive survey function as inﬂuencers can then be used by brands and gov- of available materials on civil service training platforms, ernment to shape buying habits and behaviours (Coates such as the GCS website. They include 38 webpages and et al., 2019; Kostygina et al., 2020). These local inﬂuencers documents relating to local government strategy and guid- have far greater connections and legitimacy with the ance, 56 relating to central government strategic frame- small-scale communities in which people take part online, works and training materials, 30 for supplier websites, supported by a range of metrics and expertise which is col- and 30 examples relating to speciﬁc campaigns for a total lated by management companies. These revive the early of 154 primary sources. 2000 consumer concerns for ‘authenticity’ which brand Our evidence covers three main domains - ﬁrst, gov- managers attempted to co-opt during this period, but at a ernment strategy and training documents, largely drawn microscale - enacted through the lives of the ‘inﬂuencers’ from the Government Communication Service hub, the who are themselves at the mercy of their own metrics and Local Government Association website, and broader the models of the advertisers (Duffy et al., 2021). Some strategic documents from law enforcement. These set ‘inﬂuencers’ involved in targeted advertising have been out a case for how the intersection of behaviourism and highly speciﬁc to local communities, while in the case of targeted digital marketing are incorporated into govern- recent covid public health campaigns, some have been ment policy visions and of the tools, approaches, and drawn from the large pool of Youtube and Instagram ‘inﬂu- ideas which underpin their communities of practice. encers’, most often employed for their vast marketing cap- Secondly, we studied the websites of 30 marketing con- acity to younger demographics. sultancies and other contractors from the government’s Taken together, we describe this complex arrangement contractor lists, drawing out key narratives and of technologies, companies, markets, and practices as the approaches - what they claim to be doing and their ‘Surveillance Inﬂuence Infrastructure’ (SII); a dynamic set ‘offer’ to government. Thirdly, we surveyed 30 estab- of global infrastructures built on top of the Internet for sur- lished campaigns to establish what methods were veilling and shaping behaviour. What we ﬁnd particularly employed and how they were evaluated. These were striking is the coming-together of two separate structures chosen from examples and case studies on government of power - the turn to behaviourism and prevention on and supplier websites. In analysing this material, we one hand, and the infrastructures of targeting and inﬂuence found a wide variety of practices, encompassing both (and their complex tertiary markets) on the other. We theo- well-established strategic and professional frameworks rise this as a coming-together of social marketing and BPP underwritten by central government and also far more approaches with the practices of algorithmic governance, informal and amateurish use of these techniques. 6 Big Data & Society Our exploratory methodology seeks to ﬁrst establish an platforms using conventional media metrics (such as target- outline map of government strategies around the use of ing Tiktok in order to reach younger people), and adverts digital ‘inﬂuence’ marketing, and then to explore some of targeted using broad search terms. In this mode, there is the leading edge practices emerging. Although the core little sense of a systematic theory of change other than the strategy documents by their nature provide a representative broadcasting of a message unidirectionally. However, it is view of government strategic thinking and the body of in the next stage of sophistication that the truly novel cap- knowledge systematically drawn on by government, our acities are realised. review of case studies and contractors is more partial, reliant on publicly available sources. There is a clear bias Professional practice: modern public sector in these data - unsuccessful or uncontroversial campaigns communications are less likely to be reported, and agencies are incentivised to over-claim or represent these strategies in a more positive Moving up a level of sophistication, we ﬁnd, from our light. However, we examine these critically as evidence of review of core government strategic documents, that an some of the emerging practices in the ﬁeld, rather than a advanced competency framework around SII has been representative survey of all activity across central and incorporated into the daily practice of public sector orga- local government We are also aware that the effects of nisations. In this form, both dedicated communications these campaigns may be exaggerated, misreported or have professionals and (on occasion) frontline operational not been continued. staff are learning, teaching, and employing the skills associated with SII and advanced digital marketing, engaging with effective ad buying, iterative message Mapping inﬂuence government development, sophisticated targeting, and, crucially, the From our initial exploratory research, the core strategy development and articulation of theories of social documents and evaluations invoke a wide range of different issues and behaviour change strategies. We now forms and levels of practice, constituting everything from discuss the broad shape of these approaches in the UK sophisticated, multi-site inﬂuence campaigns to simply pur- public sector - who is using them, in which frameworks, chasing Google Ads. We set out a typology of three distinct andtowhatend. ‘modes’ of practice which we observe, involving progres- At present, this body of professional expertise has its sively deeper links to the networks of power and practice home in the Government Communication Service (who which attend this work. have also helped transfer this expertise to other public In each case, we discuss the broader strategic picture set bodies). The GCS website, the repository for training, prac- out in core strategy documents, and then, where appropri- tice, and policy, now includes substantial details of how ate, explore this ‘on the ground’ in more depth through behavioural communications campaigns can be conducted, case study examples. In general, this is a picture of a including evaluation and digital delivery. There are clear move from government ‘crisis’ in the face of new digital ﬂows of expertise between governmental, private, and media to an increasingly full embrace of its potentials, prac- quasi-governmental bodies, with nationwide strategic part- tices, and modes of knowledge. nerships and procurement structures with ad buying ser- vices (which also provide core government marketing training), dedicated creative agencies, and staff moving Naive uses: targeted ads as billboard space bidirectionally between government and the private sector. The ﬁrst form this takes is its most basic - the opening up of Much of the detail of these professional frameworks and targeted advertising as a space for traditional communica- training materials is freely available, and analysis of these tions. In this model, the advertising budget of the organisa- reveals the supportive structures of a fully-ﬂedged commu- tion is simply extended to include a range of online nity of practice. Within the GCS, behaviour change cam- ‘spaces’, with campaigns running on TV, billboards and paigns are structured within what is termed the OASIS in newspapers additionally being delivered through online model, a cyclical delivery model whose steps are: ad buys. These more naive forms are not part of a coherent Objectives, Audience Insight, Strategy/Idea, ‘new media’ strategy, but simply upgrades to existing com- Implementation, Scoring and Evaluation. This borrows munications routes - where they do appear in strategic docu- from ‘agile’ implementation frameworks for evidence- mentation, the focus is on bringing existing buying based design, with messages and targeting able to ‘evolve practices up to date. We found a range of examples of organically’ (at least in theory) informed by ongoing evalu- these across policy areas, which tended to be minimally tar- ation. Evaluative strategies employed include statistical geted - often simply at national level - and with little appar- data held by the Ofﬁce of National Statistics, New Media ent iteration or audience segmentation. This includes the Organisation, OFCOM, the GCS research library, and the bulk buying of non-targeted digital adverts, limited context- Cabinet Ofﬁce’s Insight and Evaluation Team . Targeting ual buying targeting particular kinds of websites and is used both to reach the desired population group but Collier et al. 7 also to design the intervention - using research, marketing or preventative law enforcement messaging based on data, and operational data. A variety of heuristics and fra- search engine queries and language used on social media. meworks are taught, often based on the design and evalu- At the level of speciﬁc places and communities there is ation of complex interventions in public health, such as evidence of further local targeting using demographic, geo- the ‘COM-B system’ (Capability, Opportunity, and graphic, and behavioural classiﬁcations. Here we ﬁnd the Motivation to engage in Behaviour) (Michie et al., 2011) content of behavioural adverts tailored at very local levels and other systems for non-specialists to design to include particular place names or local contexts, and communication-based behaviour change programmes. other strategies, including identifying community leaders Beyond these tools, core competencies for the GCS at the hyperlocal level and encouraging them to take part now include higher-level expertise in full-spectrum in adverts themselves. In the context of the pandemic, gov- digital marketing campaigns. This involves a range of ernment has been enlisting inﬂuencers to deliver core additional capacities, including the use of inﬂuencers behavioural messaging and promote the Test and Trace pro- and practices such as countering misinformation, asses- gramme. The major SII operators, increasingly vulnerable sing questions of data use and propriety, and protecting to public sensibilities about the harms which they facilitate, government brand identity in an environment where are developing internal relationships with governments and adverts can be displayed in unexpected and undesired devoting their own resources to public goods - using their contexts. There is an awareness within central govern- own targeting capacities to counter-message against radica- ment and the GCS that these advertisements are being lisation, grooming, and misinformation as well as occasion- deployed in an online environment which is fundamen- ally providing public bodies with free space for public issue tally adversarial; other actors are attempting to counter campaigns or working with organisations like Moonshot. the messages given out by government in a range of ways. From our research, we have established that frontline Government practices in this environment go well beyond tra- operational data collected in the management of public ser- ditional communications, embedding counter-disinformation vices, such as records of ﬁre service call-outs and other approaches (through the RESIST toolkit) and strategies for public service data, are being used, in combination with using large, multi-site campaigns to achieve direct behaviour commercial data and open data, to develop targeting pro- change. ﬁles for behavioural campaigns. These communications Although the broader ways in which communication practices link with the operational work of the public practices link up with other policy areas are doubtlessly fas- body or agency, feeding data collected by the campaigns cinating, we are particularly interested in the aspects of to inform and evaluate operational practices, and collecting these campaigns which relate to the use of commercial data from the operational side to tailor and target the cam- Surveillance Inﬂuence Infrastructures. Many of the exam- paigns. This is also crucial to the evaluation of these cam- ples available show real attempts at contextual ‘in the paigns - operational data are used to establish baselines, moment’ targeting working in tandem with situational forecast predicted effects and then measure change and nudges in the built environment or in user design. There evaluate campaigns, often incorporating continuous feed- is a clear drive to demonstrate innovation and creative or back and development. ‘edgy’ approaches, for example, an early campaign from As more public-facing examples and evaluations of 2015 in which fake celebrity proﬁles ‘matched’ young these campaigns emerge, it is increasingly clear that we people on Tinder, (a popular dating and hook-up app), are observing a well-established phenomenon across the then when the target agreed to the match, they were pre- UK public sector rather than simply the front line of inno- sented with an advert inviting them to sign up for organ vation, with many government departments and agencies donation. For more centralised campaigns there is scant now retaining dedicated behaviour change communications detail on the public pages of the GCS relating to exactly teams. There is a developing body of knowledge and set of how these adverts are targeted, but more evidence of audi- professional frameworks on which practitioners can draw, ence segmentation approaches can be found elsewhere, par- and a well-established infrastructure of services and con- ticularly at the local level. We found numerous examples sultancies which provide practical delivery and support. from key policy areas including Justice, National In short - behavioural campaigns are now a core aspect of Security, Environment, Health, Welfare, and Fire Safety, government communications work. ranging back to as early as 2011, with increasing frequency of campaigns in recent years. While initially targeting Professional practice: Law enforcement approaches were based rather broadly, on particular plat- forms or demographics, more recent campaigns use far A second, separately evolving strand of practice can be more intimate and algorithmically-enhanced methods - for identiﬁed in law enforcement, particularly in the Prevent example, the Home Ofﬁce using purchasing data for counter-radicalisation programme and broader preventative people who had bought candles recently and targeting policing. As an example, we discuss the NCA’s CYBER them through their smart speakers with ﬁre safety adverts CHOICES preventative diversion programme. This 8 Big Data & Society involves a process of identifying ‘at-risk’ young people, focus more on in-person interventions in schools and com- selected based on demographic risk and patterns of behav- munity mentorship, other VRUs appear to use a compre- iour detected by surveillance; in this case, on the basis of hensive data strategy in much the same manner as the online activity which indicates a potential