What is policy assemblage?

ABSTRACT

Assemblage thinking has exploded in policy research, especially among scholars working in the policy mobilities field who are seeking to harness the potential of an assemblage approach to understand how policies move, mutate and manifest in increasingly transnational contexts. The ubiquity of assemblage, however, does not always render it clear, with the concept being variously defined and sometimes lacking conceptual strength and explanatory power. This paper seeks to conceptualize and defend an assemblage approach to policy analysis. By synthesizing core threads from existing literature, it identifies three theoretical and conceptual foundations central to a ‘policy assemblage’ approach: (1) relations of exteriority and emergence; (2) heterogeneity, relationality and flux; and (3) attention to power, politics and agency. Together, these foundations signal a coherency to assemblage thinking and suggest an assemblage approach has powerful potential, allowing researchers to see and explain things in ways that many established traditions in policy research do not. By identifying foundations and offering examples of how each might be mobilized, the paper provides the beginnings of a framework for policy assemblage research not previously articulated in a systematic form, thus inviting further discussion about what it means to undertake policy assemblage research.

KEYWORDS:

• assemblages
• public policy
• mobility
• power

INTRODUCTION

Assemblage thinking has exploded. Theoretically, conceptually and methodologically, scholars in a wide variety of fields are seeking to harness the potential of an assemblage approach to generate insights into a diverse range of social formations. Framing assemblages as relational constructs, comprised of heterogeneous and emergent component parts that are arranged together towards certain strategic ends, in particular spaces and times, scholars have engaged assemblage thinking to make sense of an array of phenomena, ranging from the formation of cities (McCann & Ward, 2011), international relations (Acuto & Curtis, 2014) and various ‘global assemblages’ (Ong & Collier, 2005), to specific policy formations such as pig farming practices in the European Union (Dunn, 2005), forest management practices in Indonesia (Li, 2007), and ‘creative industries’ policies in New Zealand (Prince, 2010). Scholars have anchored their scholarship in a related set of foundational theories. Central has been the philosophical works of Gilles Deleuze and Félix Guattari, who used the concept of assemblage (agencement in French) to understand the heterogeneous composition of complex social and non-social formations. Bruno Latour’s articulation of social assemblages in the field of actor–network theory has also featured regularly (Latour, 2005), as has Manuel DeLanda’s ‘re-constructed’ theory of assemblages, which builds on Deleuze and Guattari but also draws strongly upon insights from complexity theory (DeLanda, 2006, 2016).

The amplified interest in assemblage thinking has generated both passionate interest in exploring and further clarifying the potential of an assemblage approach (Anderson & MacFarlane, 2011; Baker & McGuirk, 2017) and has also inspired critique, ranging from those who have questioned the coherency and utility of assemblage (e.g., Brenner, Madden, & Wachsmuth, 2011) to those who defend the concept but argue contemporary articulations have unhelpfully strayed from the foundations established by key scholars such as Deleuze and Guattari (Buchanan, 2015; Nail, 2017). At the same time, assemblage has been connected to a range of similar but distinct concepts, including the Foucauldian concepts of dispositif and apparatus (Legg, 2011; Li, 2007), and other concepts in research focused on transnational networks, mobilities and topological understandings of space and power (Allen & Cochrane, 2010; Prince, 2017).

Set against the backdrop of this broader explosion in assemblage thinking, this paper takes as its object of analysis a body of scholarship that has sought to harness assemblage for the purposes of understanding contemporary policy processes: that is, processes of policy assemblage. More specifically, its interest lies in synthesizing and building on scholarship in the fields of critical geography, critical policy studies, sociology and anthropology, which has positioned assemblage as part of the emerging field of policy mobilities research. While this work is varied in form, intent and empirical foci, it is unified by a common aim and challenge to develop more nuanced approaches to understanding how policies move, mutate and manifest in particular spaces and times, in a context of intense transnational flows of policy ideas and practices (Gulson et al., 2017; McCann & Ward, 2012; Peck & Theodore, 2015). Assemblage has been positioned as a generative tool for addressing the limits of established debates and concepts, especially those relating to policy transfer, borrowing and diffusion; but it has also been framed as a corrective to rational–technical, institutionalist and state-centric accounts of policy and governance processes (Clarke, Bainton, Lendvai, & Stubbs, 2015; McCann & Ward, 2012; Peck & Theodore, 2015; Shore, Wright, & Pero, 2011; Temenos & McCann, 2013; Ureta, 2015). In some cases, assemblage is used as part of a three-part distinction between what has been termed the policy ‘assemblage, mobilities and mutation’ approach (McCann & Ward, 2013). Other work has focused on bringing together assemblage with the related concept of ‘translation’ to capture the processes of translation that policies undergo as they travel and are reassembled in new locations (Clarke et al., 2015). In other cases, an interest in mobilities and translation has been subsumed into an assemblage lens (Savage & Lewis, 2018).