interest in cyber- NCA, blending operational data, commercial data, and crime forums or the purchase of cybercrime tools. These research data into high-level and local dashboards for interventions target people before they engage in serious operational targeting, then feeding into sophisticated tar- illegal activity based on a set of risk characteristics. Once geted marketing campaigns. For example, in the VRU’s potential targets are identiﬁed by NCA surveillance, work in London with the Behavioural Insights Unit, we initial intervention is generally carried out through ‘knock can observe that the BI team’s recommendations involve and talk’ visits, where an NCA ofﬁcer visits the home of using a combination of social media and operational data, the young person and discusses their suspect behaviour further tightening the network of surveillance and messa- with them and with their parents. For those who are identi- ging around young people deemed ‘at-risk’: ﬁed as suitable, this leads to a workshop intervention, in which NCA ofﬁcers take a group of these children and The VRU and its partners have access to large swathes of give them talks and skills development in order to divert administrative data, which present a good opportunity for their ‘illicit’ skills into a legitimate career in cybersecurity. identifying behaviours or combinations of risk factors Throughout, data are gathered with the aim of not only which predict violence (as opposed to simply being asso- informing operational concerns, but contributing to a ciated with it). By drawing on advanced analytical techni- body of knowledge within the NCA about the people they ques such as algorithmic analyses and natural language are targeting and the characteristic factors which relate processing, the VRU can micro target resources where them to criminal offending pathways. risk is highest and bolster the ‘safety net’ around those This operational knowledge and data directly contributes most vulnerable to violence. In particular, we recommend to a complementary strategy involving targeted advertising, early analytical projects focus on: going missing and vio- known as ‘inﬂuence operations’, or (as previous scholarship lence; the use of social networks for predicting violence; has suggested) inﬂuence policing (Collier et al., 2021). analysing social media sentiment to predict threat online; These adverts, targeted at UK adolescents between the and exclusions and violence (Behavioural Insights Team age of 14 and 20 with an interest in gaming, are calibrated Violence in London Report) . to appear when users search for particular cybercrime ser- vices on Google, informing them that these services are Many of these VRU and NCA campaigns explicitly illegal and that they face NCA action if they purchase draw on the language and frameworks of PREVENT. them. Beginning as simple text-based adverts, the NCA developed them across a six month campaign in consult- Consultancy networks: ation with behavioural psychologists using the data they were collecting from their operational work. They addition- Although there is clear evidence of the UK Civil Service ally linked these adverts to hashtags for major gaming con- developing these capacities, much of this work is nonethe- ventions (assuming from their debrieﬁng interviews and the less outsourced to the private sector, including key suppor- academic literature a link between gaming and cybercrime), tive capacities for ad buying, creative, and market and purchased advertorials discussing the illegality of these segmentation. These services are purchased from a set of services on major gaming websites. Finally, they developed recommended or preferred suppliers as part of the profes- video adverts using their pathways data for circulation on sional practice model. However, in some cases, these con- YouTube. sultancies are contracted to take a more central role in There is evidence that the adverts themselves have been creating and shaping campaigns and running them as a effective in dissuading particular kinds of online crime, full service - from initial discussions with policymakers with a six-month NCA campaign appearing to be linked through to research, design and delivery. These agencies to a total cessation in growth in the purchase of Denial of market themselves on their capacity for deep engagement Service attacks in the UK, at a time during which these with communities on the ground, conducting focus attacks were rising sharply across in comparable nations groups, identifying micro-inﬂuencers, drawing on corporate (Collier et al., 2021). The hosting by these cybercrime ser- datasets and developing cultural and behavioural pictures of vices of Google Ads in order to secure advertising revenue often quite small target groups. These techniques, drawn means that the NCA have even managed to get these notices from marketing professional practice, also implicitly cast onto the sites themselves. the citizen as consumer - in this case, of narratives and These behaviour change campaigns have been taken up nudges. by some of the many Violence Reduction Units and Both community-focused and more centralised cam- Networks around the country, which take ‘public health’ paigns are widely evident. For example, an HIV behaviour approaches to violent crime. While campaigns in Scotland change campaign by the Hitch Marketing agency directly Collier et al. 9 involved people at a local level not only in appearing in the campaign-to-campaign basis, these proﬁles are ﬁnessed in campaign, but helping to co-design and implement it. particular local contexts, through surveys, interviews, Conversely, a particularly controversial and widely-reported ethnographic research, and other attempts to dig into the example was the SuperSisters website, a culture website for culture, ideas, and beliefs of the targeted population. This Muslim teens which was revealed to be covertly funded by develops very intimate views of communities, often at the UK Home Ofﬁce. Where these touch on criminal justice hyperlocal scale, but they are designed to render the concerns, there have been further controversies. The disas- citizen as consumer, an amalgam of tastes, qualities, and trous campaign by agency FCB Inferno and All City Media beliefs to be steered. It is at this stage that the essentialised targeting young black Londoners through chicken restau- components of culture – in the form of disconnected signif- rants showed the potential for serious backlash where tar- iers, strands of discourse, and aesthetics unmoored from geting was deemed to be discriminatory, yet reaction context – are collected, processed, and reconﬁgured as focused on the more visible, ofﬂine aspects of the cam- ‘positive’ versions of groups deemed risky. This serves paign, not the online targeted advertisements which also both a laudable practical rationale of ‘speaking to people formed a part. in their own language’ and tailoring messages for groups to be more relevant and comprehensible, but additionally a broader and more insidious function – the state shaping Theorising targeted advertising and the of culture. This represents a signiﬁcant development from state the state ‘policing the crisis’ to actively shaping cultures Having mapped the current available evidence of how the that are deemed ‘risky’ (Hall et al., 1978). UK state is making use of these technologies, we now These hybrid proﬁles, narratives, and aesthetics are then reﬂect on some of the theoretical concerns raised by this projected into the platform targeting infrastructures. This research. Ultimately, both business and states are interested reﬂects a governmentality of its own - a ‘big data’ rational- in shaping the behaviour of populations, and it is unsurpris- ity, which, in addition to classifying subjects into pre- ing to observe convergence in the practices that they are established deductive category systems of age, gender, developing to do so. The crucial development of this health, ethnicity, and sexuality, abstracts categories directly article is to observe and understand the coming together from data – clusters emerging from a collective of atomised of an increasingly well-established rationality of ‘behav- individuals (Aradau and Blanke, 2017). The data used to iourist’ government with a set of infrastructures, practices, create these proﬁles and clusters is extremely intimate – a and knowledges created by private sector ‘platform’ com- record of clicks, page visits, posts on social media, net- panies and the ecosystems which have grown up around works of interaction and behaviour whose characteristics them. cast back and forward between people within dense and We identify a common pattern between all these loose social groups, imputed through statistical aggregation approaches which is at the heart of ‘inﬂuence government’ where they are missing. Where this meets behaviourist gov- as an emerging rationality - the conjunction of three ernment, it represents a further abstraction of power towards systems of knowledge production. First, government softer and more insidious modes - which we term ‘inﬂuence departments draw on research and surveillance from government’. We believe that this represents a development their expansive operational datasets to develop proﬁles from the ‘nudge’ approach of the Behavioural Insights Unit, which are used to assess the needs, riskiness, and vulner- closing the loop between different forms of data, targeting, abilities of particular groups. These ‘proﬁles’ reﬂect not proﬁling, surveillance, and control sites at which the gov- established targets but whole groups felt to pose a higher ernment can place messages to a variety of ends. or lower risk of negative outcomes – proﬁles of patients, Our theorisation of the emerging biopolitics of ‘inﬂu- immigrants, potential offenders, prisoners, and welfare ence government’ reveals a triple system of knowledge pro- recipients. These category systems, which reﬂect well- duction which, when considered in the round, constitutes a established approaches to modern government, are those dizzyingly wide and deep array of lenses on communities produced by state institutions in the business of governing and individuals. These link the forms of biopower tradition- people’s lives and are facilitated by the large private sector ally associated with the state to the gathering of deeper cul- consultancies who provide much of the data infrastructure tural knowledge and consumer segmentation, and then to a for hospitals, immigration services, prisons, and other further range of algorithmic lenses whose classifying gaze institutions. (however partial and ﬂawed) is based on extremely intimate These institutional forms of knowledge are then surveillance of online behaviour. The rationality of risk is brought into contact with those produced in marketing written through this emerging system of power – it is at consultancies and consumer research organisations. The its heart a development of the epidemiological or ‘public proﬁles of state subjects are hybridised with a wide range of health’ approach to thinking about public policy problems consumer proﬁles - the classic demographic, cultural, and (Heath-Kelly, 2017). The context of austerity in the UK postcode-based systems used by traditional marketing. On a and moves to ‘cheap’ and ‘smart’ digital solutions, along 10 Big Data & Society with a more general imperative in the public sector for 2017). However in some cases this democratic emphasis bureaucrats to demonstrate innovation, leadership, and the seems to be in spite of, rather than as a result of the BPP use of scientiﬁc research to inform policy, has created the ‘libertarian paternalism’ philosophy (Gane, 2021) exempli- perfect conditions for behavioural psychologists, behav- ﬁed by the Behavioural Insights team. The BI campaigns ioural economists, and communications professionals to appeared much more top-down, surveillance focused, and sell inﬂuence government as a cheap, low-risk, and engaged on, rather than with, communities. The role of scientiﬁcally-grounded approach to policy and in some private sector agencies, particularly where they are cases to develop these uses below the radar of public demo- leading the work, ethical expertise, and evaluation, is trou- cratic scrutiny. As we discuss in the ﬁnal section of this bling. We have found evidence of some genuinely demo- article, ‘inﬂuence government’ is accompanied by a range cratic practices in these campaigns, where the company of potential issues and concerns which are at present insuf- works with communities to drive and design communica- ﬁciently addressed. tions through participatory, co-creative practices. However, particularly when the ‘innovative’ or ‘edgy’ side of the marketing agency dominates (rather than RIsks, ethics, and issues serious and systematic participation), some of these agen- The crucial development of this article is to observe and cies are characterised by a breathtaking naivety at best understand the coming together of an increasingly well- and serious failings at worst established rationality of ‘behaviourist’ government with The governance structure which underpins this policy a set of infrastructures, practices, and knowledges created decision-making is therefore important. Who draws on by private sector ‘platform’ companies and the ecosystems this knowledge, who sets priorities, and the transparency which have grown up around them. If inﬂuence government and accountability of these processes are crucial to their does constitute an emerging set of tools and practices for overall democratic legitimacy. The lack of transparency government communications and policy, it is clear that of these methods is of particular concern, especially there are serious practical and ethical aspects of their use where targeted advertising is used. Many of the situational which would beneﬁt from further consideration and demo- ‘tweaks’ which attend a traditional nudge or behavioural cratic discussion. In this ﬁnal substantive section, we give science-informed campaign are targeted, but they tend an overview of some of the main issues which we have also to be visible to populations beyond these targets. identiﬁed. Minimum unit pricing, changes to cigarette displays, anti- A central critique of these measures, and one which is no homeless spikes, and ‘go home’ vans are all targeted at par- stranger to ‘nudge’ and behavioural science (Ewert, 2020) ticular populations, but their broader visibility allows at is their contested relationship with democracy; that as prac- least a minimal route of accountability and critique - they ticed they are essentially top-down, providing public bodies can become subjects of public outrage and be reported on with a unidirectional capacity to shape the online environ- by journalists. Targeted advertising and other inﬂuence ment, behaviours, and cultures of their citizens (and those practices, such as the cultivation of ‘inﬂuencers’ in popula- groups who fall under their control but are denied citizen- tions, however, are only viewed (in theory) by the intended ship). Additionally, these inductive, iterative knowledge audience, reducing the capacity for broader accountability. processes can have a ‘reifying’ effect, in which the assump- Thus, while theoretically the data trails, open datasets, and tions, biases, and prejudices embedded in operational prac- auditable algorithms might make these more transparent, in tices shape the data which these practices (such as policing practice these are not as available for scrutiny. or medicine) produce about populations; data which are The development of ‘nudge’ into ‘inﬂuence government’ then fed into targeting and evaluation systems to become is much more well-established in counter-radicalisation and ‘self-fulﬁlling prophecies’. Thus a group is targeted national security research. The lessons from this body of because it is perceived to pose a higher risk, and it poses practice are as sobering when considering the application a higher risk because it is being targeted (Fraser and of these methods in the context of domestic governance Atkinson, 2014). Far from being an agile process that con- as they are in the broader global stages of power. The phe- stantly challenges assumptions and ﬁnds new and more nomenon of ‘blowback’, the violently negative reactions appropriate ways to address social problems, these practices which occur when groups realise that they are being can serve to amplify and embed ideas of ‘risky groups’ -a subject to these measures, reﬂect the fact that people’s rela- critique well-established in critical studies of existing and tionship with media is multifaceted – they know that the tar- historical forms of ‘smart’ policing (Hinton and Cook, geted inﬂuence infrastructure exists, they can often tell 2020). when it is being used and speculate as to how they are Some of this work is more ethically justiﬁable. For being targeted, and can react not only to messages to example, the VRU work emphasises the potential to incor- which they are exposed, but to the broader political dimen- porate more of the co-production values which might make sions of the messaging practices themselves. There is also such an approach bottom-up rather than top-down (Fraser, the potential for these inﬂuence approaches to in fact Collier et al. 11 serve to expose vulnerable groups to the very messages and models exist for developing and delivering this driven by narratives which policymakers are trying to counter, spread- communities themselves - participatory and co-produced ing them far wider. The potential contribution of both the approaches. Some of the examples we have found, particu- content and the targeting itself to stigma and labelling pro- larly where they are designed with an ethic of participation cesses well-established within criminology is also concern- and co-production with the targeted communities, appear to ing. Equally, there is the prospect of harming the perceived be genuinely compatible with established democratic legitimacy of the state and its institutions for targeted com- norms. Outside the domain of ‘nudge’, there are alternative munities already long-used to being on the receiving end of rationalities which could drive this which move away from state harm or neglect. the individual as the site of change and incorporate more Despite these concerns, the dangers posed by the state’s liberatory, community-based, solidaristic, and participatory use of targeted ads are accompanied by the dangers of not ideas, repurposing the platform technologies of control for using them. There are a wealth of areas in which targeted bottom-up social change. However it remains to be seen advertising and inﬂuence approaches are being used in what the role of the platforms might be in such a future - co-ordinated campaigns by malicious actors, from the whether infrastructures developed for commercial exploit- spread of illicit cybercrime services, to the targeting of vul- ation can ultimately serve social goods. nerable people with scams, to attempts by far-right, misogy- This paper is, we hope, the starting point for a much nist, racist, and queerphobic groups to spread hateful larger cross-disciplinary research project. Although there narratives and radicalise. There is a compelling argument is a wide and deep seam of research activity which to be made that the state has some duty to either counter addresses improving and measuring the efﬁcacy of different these malicious inﬂuence campaigns directly on the same kinds of behavioural interventions, there is next to none on terms, or to support communities in doing this work them- how these are being combined with digital inﬂuence tech- selves. Where the state averts its gaze intentionally there is nologies and how this hybrid approach is being realised the potential for these inﬂuence infrastructures to operate in the practices and processes of government. There is a unchecked, open as a technology of power to anyone able great deal of further research (both academic and journalis- to pay for adverts or who is able to subvert the algorithms tic) to be done: on the practices and rationalities of commu- (such as the far right communities who attempt to game nications professionals within government, on links to Youtube recommendations). Still, even in cases where private providers, and on what this means for the future there may be a clear moral imperative for the state to of government and law enforcement. Further questions employ targeted advertising and inﬂuence, robust transpar- should be asked about the extent to which the experience ency around the employment of citizen data should be of UK citizens in the online realm is being shaped by gov- prioritised. ernment inﬂuence. The international picture bears substan- Finally, it is important to note that the efﬁcacy of these tial further examination, as do the implications in global ‘inﬂuence’ interventions is extremely difﬁcult to assess. power, ethics, law, and democracy. This particular techno- Evaluation is a serious issue - unlike commercial targeting, logical future has already begun to arrive and the role of the where conversion to sale offers a fairly clear metric, it is academy should not only be to administer it, but to critique often extremely difﬁcult, despite the promises of the and challenge it as well. OASIS model and the access to administrative data pos- sessed by government, to evidence effects robustly. The Acknowledgments reliance on a tracking and targeting infrastructure that is The authors would like to thank the Scottish Centre for Crime and still fairly unreliable means that except for particularly self- Justice Research for their support with this project, particularly evident forms of targeting (such as the NCA’s approach, Alistair Fraser and Rachelle Cobain. which targets people searching for particular illegal ser- vices), many of these messages may be seen by the Declaration of conﬂicting interests wrong people, or not seen by the right people. Where these proﬁles act in the delivery of government policy, it The author(s) declared no potential conﬂicts of interest with respect to the research, authorship, and/or publication of this is crucial to account for the ways in which algorithmic article. bias or inaccuracy might shape who gets which messages, and the consequences of this. Funding The author(s) received no ﬁnancial support for the research, Concluding thoughts and possible futures authorship and/or publication of this article. Although we draw out a critical perspective on these approaches in this article, we do not argue that there are ORCID iD no possible positive futures of targeted messaging. Government will always involve communication, and Ben Collier https://orcid.org/0000-0002-9207-3068 12 Big Data & Society Gane N (2021) Nudge economics as libertarian paternalism. Notes Theory, Culture & Society 38(6): 119–142. 1. https://www.emarketer.com/content/uk-programmatic-digital- Garland D (1997) Governmentality’ and the problem of crime: display-ad-spending. foucault, criminology, sociology. Theoretical criminology 2. https://gcs.civilservice.gov.uk/publications/evaluation- 1(2): 173–214. framework/. Garland D (2001) The Culture of Control: Crime and Social 3. https://gcs.civilservice.gov.uk/publications/strategic- Order in Contemporary Society. Oxford: Oxford University communications-a-behavioural-approach/. Press. 4. https://www.bi.team/wp-content/uploads/2020/02/BIT- Giddens A (2013) The Third way: The Renewal of Social London-Violence-Reduction.pdf. Democracy. Cambridge: Polity Press, John Wiley & Sons. Gofen A, Moseley A, Thomann E, et al. (2021) Behavioural gov- ernance in the policy process: Introduction to the special issue. References Journal of European Public Policy 28(5): 633–657. Aradau C and Blanke T (2017) Politics of prediction: security and Grier S and Bryant CA (2005) Social marketing In public health. the time/space of governmentality in the age of big data. Annual Review of Public Health 26(1): 319–339. European Journal of Social Theory 20(3): 373–391. Grier SA and Kumanyika S (2010) Targeted marketing and public Atkin CK and Wallack L (1990) Mass Communication and Public health. Annual Review of Public Health 31(1): 349–369. Health: Complexities and Conﬂicts (Vol. 121). California: Hall S, Critcher C, Jefferson T, et al. (1978) Policing the Crisis: Sage. Mugging, the State and law and Order. Macmillan, London: Baggio M, et al. (2021) The evolution of behaviourally informed Macmillan International Higher Education. policy-making in the EU. Journal of European Public Policy Halpern D (2015) Inside the Nudge Unit: How small changes can 28: 1–19. make a big difference. London, UK: Random House. Barberis P (1998), The new public management and a new Hausman DM and Welch B (2010) Debate: to nudge or not to accountability. Public Administration 76(3): 450–471. nudge. Journal of Political Philosophy 18(1): 123–136. Coates AE, Hardman CA, Halford JC, et al. (2019) Social media Heath-Kelly C (2017) The geography of pre-criminal space: epidemio- inﬂuencer marketing and children’s food intake: A randomized logical imaginations of radicalisation risk in the UK prevent strategy, trial. Pediatrics 143(4): 1–9. 2007–2017. Critical studies on terrorism 10(2): 297–319. Collier B, Thomas DR, Clayton R, et al. (2021) Inﬂuence, infra- Hinton E and Cook D (2020) The mass criminalization of black structure, and recentering cybercrime policing: Evaluating Americans: A historical overview. Annual Review of Criminology emerging approaches to online law enforcement through a 4: 261–286. market for cybercrime services. Policing and Society:1–22. Jones R and Whitehead M (2018) “Politics done like science”: Cook J, Lynes J and Fries S (2021) Exploring mistakes and fail- critical perspectives on psychological governance and the ures in social marketing: The inside story. Social Marketing experimental state. Environment and Planning D: Society Quarterly 27(1): 13–31. and Space 36(2): 313–330. Crain M (2019) A critical political economy of Web advertising Katz J, Peberdy A and Douglas J (2000) Promoting Health: history. In: 2019. The SAGE Handbook of Web History.55 Knowledge and Practice, 2nd edition. Milton Keynes: The City Road, London: Sage Publications Ltd, 330–343. Open University, in association with Palgrave. Crawford A (2006) Networked governance and the post- Kostygina G, et al. (2020) Boosting health campaign reach and regulatory state? Steering, rowing and anchoring the provision engagement through Use of social Media inﬂuencers and of policing and security. Theoretical Criminology 10(4): 449– memes. Social Media + Society 6(2): 2056305120912475. 479. Leggett W (2014) The politics of behaviour change: Nudge, neo- Duffy BE, Pinch A, Sannon S, et al. (2021) The nested precarities of liberalism and the state. Policy & Politics 42(1): 3–19. creative labor on social Media. Social Media+ Society 7(2): : 1–12. Lepenies R and Małecka M (2018) The ethics of behavioural public Ewert B (2020) Moving beyond the obsession with nudging indi- policy. In: Lever A and Poama A (eds) The Routledge Handbook vidual behaviour: Towards a broader understanding of of Ethics and Public Policy. Abingdon, Oxon: Routledge, 513– Behavioural Public Policy. Public Policy and Administration 525. https://doi.org/10.4324/9781315461731-41 35(3): 337–360. Loader I (2006) Fall of the ‘platonic guardians’ liberalism, crimin- Feitsma J (2019) Brokering behaviour change: the work of behav- ology and political responses to crime in England and Wales. ioural insights experts in government. Policy and Politics British Journal of Criminology 46(4): 561–586. 47(1): 37–56. Lupton D (1995) The Imperative of Health: Public Health and the Foucault M (1982) The subject and power. Critical Inquiry 8(4): Regulated Body. London: Sage. 777–795. Mbembe A (2008) Necropolitics. In: Foucault in an Age of Terror. Fraser A (2017) Gangs & Crime: Critical Alternatives. London: London: Palgrave Macmillan, 152–182. Sage. Michie S, van Stralen MM and West R (2011) The behaviour Fraser A and Atkinson C (2014) Making up gangs: looping, label- change wheel: A new method for characterising and ling and the new politics of intelligence-led policing. Youth designing behaviour change interventions. Implementation Sci Justice 14(2): 154–170. 6(1): 1–12. Gandy OH and Nemorin S (2019) ‘Toward a political economy of Mols F, et al. (2015) Why a nudge is not enough: A social identity nudge: Smart city variations’, information. Communication & critique of governance by stealth. European Journal of Society 22(14): 2112–2126. Political Research 54(1): 81–98. Collier et al. 13 Ofcom (2020) Media Nations 2020: UK report, p. 101. Available Thaler RH and Sunstein CR (2009) Nudge: Improving at: https://www.ofcom.org.uk/__data/assets/pdf_ﬁle/0010/200 Decisions=About Health, Wealth, and Happiness. London: 503/media-nations-2020-uk-report.pdf. Penguin. Rose JW (2000) Making” Pictures in our Heads": Government WARC (2021) Advertising Association/WARC Expenditure Advertising in Canada. Greenwood, Connecticut: Greenwood Report http://expenditurereport.warc.com/. Publishing Group. West SM (2019) Data capitalism: redeﬁning the Schmidt AT and Engelen B (2020) The ethics of nudging: An logics of surveillance and privacy. Business & society 58(1): overview. Philosophy Compass 15(4): e12658. 20–41. Schubert C (2017) Exploring the (behavioural) political economy of Wallack L (2002) Public health, social nudging. Journal of Institutional Economics 13(3): 499–522. change, and Media advocacy. Social Marketing Quarterly Srnicek N (2017) Platform Capitalism. Cambridge, UK: Polity Press. 8(2): 25–31. Sunstein CR (2016) The Ethics of Inﬂuence: Government Zuboff S (2015) Big other: Surveillance capitalism and the pro- in the Age of Behavioral Science. Cambridge: Cambridge spects of an information civilization. Journal of Information University Press. https://doi.org/10.1017/CBO9781316493021 Technology 30(1): 75–89. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Big Data & Society SAGE http://www.deepdyve.com/lp/sage/influence-government-exploring-practices-ethics-and-power-in-the-use-UIKww20Ghq

Loading next page...