Despite the ballooning popularity of assemblage in policy research and the horizons of hope it promises to offer policy researchers, the concept remains variously defined and, in some cases, lacks conceptual or methodological precision (Baker & McGuirk, 2017; Savage, 2018). Moreover, as Allen (2011) argues, there is also a strong tendency for assemblage to be used in overly descriptive ways, leading, at worst, to analysis that represents ‘a simple joining up exercise’ (p. 156) – or what he terms ‘thin’ and ‘endless description’ (p. 154). In other words, an assemblage approach risks becoming a justification for (and process of) just mapping out and describing various bits and pieces of policy, but in a way that offers ‘weak conceptualisation’ (p. 154) and little in terms of explaining phenomena in new ways or advancing useful normative arguments (Brenner et al., 2011). Even worse, as the concept balloons in popularity and diversity of usage, it risks emerging as an empty signifier, meaning anything and everything at the same time (Anderson & MacFarlane, 2011, p. 125), and potentially in ways that are de-anchored from work that has sought to articulate carefully the strengths and limits of the concept. Ultimately, therefore, the ubiquity of assemblage does not necessarily render it clear. This is not to suggest assemblage should have one fixed meaning or be anchored exclusively in the work of particular scholars (Allen, 2011; Anderson & MacFarlane, 2011), but if it lacks coherent and ‘thoughtful conceptualisation’ (Allen, 2011, p. 156), then it is difficult for scholars to articulate, defend and extend its use productively in policy research.

In the light of these well-established concerns, this paper seeks to conceptualize and defend an assemblage approach to policy analysis by demonstrating that assemblage does indeed have strong generative potential for policy researchers, especially those working within the policy mobilities field. To address concerns about a lack of conceptual precision and explanatory power, this paper synthesizes existing literature and identifies three core foundations that are theoretically and conceptually central to a policy assemblage approach: (1) relations of exteriority and emergence; (2) heterogeneity, relationality and flux; and (3) attention to power, politics and agency. Together, these foundations signal a coherency to assemblage thinking and suggest an assemblage approach has powerful potential, allowing researchers to see and explain things in ways that many established traditions in policy research do not. By identifying foundations and offering examples of how each might be mobilized, the paper provides the beginnings of a framework for policy assemblage research not previously articulated in a systematic form, thus inviting further discussion about what it means to undertake policy assemblage research. In a sense, therefore, the aim is to do, in theoretical and conceptual terms, what Baker and McGuirk (2017) have recently attempted in methodological terms by outlining what they see as core elements of a methodological framework for assemblage research. In another way, the analysis also seeks to provide, but in more detail, an articulation of the commonalities that exist across assemblage research, in ways that extend Anderson and MacFarlane’s (2011) work, but with a specific focus on policy research (i.e., rather than critical geography specifically). Ultimately, the hope is that this paper will fruitfully respond to Allen’s (2011) concerns by showing that assemblage does indeed ‘allow us to do certain things and enable us to think in certain ways that were not possible before’; and can therefore amount to much more than ‘a passing fad or intellectual fashion’ (p. 154).

POLICY ASSEMBLAGE: THREE CORE FOUNDATIONS

While all theories and concepts are necessarily incomplete and are ideally creative works in motion towards greater explanatory power, it is nevertheless important for scholars working with shared theories and concepts to seek some measure of shared understanding about the foundations that broadly define their approach. This is not only important for forging a common and more refined language, but also for driving scholarship forward through providing analytical structures that can be targeted, challenged and revised. As the use of assemblage rapidly expands in policy research, a definable field of ‘policy assemblage’ research is emerging (Savage, 2018). Yet, while a ‘broad consensus’ (Baker & McGuirk, 2017, p. 428) exists regarding the usefulness of assemblage thinking for policy research, there is not any systematic attempt to take stock of ‘the remarkable diversity with which the term has come to use’ (Anderson & MacFarlane, 2011, p. 124). In what follows, this paper seeks to generate the beginnings of a framework for policy assemblage research by arguing that the following three core foundations are central to a policy assemblage approach.

Relations of exteriority and emergence

The concept assemblage reflects a distinctive understanding of the relationship between parts and wholes, which is frequently articulated through what has been termed ‘relations of exteriority’. This unique take on the relationship between parts and wholes reflects core tenets of complexity theory concerning the concept of ‘emergence’, and is reflected either implicitly or explicitly in nearly all policy research that adopts an assemblage analytic. Recognizing relations of exteriority and emergence has significant implications for how we might understand processes of policy assemblage, as well as broader political systems and other social formations.

DeLanda’s (2006, 2016) ‘reconstructed theory of assemblages’ (DeLanda, 2006, p. 4) builds directly on the work of Deleuze and Guattari and provides one of the most fleshed-out examinations of relations of exteriority and associated implications for understanding the relationship between parts and wholes. While DeLanda does not focus on policy in an explicit sense, his analysis of nations, governments, organizations and networks translates powerfully into a discussion of how the concept assemblage can inform policy research. For this reason, his insights have been taken up by a number of policy scholars working with an assemblage approach. Central to DeLanda’s articulation of assemblage is a critique of what he terms ‘the organismic metaphor’. Drawing strongly on complexity theory, DeLanda argues that social scientists too often treat social formations