References (63)

M. Foucault (1982)
The Subject and Power
Critical Inquiry, 8
A. Fraser, Colin Atkinson (2014)
Making Up Gangs: Looping, Labelling and the New Politics of Intelligence-led Policing
Youth Justice, 14
Charlotte Heath-Kelly (2017)
The geography of pre-criminal space: epidemiological imaginations of radicalisation risk in the UK Prevent Strategy, 2007–2017
Critical Studies on Terrorism, 10
S. Michie, M. Stralen, R. West (2011)
The behaviour change wheel: A new method for characterising and designing behaviour change interventions
Implementation Science : IS, 6
Luís Souza (2003)
The culture of control: crime and social order in contemporary, society
Revista de Sociologia e Política
D. Garland (1997)
`Governmentality' and the Problem of Crime:
Theoretical Criminology, 1
Julie Cook, J. Lynes, S. Fries (2021)
Exploring Mistakes and Failures in Social Marketing: The Inside Story
Social Marketing Quarterly, 27
O. Gandy, Selena Nemorin (2018)
Toward a political economy of nudge: smart city variations
Information, Communication & Society, 22
J. Feitsma (2019)
Brokering behaviour change: the work of behavioural insights experts in government
Policy & Politics
?Toward a political economy of nudge: Smart city variations?, information
A. Turk (1980)
Policing the Crisis: Mugging, the State, and Law and Order.Stuart Hall , Chas Critcher , Tony Jefferson , John Clarke , Brian Roberts
American Journal of Sociology, 86
Ganna Kostygina, Hy Tran, Steven Binns, G. Szczypka, S. Emery, D. Vallone, E. Hair (2020)
Boosting Health Campaign Reach and Engagement Through Use of Social Media Influencers and Memes
Social Media + Society, 6
(2021)
Oxford textbook of violence prevention: Epidemiology, evidence, and policy
A. Coates, C. Hardman, J. Halford, P. Christiansen, E. Boyland (2019)
Social Media Influencer Marketing and Children’s Food Intake: A Randomized Trial
Pediatrics, 143
Will Leggett (2014)
The politics of behaviour change: nudge, neoliberalism and the state
Policy and Politics, 42
J. Katz, Alyson Peberdy, J. Douglas (2000)
Promoting health : knowledge and practice
I. Loader (2006)
Fall of the 'Platonic Guardians': Liberalism, Criminology and Political Responses to Crime in England and Wales
University of Oxford Legal Studies Research Paper Series
M. Baggio, Emanuele Ciriolo, Ginevra Marandola, René Bavel (2021)
The evolution of behaviourally informed policy-making in the EU
Journal of European Public Policy, 28
Stuart Hall, C. Critcher, T. Jefferson, John Clarke, B. Roberts (2018)
Policing the Crisis:
Essential Essays, Volume 1
Anat Gofen, A. Moseley, Eva Thomann, R. Weaver (2021)
Behavioural governance in the policy process: introduction to the special issue
Journal of European Public Policy, 28
(1990)
1990)Mass Communication and Public
A critical political economy of Web advertising history
D. Hausman, Brynn Welch (2010)
Debate: To Nudge or Not to Nudge*
Journal of Political Philosophy, 18
R. Sills (1998)
The third way: the renewal of social democracy
Lorraine Hawick, Alana Thomson, J. Cleland (2019)
A culture of control?
Medical Education, 53
Sonya Grier, S. Kumanyika (2010)
Targeted marketing and public health.
Annual review of public health, 31
A. Fraser (2017)
Gangs & Crime: Critical Alternatives
Christian Schubert (2017)
Exploring the (behavioural) political economy of nudging
Journal of Institutional Economics, 13
(2015)
Inside the Nudge Unit: How small changes
S. Yesiloglu (2020)
The rise of influencers and influencer marketing
Influencer Marketing
L. Wallack (2002)
Public Health, Social Change, and Media Advocacy
Social Marketin Quarterly, 8
Robert Lepenies, M. Małecka (2018)
The ethics of behavioural public policy
The Routledge Handbook of Ethics and Public Policy
P. Barberis (1998)
The New Public Management And A New Accountability
Public Administration, 76
R. Jones, M. Whitehead (2018)
‘Politics done like science’: Critical perspectives on psychological governance and the experimental state
Environment and Planning D: Society and Space, 36
Sonya Grier, C. Bryant (2005)
Social marketing in public health.
Annual review of public health, 26
A. Lever, A. Poama (2018)
The Routledge Handbook of Ethics and Public Policy
A. Crawford (2006)
Networked governance and the post-regulatory state?
Theoretical Criminology, 10
Networked governance and the post-regulatory state? Steering, rowing and anchoring the provision of policing and security
Frank Mols, S. Haslam, J. Jetten, Niklas Steffens (2015)
Why a nudge is not enough: a social identity critique of governance by stealth
European Journal of Political Research, 54
Necropolitics
Soon Chun, Stuart Shulman, Rodrigo Sandoval, E. Hovy (2010)
Government 2.0: Making connections between citizens, data and government
Inf. Polity, 15
S. West (2019)
Data Capitalism: Redefining the Logics of Surveillance and Privacy
Business & Society, 58
(2015)
Identity Ecosystem: Decentralised Citizens Engagement Technologies
Victo Neto (2019)
Platform capitalism
Revista Brasileira de Inovação
(1988)
Permission is granted to reproduce any part or all of this report for personal and educational use only
Claudia Aradau, Tobias Blanke (2017)
Politics of prediction
European Journal of Social Theory, 20
(2019)
How Propaganda Became Public Relations: Foucault and the Corporate Government of the Public
N. Gane (2021)
Nudge Economics as Libertarian Paternalism
Theory, Culture & Society, 38
Governmentality? and the problem of crime: foucault, criminology, sociology
Ben Collier, Daniel Thomas, R. Clayton, Alice Hutchings, Y. Chua (2021)
Influence, infrastructure, and recentering cybercrime policing: evaluating emerging approaches to online law enforcement through a market for cybercrime services
Policing and Society, 32
B. Duffy, Annika Pinch, Shruti Sannon, M. Sawey (2021)
The Nested Precarities of Creative Labor on Social Media
Social Media + Society, 7
Matthew Crain (2019)
A Critical Political Economy of Web Advertising History
The SAGE Handbook of Web History
Shoshana Zuboff (2015)
Big other: surveillance capitalism and the prospects of an information civilization
Journal of Information Technology, 30
D. Lupton (1995)
The Imperative of Health: Public Health and the Regulated Body
(2016)
Politics of prediction: Security and the time/space of governmentality in the age of big data
Benjamin Ewert (2020)
Moving beyond the obsession with nudging individual behaviour: Towards a broader understanding of Behavioural Public Policy
Public Policy and Administration, 35
Elaine Hoffman (2001)
Making “Pictures in Our Heads”: Government Advertising in Canada
Journal of Government Information, 28
Andreas Schmidt, Bart Engelen (2020)
The ethics of nudging: An overview
Philosophy Compass, 15
Chris Arney (2015)
Nudge: Improving Decisions about Health, Wealth, and Happiness
Mathematics and Computer Education, 49
Elizabeth Hinton, D. Cook (2021)
The Mass Criminalization of Black Americans: A Historical Overview
, 4
Programmatic Supply Chain Transparency Study, Incorporated Society of British Advertisers
Mona Lynch (2002)
The Culture of Control: Crime and Social Order in Contemporary Society
, 25
K. Dean (1997)
The imperative of health: Public health and the regulated body
Social Science & Medicine, 44

Publisher: SAGE
Copyright: Copyright © 2022 by SAGE Publications Ltd, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses.
ISSN: 2053-9517
eISSN: 2053-9517
DOI: 10.1177/20539517221078756
Publisher site: See Article on Publisher Site

Abstract

We have identiﬁed an emerging tool being used by the UK government across a range of public bodies in the service of public policy - the online targeted advertising infrastructure and the practices, consultancy ﬁrms, and forms of expertise which have grown up around it. This reﬂects an intensiﬁcation and adaptation of a broader ‘behavioural turn’ in the gov- ernmentality of the UK state and the increasing sophistication of everyday government communications. Contemporary UK public policy is fusing with the powerful tools for behaviour change created by the platform economy. Operational data and associated systems of classiﬁcation and proﬁling from public bodies are being hybridised with traditional con- sumer marketing proﬁles and then ‘projected’ onto the classiﬁcation systems of the targeted advertising infrastructures. This is not simply a case of algorithms being used for sorting, surveilling, and scoring; rather this suggests that targeted interventions in the cultural and behavioural life of communities are now a core part of governmental power which is being algorithmically-driven, in combination with inﬂuencer networks, traditional forms of messaging, and frontline opera- tional practices. We map these uses and practices of what we describe as the ‘Surveillance Inﬂuence Infrastructure’, iden- tifying key ethical issues and implications which we believe have yet to be fully investigated or considered. What we ﬁnd particularly striking is the coming-together of two separate structures of power - the governmental turn to behaviourism and prevention on one hand, and the infrastructures of targeting and inﬂuence (and their complex tertiary markets) on the other. We theorise this as a move beyond ‘nudge’ or ‘behavioural science’ approaches, towards a programme which we term ‘inﬂuence government’. Keywords Targeted advertising, government, power, marketing, criminal justice, public policy This article serves to set out our initial, exploratory Introduction ﬁndings about the use of these techniques in the UK The practices of private sector advertising and marketing public sector, discussing some of the emerging ways have long existed in a mutual relationship with government in which public bodies are using what we term the - from wartime propaganda to public health messaging. In Surveillance Inﬂuence Infrastructure (SII), developed their contemporary forms, marketing practices have for targeted advertising, to facilitate public policy out- evolved substantially beyond postcode-based demographic comes through ‘behaviour change’ strategies. We ﬁrst targeting, supported by the proliferation of online advertis- set out relevant context, explaining the thinking ing infrastructures which allow continually-updated target- behind these approaches to public policy - approaches ing based on behaviour and online activity. Our empirical which apply behavioural science and ‘public health’ research shows that these advanced marketing techniques are now being incorporated into the business of government and law enforcement. Although the ‘algorithmic turn’ and University of Edinburgh, Edinburgh, Scotland University of Strathclyde, Glasgow, Scotland ‘behaviourist turn’ are both well-established within UK governance, their combination in an emerging set of prac- Corresponding author: tices represents a novel, powerful, and in some cases poten- Ben Collier, University of Edinburgh, Edinburgh, Scotland. tially concerning frontier of government policy. Email: collier@ed.ac.uk Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https:// creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as speciﬁed on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Big Data & Society logics to a wide range of social issues. We then discuss housing, and health at poor communities and communities the complex network of online services and infrastruc- of colour. State engagement with ‘traditional’ media (parti- tures which generate databases and algorithmic models cularly newspapers) was crucial to manufacturing broad that facilitate the targeting of adverts. We then identify social consent for these approaches through concerted (drawing from an analysis of publicly available docu- media campaigns and moral panics which transformed the ments) a range of examples of how contemporary tar- cultural environment - around muggings, ‘hoodies’, anti- geted advertising though the SII is being used in social behaviour, asylum seekers, or the ‘beneﬁt cheat’ practice in the UK public sector. We provide some (Hall et al., 1978). initial explorations of what this means for public The rise of a ‘risk’ model of social groups and issues was policy and the character of state governance more gen- reﬂected in a broader managerialisation of the business of erally. Our paper concludes with a critical reﬂection government across the 1980s, becoming known as New on emergent ethical issues and some of the areas Public Management (Barberis, 1998). This managerialisa- which might conceivably beneﬁt from these techniques tion entailed the increasing collection of data and categor- -the ‘potential futures’ of surveillance-targeted behav- isation of publics (through patient databases, offender ioural messaging. matrices, and other systems) with a view to managing the public through metrics, outcomes, and measurable pro- cesses. Many of these metrics relate as much to the Governmental modernity: From social targets and outcomes used to monitor the intermediary marketing to responsibilisation ﬁrms granted public contracts under the neoliberal model To understand these developments it is useful to situate as to the public themselves. This also included the appropri- these practices within the history of government in the ation of the tools, standardised techniques, and models of UK. Governmental modernity, at the heart of much of private industry, including market segmentation and Foucault’s scholarship, is generally linked to the ascend- postcode-based marketing using the Mosaic or Acorn clas- ancy of a distinct rationality of power. This rationality con- siﬁcations. Foucauldian scholarship generally makes sense tends that, as the empirical study of human societies of this through the lens of biopolitics - the protective power progresses, their essential dynamics will be understood to and subjectifying force which the state exerts by gathering the extent that they can eventually be designed to function data about its citizens, with its companion, necropolitics - in more beneﬁcial and harmonious ways (Foucault, 1982; where the state represses through dramatically overextend- Garland, 1997). ing or tactically withdrawing its gaze (Mbembe, 2008). In the 1970s and 1980s, the adoption of the ideals of Modernist ideas of transforming society had given way to modern government - that the state could shape important forms of intervention and shaping based around surveil- aspects of experience and personhood - continued to lance and the management of ever more diffuse and globa- develop. Within public health, the rise of a ‘social market- lised forms of risk - maintaining the existing social order ing’ approach in this period saw commercial marketing and mitigating its worst effects. techniques deployed using the limited targeting available As the modes of governance in the UK and US shifted in through the mass media to transmit public health messages the neoliberal era, the state’s role became less the centra- (Atkin and Wallack, 1990; Lupton, 1995, p110; Grier and lised design of secure societies and prosocial citizens, and Bryant, 2005). These were designed by partnerships of ad more concerned with the responsibilisation (Garland, agencies, community groups, government agencies and 2001) of private citizens and businesses, who could pur- academics (Atkin and Wallack, 1990). While health lit- chase services from private sector providers. This was eracy, empowerment, social norm adaptation and collective coupled with the continuing ascendancy of repressive action may have been the aim of some of these pro- force targeted against communities deemed a risk to main- grammes, they were often designed with a model of the stream society - in health, crime, welfare, and other areas of ‘heroic’ individual able to change their own behaviour social policy. The role of the state in social design in this while ignoring more structural features (Katz et al., 2000; ‘marketised’ mode, though distanced and softened, was Wallack, 2002). Despite notable successes, these pro- not necessarily diminished; in idealised neoliberal societies grammes suffered from problems such as poor design, inad- the state takes the form of a ‘steering’, not a ‘rowing’ force, equate research and poor stakeholder coordination (Cook in which the delivery of government policy is devolved to et al., 2021). the private sector and civil society, but the state still ulti- Another approach sought to build public consensus mately sets the goals and agenda (Crawford, 2006). This around legitimacy for policy action. This was accompanied privatisation coincided with an increasing perception throughout the second half of the 20th Century by an expan- within government of the apparent intractability of social sion of the reach, severity, and targeting of the repressive issues, leaving only individuals, able to protect themselves, force of the state, particularly through the targeting of but with no sense of a wider possibility of collective social violent policing and punitive policies within education, change (Loader, 2006). Collier et al. 3 reﬂect capitalist, entrepreneurial, and ‘resilient’ models of Behaviourism and government the good citizen. With the election of New Labour in 1997 and the ascend- There is now a well-developed research literature, ancy of ‘third way’ politics came a reinvigoration of inter- including a number of Nobel prizes, on the potential and ventionist social policy. A key feature was application of use of BPP in a government context (Baggio et al., 2021; scientiﬁc evidence, expertise, technocratic methods and Gofen et al., 2021; Lepenies and Malecka, 2018) - this ‘e-government’ to the business of public policy and itself is far from new. The literature focuses not only on public services (Giddens, 2013). Within this approach, the fundamentals of behavioural psychology, but also on communications was generally seen as an important but analysis of the policy actors who undertake these interven- separate aspect of government - gaining consent and aware- tions (Gofen et al., 2021). A whole range of behavioural ness for government policy and judging the public mood, levers have become well-established within government, rather than constituting a policy ‘lever’ in its own right. not only at the level of policy, but also in formulation of However, under David Cameron’s Coalition government law (Thaler and Sunstein, 2009). In part this follows the (post-2010), preventative policy was re-imagined and turn to evidence based/informed policy and is associated brought together with communications practices in the with a drive across government of experts and policymakers form of the Behavioural Insights Team, also known as the promoting these methods (Feitsma, 2019; Jones and ‘nudge’ unit. Whitehead, 2018). BPP approaches in government integrate Nudge, a term coined by Thaler and Sunstein (2009), is a heterogeneous set of theories - incorporating ideas from one of the better-known parts of a preventative turn in gov- cognitive psychology, social psychology, experimental ernment social policy, and involves reshaping the ‘choice and evidence-based government techniques such as the architecture’ in which individuals make decisions. The pro- RCT, and behavioural economics. This has seen it emerge vision of information by government is complemented with as a dominant policy paradigm in the UK in recent years, direct attempts to leverage existing social capital, repurpos- commanding signiﬁcant rhetorical and practical power - ing of ‘deviant’ social norms, and interventions in the built what Mols et al., describe as a ‘ﬁfth’ mode of governance environment and in consumer choice (Halpern, 2015). This alongside ‘hierarchy, markets, networks, and persuasion’ includes economic levers, such as changing the price of (Mols et al., 2015). tobacco, architectural levers, including design elements of In part, BPP has succeeded in establishing itself follow- the built environment, and, in addition to these older ‘situa- ing the government-wide shift to evidence-based tional’ approaches, the targeting of messaging at particular approaches to policy. These generally favour targeted inter- groups to inﬂuence the psychological and behavioural pro- ventions that are different for different groups, which can be cesses involved in making decisions. This turn to a psycho- studied semi-empirically. A skeptical critique might logical or ‘Behavioural Public Policy’ (BPP) draws on suggest that nudge in particular packages up simple, expertise from behavioural economics, psychology and easy-to-grasp ‘scientiﬁc’ theories, non-radical ‘tweaking’ neuroscience, often attempting to exploit hypothecated forms of policy change, and self-referential measures of unconscious biases in the brain to shape attitudes and performance which make them easy to ﬁnance, justify, behaviour (Halpern, 2015). In broader preventative enact, and evaluate. This increasingly established policy policy, this ‘in the moment’ behavioural shaping is some- paradigm is accompanied by a series of more critical times supported by more abstract attempts to shape ‘risky debates: around the legitimacy and rigour of behavioural cultures’,where the ‘culture’ (loosely deﬁned) of particu- science as an intellectual project, the practical efﬁcacy of lar social groups is seen (in a problematic and often nudge interventions, and what this development means implicitly racist or classist sense) to contribute to social for broader frameworks of political economy (Schubert, problems. 2017) and the evolving rationalities of government This can be seen across a range of policy areas, perhaps (Leggett, 2014). BPP’s status as an intellectual project has none more controversial than the UK’s approach to the come in for substantial critique, particularly from within domestic `War on Terror’ which has leaned heavily on sur- psychology itself, reﬂecting a tension between two distinct veillance and communications in addition to the more direct bodies of psychological knowledge and practice (Mols exercise of disruptive force, typiﬁed as the ‘inﬂuence opera- et al., 2015). One emphasises an individualist psychology, tion’. In this account, radicalisation (and other social issues) often based in neuroscientiﬁc research about individual can be tackled through the logics of public health, through a decision-making, which contrasts a countervailing combination of surveillance, individual behaviour change, approach rooted in social psychology that often informed cultural, and structural interventions (Heath-Kelly, 2017). older social marketing approaches. Individual-focused Where they are rolled out from the centre, rather than devel- understandings of nudges have generally predominated in oped locally, cultural programmes in the UK have often government. Social psychologists, on the other hand, promoted ‘state sanctioned’ versions of the cultures of com- advancing the ‘social identity approach’, generally critique munities deemed by the state to be risky, which tend to individual-focused forms of nudge, instead arguing that 4 Big Data & Society individuals should be understood within broader social con- Digital communications, tracking and texts and communities, whereby human decisions are advertising infrastructures viewed as shaped in large measure by prevailing social Thus, communications forms an important (and, crucially, norms, and to a lesser extent by the choice architecture often the cheapest) part of BPP. Targeted marketing has the individual is presented with when being nudged its roots in the 1920s (Grier and Kumanyika, 2010) with (Mols et al., 2015). Thus, they argue, successful lasting the aim of increasing the relevance of the messaging by behaviour change depends instead on shifting norms and market segment, and has been exploited by commerce broader changes to social cultural context. Ultimately, using a whole range of techniques and models of behaviour however, the responsibility for policy success in a behav- change, attitude change and reinforcement. However, com- iourist frame still relies on the citizen, with the role of the munication in the Internet age need not be the one-to-many state being to shape their behaviour (Gandy and Nemorin, style of the billboard, cigarette packet, or television advert 2019). (though these also use rudimentary forms of targeting); The stealth-based aspects of traditional nudges have the increasingly personalised and fragmented online also come in for critique (Jones and Whitehead, 2018). media has transformed how commerce addresses its BPP has been framed as ‘Liberal Paternalism’ markets, and as we argue, how governments communicate (Lepenies and Malecka, 2018; Sunstein, 2016; with their citizens. Hausman and Welch, 2010), with ﬁerce debates over As consumption of media shifts online, legacy commu- both its legitimacy and the scope for consent for nications channels increasingly fail to reach many groups in nudges that are in some tellings supposed to work only society (Ofcom, 2020), so advertisers have turned to online when the citizen is not actually aware of them (Schmidt channels - with over 3/5ths of UK ad spending pre- and Engelen, 2020). The secrecy of some nudge practices pandemic being spent on online channels, spending that can cause severe negative reactions when revealed, shifted even more to online markets over 2020-21 which can undermine the broader institutional legitimacy (WARC, 2021). ‘Top down’ advertising communications on which such interventions depend and stigmatise tar- practices have developed further in three key ways with geted groups (Mols et al., 2015). This phenomenon is the rise of digital platforms. First is the reﬁnement of part of a wider set of unexpected consequences which detailed real-time metrics about the communications avail- can result from this approach, collectively termed ‘blow- able to those who use the services - from simple views and back’. The complex and harmful side-effects of these likes, to rafts of data related to location, time and many approaches contest the idea of the ‘biddable’ citizen other characteristics of the individuals engaging with each who can be nudged or messaged in a ‘hypodermic’ communication, including successful sales, or ‘conversion’; model, as a passive recipient of government power, The second is the creation of tools, dashboards and ‘analy- who makes choices within a decision environment but tics’, to interpret and visualise this data and shape ongoing is powerless to change, re-imagine, or re-interpret that communications programmes. The third, since many of the environment (Hausman and Welch, 2010). Instead, the platform businesses work on an advertising model, is the citizen often proves far more active and critical a offer of paid channels to reach audiences, targeted and per- subject of messaging than nudge generally assumes. sonalised from second-to-second using the data and analy- The apparent successes and failures of the ‘nudge tics tools that continually collect information on unit’ perhaps obscure the broader movement of BPP individual’s behaviours, interests, and personal network - ideas throughout the UK Civil Service, including the “surveillance advertising” model (Crain, 2019). These within communications work. These forms of knowl- developments are part of a broader evolution of the business edge have long-standing roots in the civil service and models of the large international companies which provide other major government institutions, such as the NHS. most Internet services. Whether this is viewed through the While communications includes announcing govern- lens of surveillance capitalism (Zuboff, 2015), platform ment policy, perennial public information campaigns, capitalism (Srnicek, 2017), or data capitalism (West, attempts to shaping public opinion, or the broader 2019), this represents a change in some of the core ethics shaping of a sense of nationhood (Rose, 2000), commu- of marketing, with users not only segmented by sociodemo- nications units have developed over the past several graphic characteristics but also by the emergent properties decades a series of their own policy ‘levers’, wielding of enormous datasets of collected behaviours - clusters sur- the power to achieve policy aims in their own right by faced automatically and at scale through ‘big data’ and attempting to shape behaviour. In developing sets of ‘algorithmic’ techniques. professional standards around these practices, behav- To reach the online ‘eyeballs’ of those targeted and to ioural psychology and behavioural economics have inﬂuence their attitudes and behaviour, there is a rich and become a core body of professional expertise on which public policy and communications can draw to diverse ecosystem of channels, now dominated by practices enact BPP. known as “programmatic digital display advertising” .In Collier et al. 5 this model, advertising ‘space’ is sold in complex secondary towards a programme of control which we term ‘inﬂuence and tertiary markets and these proﬁles are not only collected government’. We now discuss the empirical case for this directly based on behaviour (as well as more traditional in depth. demographics), but also using data gathered from other data brokers to infer characteristics and behaviours where Methods this data is missing. Connecting these individual proﬁles to the connections in a person’s online social network Although considering the ways in which control technolo- allows messages not only to be targeted at the individual, gies such as these might potentially be abused provides a but to those around them (their family, friends, and collea- useful hook for critique, it is vital also to understand the gues) in order to shape their behaviour indirectly (Crain, reality of how they are being used in practice. Focusing 2019). Wider context for this targeting can be provided on the UK (though we also found evidence of these by search terms (for example, searching for a particular approaches elsewhere), we draw on publicly available product or service), visiting sites and services, ‘social’ documents to map out the evidence on how governments engagement, geographical location, characteristics of are already using these technologies to address contempor- people in an individual’s close network, and characteristics ary challenges of governance and control. In doing so, we of other people in a location in order to tailor messaging seek to build an empirical case for our argument that two even more effectively (Crain, 2019). pre-existing trends, the algorithmic turn and behavioural The targeting of adverts is only one part of what has turn in government, are fusing in the practices of become established as an infrastructure of inﬂuence government. methods facilitated by digital platforms - a whole set of The source documents were obtained through Internet standardised tools, processes, business services and search, initially using the keywords ‘behaviour change’, metrics can be called on by anyone willing to pay. ‘targeted advertising’ and ‘digital marketing’ in .gov Particularly inﬂuential individuals who sit as opinion space, and subsequent snowball sampling. They are all leaders or tastemakers at the centre of local networks (and accessible on gov.uk or other UK public sector domains, more successful national and international digital celebri- including the subdomains local.gov.uk and nationalcrimea- ties) can be identiﬁed using advanced analytics, and their gency.gov.uk. We also conducted a comprehensive survey function as inﬂuencers can then be used by brands and gov- of available materials on civil service training platforms, ernment to shape buying habits and behaviours (Coates such as the GCS website. They include 38 webpages and et al., 2019; Kostygina et al., 2020). These local inﬂuencers documents relating to local government strategy and guid- have far greater connections and legitimacy with the ance, 56 relating to central government strategic frame- small-scale communities in which people take part online, works and training materials, 30 for supplier websites, supported by a range of metrics and expertise which is col- and 30 examples relating to speciﬁc campaigns for a total lated by management companies. These revive the early of 154 primary sources. 2000 consumer concerns for ‘authenticity’ which brand Our evidence covers three main domains - ﬁrst, gov- managers attempted to co-opt during this period, but at a ernment strategy and training documents, largely drawn microscale - enacted through the lives of the ‘inﬂuencers’ from the Government Communication Service hub, the who are themselves at the mercy of their own metrics and Local Government Association website, and broader the models of the advertisers (Duffy et al., 2021). Some strategic documents from law enforcement. These set ‘inﬂuencers’ involved in targeted advertising have been out a case for how the intersection of behaviourism and highly speciﬁc to local communities, while in the case of targeted digital marketing are incorporated into govern- recent covid public health campaigns, some have been ment policy visions and of the tools, approaches, and drawn from the large pool of Youtube and Instagram ‘inﬂu- ideas which underpin their communities of practice. encers’, most often employed for their vast marketing cap- Secondly, we studied the websites of 30 marketing con- acity to younger demographics. sultancies and other contractors from the government’s Taken together, we describe this complex arrangement contractor lists, drawing out key narratives and of technologies, companies, markets, and practices as the approaches - what they claim to be doing and their ‘Surveillance Inﬂuence Infrastructure’ (SII); a dynamic set ‘offer’ to government. Thirdly, we surveyed 30 estab- of global infrastructures built on top of the Internet for sur- lished campaigns to establish what methods were veilling and shaping behaviour. What we ﬁnd particularly employed and how they were evaluated. These were striking is the coming-together of two separate structures chosen from examples and case studies on government of power - the turn to behaviourism and prevention on and supplier websites. In analysing this material, we one hand, and the infrastructures of targeting and inﬂuence found a wide variety of practices, encompassing both (and their complex tertiary markets) on the other. We theo- well-established strategic and professional frameworks rise this as a coming-together of social marketing and BPP underwritten by central government and also far more approaches with the practices of algorithmic governance, informal and amateurish use of these techniques. 6 Big Data & Society Our exploratory methodology seeks to ﬁrst establish an platforms using conventional media metrics (such as target- outline map of government strategies around the use of ing Tiktok in order to reach younger people), and adverts digital ‘inﬂuence’ marketing, and then to explore some of targeted using broad search terms. In this mode, there is the leading edge practices emerging. Although the core little sense of a systematic theory of change other than the strategy documents by their nature provide a representative broadcasting of a message unidirectionally. However, it is view of government strategic thinking and the body of in the next stage of sophistication that the truly novel cap- knowledge systematically drawn on by government, our acities are realised. review of case studies and contractors is more partial, reliant on publicly available sources. There is a clear bias Professional practice: modern public sector in these data - unsuccessful or uncontroversial campaigns communications are less likely to be reported, and agencies are incentivised to over-claim or represent these strategies in a more positive Moving up a level of sophistication, we ﬁnd, from our light. However, we examine these critically as evidence of review of core government strategic documents, that an some of the emerging practices in the ﬁeld, rather than a advanced competency framework around SII has been representative survey of all activity across central and incorporated into the daily practice of public sector orga- local government We are also aware that the effects of nisations. In this form, both dedicated communications these campaigns may be exaggerated, misreported or have professionals and (on occasion) frontline operational not been continued. staff are learning, teaching, and employing the skills associated with SII and advanced digital marketing, engaging with effective ad buying, iterative message Mapping inﬂuence government development, sophisticated targeting, and, crucially, the From our initial exploratory research, the core strategy development and articulation of theories of social documents and evaluations invoke a wide range of different issues and behaviour change strategies. We now forms and levels of practice, constituting everything from discuss the broad shape of these approaches in the UK sophisticated, multi-site inﬂuence campaigns to simply pur- public sector - who is using them, in which frameworks, chasing Google Ads. We set out a typology of three distinct andtowhatend. ‘modes’ of practice which we observe, involving progres- At present, this body of professional expertise has its sively deeper links to the networks of power and practice home in the Government Communication Service (who which attend this work. have also helped transfer this expertise to other public In each case, we discuss the broader strategic picture set bodies). The GCS website, the repository for training, prac- out in core strategy documents, and then, where appropri- tice, and policy, now includes substantial details of how ate, explore this ‘on the ground’ in more depth through behavioural communications campaigns can be conducted, case study examples. In general, this is a picture of a including evaluation and digital delivery. There are clear move from government ‘crisis’ in the face of new digital ﬂows of expertise between governmental, private, and media to an increasingly full embrace of its potentials, prac- quasi-governmental bodies, with nationwide strategic part- tices, and modes of knowledge. nerships and procurement structures with ad buying ser- vices (which also provide core government marketing training), dedicated creative agencies, and staff moving Naive uses: targeted ads as billboard space bidirectionally between government and the private sector. The ﬁrst form this takes is its most basic - the opening up of Much of the detail of these professional frameworks and targeted advertising as a space for traditional communica- training materials is freely available, and analysis of these tions. In this model, the advertising budget of the organisa- reveals the supportive structures of a fully-ﬂedged commu- tion is simply extended to include a range of online nity of practice. Within the GCS, behaviour change cam- ‘spaces’, with campaigns running on TV, billboards and paigns are structured within what is termed the OASIS in newspapers additionally being delivered through online model, a cyclical delivery model whose steps are: ad buys. These more naive forms are not part of a coherent Objectives, Audience Insight, Strategy/Idea, ‘new media’ strategy, but simply upgrades to existing com- Implementation, Scoring and Evaluation. This borrows munications routes - where they do appear in strategic docu- from ‘agile’ implementation frameworks for evidence- mentation, the focus is on bringing existing buying based design, with messages and targeting able to ‘evolve practices up to date. We found a range of examples of organically’ (at least in theory) informed by ongoing evalu- these across policy areas, which tended to be minimally tar- ation. Evaluative strategies employed include statistical geted - often simply at national level - and with little appar- data held by the Ofﬁce of National Statistics, New Media ent iteration or audience segmentation. This includes the Organisation, OFCOM, the GCS research library, and the bulk buying of non-targeted digital adverts, limited context- Cabinet Ofﬁce’s Insight and Evaluation Team . Targeting ual buying targeting particular kinds of websites and is used both to reach the desired population group but Collier et al. 7 also to design the intervention - using research, marketing or preventative law enforcement messaging based on data, and operational data. A variety of heuristics and fra- search engine queries and language used on social media. meworks are taught, often based on the design and evalu- At the level of speciﬁc places and communities there is ation of complex interventions in public health, such as evidence of further local targeting using demographic, geo- the ‘COM-B system’ (Capability, Opportunity, and graphic, and behavioural classiﬁcations. Here we ﬁnd the Motivation to engage in Behaviour) (Michie et al., 2011) content of behavioural adverts tailored at very local levels and other systems for non-specialists to design to include particular place names or local contexts, and communication-based behaviour change programmes. other strategies, including identifying community leaders Beyond these tools, core competencies for the GCS at the hyperlocal level and encouraging them to take part now include higher-level expertise in full-spectrum in adverts themselves. In the context of the pandemic, gov- digital marketing campaigns. This involves a range of ernment has been enlisting inﬂuencers to deliver core additional capacities, including the use of inﬂuencers behavioural messaging and promote the Test and Trace pro- and practices such as countering misinformation, asses- gramme. The major SII operators, increasingly vulnerable sing questions of data use and propriety, and protecting to public sensibilities about the harms which they facilitate, government brand identity in an environment where are developing internal relationships with governments and adverts can be displayed in unexpected and undesired devoting their own resources to public goods - using their contexts. There is an awareness within central govern- own targeting capacities to counter-message against radica- ment and the GCS that these advertisements are being lisation, grooming, and misinformation as well as occasion- deployed in an online environment which is fundamen- ally providing public bodies with free space for public issue tally adversarial; other actors are attempting to counter campaigns or working with organisations like Moonshot. the messages given out by government in a range of ways. From our research, we have established that frontline Government practices in this environment go well beyond tra- operational data collected in the management of public ser- ditional communications, embedding counter-disinformation vices, such as records of ﬁre service call-outs and other approaches (through the RESIST toolkit) and strategies for public service data, are being used, in combination with using large, multi-site campaigns to achieve direct behaviour commercial data and open data, to develop targeting pro- change. ﬁles for behavioural campaigns. These communications Although the broader ways in which communication practices link with the operational work of the public practices link up with other policy areas are doubtlessly fas- body or agency, feeding data collected by the campaigns cinating, we are particularly interested in the aspects of to inform and evaluate operational practices, and collecting these campaigns which relate to the use of commercial data from the operational side to tailor and target the cam- Surveillance Inﬂuence Infrastructures. Many of the exam- paigns. This is also crucial to the evaluation of these cam- ples available show real attempts at contextual ‘in the paigns - operational data are used to establish baselines, moment’ targeting working in tandem with situational forecast predicted effects and then measure change and nudges in the built environment or in user design. There evaluate campaigns, often incorporating continuous feed- is a clear drive to demonstrate innovation and creative or back and development. ‘edgy’ approaches, for example, an early campaign from As more public-facing examples and evaluations of 2015 in which fake celebrity proﬁles ‘matched’ young these campaigns emerge, it is increasingly clear that we people on Tinder, (a popular dating and hook-up app), are observing a well-established phenomenon across the then when the target agreed to the match, they were pre- UK public sector rather than simply the front line of inno- sented with an advert inviting them to sign up for organ vation, with many government departments and agencies donation. For more centralised campaigns there is scant now retaining dedicated behaviour change communications detail on the public pages of the GCS relating to exactly teams. There is a developing body of knowledge and set of how these adverts are targeted, but more evidence of audi- professional frameworks on which practitioners can draw, ence segmentation approaches can be found elsewhere, par- and a well-established infrastructure of services and con- ticularly at the local level. We found numerous examples sultancies which provide practical delivery and support. from key policy areas including Justice, National In short - behavioural campaigns are now a core aspect of Security, Environment, Health, Welfare, and Fire Safety, government communications work. ranging back to as early as 2011, with increasing frequency of campaigns in recent years. While initially targeting Professional practice: Law enforcement approaches were based rather broadly, on particular plat- forms or demographics, more recent campaigns use far A second, separately evolving strand of practice can be more intimate and algorithmically-enhanced methods - for identiﬁed in law enforcement, particularly in the Prevent example, the Home Ofﬁce using purchasing data for counter-radicalisation programme and broader preventative people who had bought candles recently and targeting policing. As an example, we discuss the NCA’s CYBER them through their smart speakers with ﬁre safety adverts CHOICES preventative diversion programme. This 8 Big Data & Society involves a process of identifying ‘at-risk’ young people, focus more on in-person interventions in schools and com- selected based on demographic risk and patterns of behav- munity mentorship, other VRUs appear to use a compre- iour detected by surveillance; in this case, on the basis of hensive data strategy in much the same manner as the online activity which indicates a potential interest in cyber- NCA, blending operational data, commercial data, and crime forums or the purchase of cybercrime tools. These research data into high-level and local dashboards for interventions target people before they engage in serious operational targeting, then feeding into sophisticated tar- illegal activity based on a set of risk characteristics. Once geted marketing campaigns. For example, in the VRU’s potential targets are identiﬁed by NCA surveillance, work in London with the Behavioural Insights Unit, we initial intervention is generally carried out through ‘knock can observe that the BI team’s recommendations involve and talk’ visits, where an NCA ofﬁcer visits the home of using a combination of social media and operational data, the young person and discusses their suspect behaviour further tightening the network of surveillance and messa- with them and with their parents. For those who are identi- ging around young people deemed ‘at-risk’: ﬁed as suitable, this leads to a workshop intervention, in which NCA ofﬁcers take a group of these children and The VRU and its partners have access to large swathes of give them talks and skills development in order to divert administrative data, which present a good opportunity for their ‘illicit’ skills into a legitimate career in cybersecurity. identifying behaviours or combinations of risk factors Throughout, data are gathered with the aim of not only which predict violence (as opposed to simply being asso- informing operational concerns, but contributing to a ciated with it). By drawing on advanced analytical techni- body of knowledge within the NCA about the people they ques such as algorithmic analyses and natural language are targeting and the characteristic factors which relate processing, the VRU can micro target resources where them to criminal offending pathways. risk is highest and bolster the ‘safety net’ around those This operational knowledge and data directly contributes most vulnerable to violence. In particular, we recommend to a complementary strategy involving targeted advertising, early analytical projects focus on: going missing and vio- known as ‘inﬂuence operations’, or (as previous scholarship lence; the use of social networks for predicting violence; has suggested) inﬂuence policing (Collier et al., 2021). analysing social media sentiment to predict threat online; These adverts, targeted at UK adolescents between the and exclusions and violence (Behavioural Insights Team age of 14 and 20 with an interest in gaming, are calibrated Violence in London Report) . to appear when users search for particular cybercrime ser- vices on Google, informing them that these services are Many of these VRU and NCA campaigns explicitly illegal and that they face NCA action if they purchase draw on the language and frameworks of PREVENT. them. Beginning as simple text-based adverts, the NCA developed them across a six month campaign in consult- Consultancy networks: ation with behavioural psychologists using the data they were collecting from their operational work. They addition- Although there is clear evidence of the UK Civil Service ally linked these adverts to hashtags for major gaming con- developing these capacities, much of this work is nonethe- ventions (assuming from their debrieﬁng interviews and the less outsourced to the private sector, including key suppor- academic literature a link between gaming and cybercrime), tive capacities for ad buying, creative, and market and purchased advertorials discussing the illegality of these segmentation. These services are purchased from a set of services on major gaming websites. Finally, they developed recommended or preferred suppliers as part of the profes- video adverts using their pathways data for circulation on sional practice model. However, in some cases, these con- YouTube. sultancies are contracted to take a more central role in There is evidence that the adverts themselves have been creating and shaping campaigns and running them as a effective in dissuading particular kinds of online crime, full service - from initial discussions with policymakers with a six-month NCA campaign appearing to be linked through to research, design and delivery. These agencies to a total cessation in growth in the purchase of Denial of market themselves on their capacity for deep engagement Service attacks in the UK, at a time during which these with communities on the ground, conducting focus attacks were rising sharply across in comparable nations groups, identifying micro-inﬂuencers, drawing on corporate (Collier et al., 2021). The hosting by these cybercrime ser- datasets and developing cultural and behavioural pictures of vices of Google Ads in order to secure advertising revenue often quite small target groups. These techniques, drawn means that the NCA have even managed to get these notices from marketing professional practice, also implicitly cast onto the sites themselves. the citizen as consumer - in this case, of narratives and These behaviour change campaigns have been taken up nudges. by some of the many Violence Reduction Units and Both community-focused and more centralised cam- Networks around the country, which take ‘public health’ paigns are widely evident. For example, an HIV behaviour approaches to violent crime. While campaigns in Scotland change campaign by the Hitch Marketing agency directly Collier et al. 9 involved people at a local level not only in appearing in the campaign-to-campaign basis, these proﬁles are ﬁnessed in campaign, but helping to co-design and implement it. particular local contexts, through surveys, interviews, Conversely, a particularly controversial and widely-reported ethnographic research, and other attempts to dig into the example was the SuperSisters website, a culture website for culture, ideas, and beliefs of the targeted population. This Muslim teens which was revealed to be covertly funded by develops very intimate views of communities, often at the UK Home Ofﬁce. Where these touch on criminal justice hyperlocal scale, but they are designed to render the concerns, there have been further controversies. The disas- citizen as consumer, an amalgam of tastes, qualities, and trous campaign by agency FCB Inferno and All City Media beliefs to be steered. It is at this stage that the essentialised targeting young black Londoners through chicken restau- components of culture – in the form of disconnected signif- rants showed the potential for serious backlash where tar- iers, strands of discourse, and aesthetics unmoored from geting was deemed to be discriminatory, yet reaction context – are collected, processed, and reconﬁgured as focused on the more visible, ofﬂine aspects of the cam- ‘positive’ versions of groups deemed risky. This serves paign, not the online targeted advertisements which also both a laudable practical rationale of ‘speaking to people formed a part. in their own language’ and tailoring messages for groups to be more relevant and comprehensible, but additionally a broader and more insidious function – the state shaping Theorising targeted advertising and the of culture. This represents a signiﬁcant development from state the state ‘policing the crisis’ to actively shaping cultures Having mapped the current available evidence of how the that are deemed ‘risky’ (Hall et al., 1978). UK state is making use of these technologies, we now These hybrid proﬁles, narratives, and aesthetics are then reﬂect on some of the theoretical concerns raised by this projected into the platform targeting infrastructures. This research. Ultimately, both business and states are interested reﬂects a governmentality of its own - a ‘big data’ rational- in shaping the behaviour of populations, and it is unsurpris- ity, which, in addition to classifying subjects into pre- ing to observe convergence in the practices that they are established deductive category systems of age, gender, developing to do so. The crucial development of this health, ethnicity, and sexuality, abstracts categories directly article is to observe and understand the coming together from data – clusters emerging from a collective of atomised of an increasingly well-established rationality of ‘behav- individuals (Aradau and Blanke, 2017). The data used to iourist’ government with a set of infrastructures, practices, create these proﬁles and clusters is extremely intimate – a and knowledges created by private sector ‘platform’ com- record of clicks, page visits, posts on social media, net- panies and the ecosystems which have grown up around works of interaction and behaviour whose characteristics them. cast back and forward between people within dense and We identify a common pattern between all these loose social groups, imputed through statistical aggregation approaches which is at the heart of ‘inﬂuence government’ where they are missing. Where this meets behaviourist gov- as an emerging rationality - the conjunction of three ernment, it represents a further abstraction of power towards systems of knowledge production. First, government softer and more insidious modes - which we term ‘inﬂuence departments draw on research and surveillance from government’. We believe that this represents a development their expansive operational datasets to develop proﬁles from the ‘nudge’ approach of the Behavioural Insights Unit, which are used to assess the needs, riskiness, and vulner- closing the loop between different forms of data, targeting, abilities of particular groups. These ‘proﬁles’ reﬂect not proﬁling, surveillance, and control sites at which the gov- established targets but whole groups felt to pose a higher ernment can place messages to a variety of ends. or lower risk of negative outcomes – proﬁles of patients, Our theorisation of the emerging biopolitics of ‘inﬂu- immigrants, potential offenders, prisoners, and welfare ence government’ reveals a triple system of knowledge pro- recipients. These category systems, which reﬂect well- duction which, when considered in the round, constitutes a established approaches to modern government, are those dizzyingly wide and deep array of lenses on communities produced by state institutions in the business of governing and individuals. These link the forms of biopower tradition- people’s lives and are facilitated by the large private sector ally associated with the state to the gathering of deeper cul- consultancies who provide much of the data infrastructure tural knowledge and consumer segmentation, and then to a for hospitals, immigration services, prisons, and other further range of algorithmic lenses whose classifying gaze institutions. (however partial and ﬂawed) is based on extremely intimate These institutional forms of knowledge are then surveillance of online behaviour. The rationality of risk is brought into contact with those produced in marketing written through this emerging system of power – it is at consultancies and consumer research organisations. The its heart a development of the epidemiological or ‘public proﬁles of state subjects are hybridised with a wide range of health’ approach to thinking about public policy problems consumer proﬁles - the classic demographic, cultural, and (Heath-Kelly, 2017). The context of austerity in the UK postcode-based systems used by traditional marketing. On a and moves to ‘cheap’ and ‘smart’ digital solutions, along 10 Big Data & Society with a more general imperative in the public sector for 2017). However in some cases this democratic emphasis bureaucrats to demonstrate innovation, leadership, and the seems to be in spite of, rather than as a result of the BPP use of scientiﬁc research to inform policy, has created the ‘libertarian paternalism’ philosophy (Gane, 2021) exempli- perfect conditions for behavioural psychologists, behav- ﬁed by the Behavioural Insights team. The BI campaigns ioural economists, and communications professionals to appeared much more top-down, surveillance focused, and sell inﬂuence government as a cheap, low-risk, and engaged on, rather than with, communities. The role of scientiﬁcally-grounded approach to policy and in some private sector agencies, particularly where they are cases to develop these uses below the radar of public demo- leading the work, ethical expertise, and evaluation, is trou- cratic scrutiny. As we discuss in the ﬁnal section of this bling. We have found evidence of some genuinely demo- article, ‘inﬂuence government’ is accompanied by a range cratic practices in these campaigns, where the company of potential issues and concerns which are at present insuf- works with communities to drive and design communica- ﬁciently addressed. tions through participatory, co-creative practices. However, particularly when the ‘innovative’ or ‘edgy’ side of the marketing agency dominates (rather than RIsks, ethics, and issues serious and systematic participation), some of these agen- The crucial development of this article is to observe and cies are characterised by a breathtaking naivety at best understand the coming together of an increasingly well- and serious failings at worst established rationality of ‘behaviourist’ government with The governance structure which underpins this policy a set of infrastructures, practices, and knowledges created decision-making is therefore important. Who draws on by private sector ‘platform’ companies and the ecosystems this knowledge, who sets priorities, and the transparency which have grown up around them. If inﬂuence government and accountability of these processes are crucial to their does constitute an emerging set of tools and practices for overall democratic legitimacy. The lack of transparency government communications and policy, it is clear that of these methods is of particular concern, especially there are serious practical and ethical aspects of their use where targeted advertising is used. Many of the situational which would beneﬁt from further consideration and demo- ‘tweaks’ which attend a traditional nudge or behavioural cratic discussion. In this ﬁnal substantive section, we give science-informed campaign are targeted, but they tend an overview of some of the main issues which we have also to be visible to populations beyond these targets. identiﬁed. Minimum unit pricing, changes to cigarette displays, anti- A central critique of these measures, and one which is no homeless spikes, and ‘go home’ vans are all targeted at par- stranger to ‘nudge’ and behavioural science (Ewert, 2020) ticular populations, but their broader visibility allows at is their contested relationship with democracy; that as prac- least a minimal route of accountability and critique - they ticed they are essentially top-down, providing public bodies can become subjects of public outrage and be reported on with a unidirectional capacity to shape the online environ- by journalists. Targeted advertising and other inﬂuence ment, behaviours, and cultures of their citizens (and those practices, such as the cultivation of ‘inﬂuencers’ in popula- groups who fall under their control but are denied citizen- tions, however, are only viewed (in theory) by the intended ship). Additionally, these inductive, iterative knowledge audience, reducing the capacity for broader accountability. processes can have a ‘reifying’ effect, in which the assump- Thus, while theoretically the data trails, open datasets, and tions, biases, and prejudices embedded in operational prac- auditable algorithms might make these more transparent, in tices shape the data which these practices (such as policing practice these are not as available for scrutiny. or medicine) produce about populations; data which are The development of ‘nudge’ into ‘inﬂuence government’ then fed into targeting and evaluation systems to become is much more well-established in counter-radicalisation and ‘self-fulﬁlling prophecies’. Thus a group is targeted national security research. The lessons from this body of because it is perceived to pose a higher risk, and it poses practice are as sobering when considering the application a higher risk because it is being targeted (Fraser and of these methods in the context of domestic governance Atkinson, 2014). Far from being an agile process that con- as they are in the broader global stages of power. The phe- stantly challenges assumptions and ﬁnds new and more nomenon of ‘blowback’, the violently negative reactions appropriate ways to address social problems, these practices which occur when groups realise that they are being can serve to amplify and embed ideas of ‘risky groups’ -a subject to these measures, reﬂect the fact that people’s rela- critique well-established in critical studies of existing and tionship with media is multifaceted – they know that the tar- historical forms of ‘smart’ policing (Hinton and Cook, geted inﬂuence infrastructure exists, they can often tell 2020). when it is being used and speculate as to how they are Some of this work is more ethically justiﬁable. For being targeted, and can react not only to messages to example, the VRU work emphasises the potential to incor- which they are exposed, but to the broader political dimen- porate more of the co-production values which might make sions of the messaging practices themselves. There is also such an approach bottom-up rather than top-down (Fraser, the potential for these inﬂuence approaches to in fact Collier et al. 11 serve to expose vulnerable groups to the very messages and models exist for developing and delivering this driven by narratives which policymakers are trying to counter, spread- communities themselves - participatory and co-produced ing them far wider. The potential contribution of both the approaches. Some of the examples we have found, particu- content and the targeting itself to stigma and labelling pro- larly where they are designed with an ethic of participation cesses well-established within criminology is also concern- and co-production with the targeted communities, appear to ing. Equally, there is the prospect of harming the perceived be genuinely compatible with established democratic legitimacy of the state and its institutions for targeted com- norms. Outside the domain of ‘nudge’, there are alternative munities already long-used to being on the receiving end of rationalities which could drive this which move away from state harm or neglect. the individual as the site of change and incorporate more Despite these concerns, the dangers posed by the state’s liberatory, community-based, solidaristic, and participatory use of targeted ads are accompanied by the dangers of not ideas, repurposing the platform technologies of control for using them. There are a wealth of areas in which targeted bottom-up social change. However it remains to be seen advertising and inﬂuence approaches are being used in what the role of the platforms might be in such a future - co-ordinated campaigns by malicious actors, from the whether infrastructures developed for commercial exploit- spread of illicit cybercrime services, to the targeting of vul- ation can ultimately serve social goods. nerable people with scams, to attempts by far-right, misogy- This paper is, we hope, the starting point for a much nist, racist, and queerphobic groups to spread hateful larger cross-disciplinary research project. Although there narratives and radicalise. There is a compelling argument is a wide and deep seam of research activity which to be made that the state has some duty to either counter addresses improving and measuring the efﬁcacy of different these malicious inﬂuence campaigns directly on the same kinds of behavioural interventions, there is next to none on terms, or to support communities in doing this work them- how these are being combined with digital inﬂuence tech- selves. Where the state averts its gaze intentionally there is nologies and how this hybrid approach is being realised the potential for these inﬂuence infrastructures to operate in the practices and processes of government. There is a unchecked, open as a technology of power to anyone able great deal of further research (both academic and journalis- to pay for adverts or who is able to subvert the algorithms tic) to be done: on the practices and rationalities of commu- (such as the far right communities who attempt to game nications professionals within government, on links to Youtube recommendations). Still, even in cases where private providers, and on what this means for the future there may be a clear moral imperative for the state to of government and law enforcement. Further questions employ targeted advertising and inﬂuence, robust transpar- should be asked about the extent to which the experience ency around the employment of citizen data should be of UK citizens in the online realm is being shaped by gov- prioritised. ernment inﬂuence. The international picture bears substan- Finally, it is important to note that the efﬁcacy of these tial further examination, as do the implications in global ‘inﬂuence’ interventions is extremely difﬁcult to assess. power, ethics, law, and democracy. This particular techno- Evaluation is a serious issue - unlike commercial targeting, logical future has already begun to arrive and the role of the where conversion to sale offers a fairly clear metric, it is academy should not only be to administer it, but to critique often extremely difﬁcult, despite the promises of the and challenge it as well. OASIS model and the access to administrative data pos- sessed by government, to evidence effects robustly. The Acknowledgments reliance on a tracking and targeting infrastructure that is The authors would like to thank the Scottish Centre for Crime and still fairly unreliable means that except for particularly self- Justice Research for their support with this project, particularly evident forms of targeting (such as the NCA’s approach, Alistair Fraser and Rachelle Cobain. which targets people searching for particular illegal ser- vices), many of these messages may be seen by the Declaration of conﬂicting interests wrong people, or not seen by the right people. Where these proﬁles act in the delivery of government policy, it The author(s) declared no potential conﬂicts of interest with respect to the research, authorship, and/or publication of this is crucial to account for the ways in which algorithmic article. bias or inaccuracy might shape who gets which messages, and the consequences of this. Funding The author(s) received no ﬁnancial support for the research, Concluding thoughts and possible futures authorship and/or publication of this article. Although we draw out a critical perspective on these approaches in this article, we do not argue that there are ORCID iD no possible positive futures of targeted messaging. Government will always involve communication, and Ben Collier https://orcid.org/0000-0002-9207-3068 12 Big Data & Society Gane N (2021) Nudge economics as libertarian paternalism. Notes Theory, Culture & Society 38(6): 119–142. 1. https://www.emarketer.com/content/uk-programmatic-digital- Garland D (1997) Governmentality’ and the problem of crime: display-ad-spending. foucault, criminology, sociology. Theoretical criminology 2. https://gcs.civilservice.gov.uk/publications/evaluation- 1(2): 173–214. framework/. Garland D (2001) The Culture of Control: Crime and Social 3. https://gcs.civilservice.gov.uk/publications/strategic- Order in Contemporary Society. Oxford: Oxford University communications-a-behavioural-approach/. Press. 4. https://www.bi.team/wp-content/uploads/2020/02/BIT- Giddens A (2013) The Third way: The Renewal of Social London-Violence-Reduction.pdf. Democracy. Cambridge: Polity Press, John Wiley & Sons. Gofen A, Moseley A, Thomann E, et al. (2021) Behavioural gov- ernance in the policy process: Introduction to the special issue. References Journal of European Public Policy 28(5): 633–657. Aradau C and Blanke T (2017) Politics of prediction: security and Grier S and Bryant CA (2005) Social marketing In public health. the time/space of governmentality in the age of big data. Annual Review of Public Health 26(1): 319–339. European Journal of Social Theory 20(3): 373–391. Grier SA and Kumanyika S (2010) Targeted marketing and public Atkin CK and Wallack L (1990) Mass Communication and Public health. Annual Review of Public Health 31(1): 349–369. Health: Complexities and Conﬂicts (Vol. 121). California: Hall S, Critcher C, Jefferson T, et al. (1978) Policing the Crisis: Sage. Mugging, the State and law and Order. Macmillan, London: Baggio M, et al. (2021) The evolution of behaviourally informed Macmillan International Higher Education. policy-making in the EU. Journal of European Public Policy Halpern D (2015) Inside the Nudge Unit: How small changes can 28: 1–19. make a big difference. London, UK: Random House. Barberis P (1998), The new public management and a new Hausman DM and Welch B (2010) Debate: to nudge or not to accountability. Public Administration 76(3): 450–471. nudge. Journal of Political Philosophy 18(1): 123–136. Coates AE, Hardman CA, Halford JC, et al. (2019) Social media Heath-Kelly C (2017) The geography of pre-criminal space: epidemio- inﬂuencer marketing and children’s food intake: A randomized logical imaginations of radicalisation risk in the UK prevent strategy, trial. Pediatrics 143(4): 1–9. 2007–2017. Critical studies on terrorism 10(2): 297–319. Collier B, Thomas DR, Clayton R, et al. (2021) Inﬂuence, infra- Hinton E and Cook D (2020) The mass criminalization of black structure, and recentering cybercrime policing: Evaluating Americans: A historical overview. Annual Review of Criminology emerging approaches to online law enforcement through a 4: 261–286. market for cybercrime services. Policing and Society:1–22. Jones R and Whitehead M (2018) “Politics done like science”: Cook J, Lynes J and Fries S (2021) Exploring mistakes and fail- critical perspectives on psychological governance and the ures in social marketing: The inside story. Social Marketing experimental state. Environment and Planning D: Society Quarterly 27(1): 13–31. and Space 36(2): 313–330. Crain M (2019) A critical political economy of Web advertising Katz J, Peberdy A and Douglas J (2000) Promoting Health: history. In: 2019. The SAGE Handbook of Web History.55 Knowledge and Practice, 2nd edition. Milton Keynes: The City Road, London: Sage Publications Ltd, 330–343. Open University, in association with Palgrave. Crawford A (2006) Networked governance and the post- Kostygina G, et al. (2020) Boosting health campaign reach and regulatory state? Steering, rowing and anchoring the provision engagement through Use of social Media inﬂuencers and of policing and security. Theoretical Criminology 10(4): 449– memes. Social Media + Society 6(2): 2056305120912475. 479. Leggett W (2014) The politics of behaviour change: Nudge, neo- Duffy BE, Pinch A, Sannon S, et al. (2021) The nested precarities of liberalism and the state. Policy & Politics 42(1): 3–19. creative labor on social Media. Social Media+ Society 7(2): : 1–12. Lepenies R and Małecka M (2018) The ethics of behavioural public Ewert B (2020) Moving beyond the obsession with nudging indi- policy. In: Lever A and Poama A (eds) The Routledge Handbook vidual behaviour: Towards a broader understanding of of Ethics and Public Policy. Abingdon, Oxon: Routledge, 513– Behavioural Public Policy. Public Policy and Administration 525. https://doi.org/10.4324/9781315461731-41 35(3): 337–360. Loader I (2006) Fall of the ‘platonic guardians’ liberalism, crimin- Feitsma J (2019) Brokering behaviour change: the work of behav- ology and political responses to crime in England and Wales. ioural insights experts in government. Policy and Politics British Journal of Criminology 46(4): 561–586. 47(1): 37–56. Lupton D (1995) The Imperative of Health: Public Health and the Foucault M (1982) The subject and power. Critical Inquiry 8(4): Regulated Body. London: Sage. 777–795. Mbembe A (2008) Necropolitics. In: Foucault in an Age of Terror. Fraser A (2017) Gangs & Crime: Critical Alternatives. London: London: Palgrave Macmillan, 152–182. Sage. Michie S, van Stralen MM and West R (2011) The behaviour Fraser A and Atkinson C (2014) Making up gangs: looping, label- change wheel: A new method for characterising and ling and the new politics of intelligence-led policing. Youth designing behaviour change interventions. Implementation Sci Justice 14(2): 154–170. 6(1): 1–12. Gandy OH and Nemorin S (2019) ‘Toward a political economy of Mols F, et al. (2015) Why a nudge is not enough: A social identity nudge: Smart city variations’, information. Communication & critique of governance by stealth. European Journal of Society 22(14): 2112–2126. Political Research 54(1): 81–98. Collier et al. 13 Ofcom (2020) Media Nations 2020: UK report, p. 101. Available Thaler RH and Sunstein CR (2009) Nudge: Improving at: https://www.ofcom.org.uk/__data/assets/pdf_ﬁle/0010/200 Decisions=About Health, Wealth, and Happiness. London: 503/media-nations-2020-uk-report.pdf. Penguin. Rose JW (2000) Making” Pictures in our Heads": Government WARC (2021) Advertising Association/WARC Expenditure Advertising in Canada. Greenwood, Connecticut: Greenwood Report http://expenditurereport.warc.com/. Publishing Group. West SM (2019) Data capitalism: redeﬁning the Schmidt AT and Engelen B (2020) The ethics of nudging: An logics of surveillance and privacy. Business & society 58(1): overview. Philosophy Compass 15(4): e12658. 20–41. Schubert C (2017) Exploring the (behavioural) political economy of Wallack L (2002) Public health, social nudging. Journal of Institutional Economics 13(3): 499–522. change, and Media advocacy. Social Marketing Quarterly Srnicek N (2017) Platform Capitalism. Cambridge, UK: Polity Press. 8(2): 25–31. Sunstein CR (2016) The Ethics of Inﬂuence: Government Zuboff S (2015) Big other: Surveillance capitalism and the pro- in the Age of Behavioral Science. Cambridge: Cambridge spects of an information civilization. Journal of Information University Press. https://doi.org/10.1017/CBO9781316493021 Technology 30(1): 75–89.

Journal

Big Data & Society – SAGE

Published: Feb 24, 2022

Keywords: Targeted advertising; government; power; marketing; criminal justice; public policy

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

Influence government: Exploring practices, ethics, and power in the use of targeted advertising by the UK state:

References (63)

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies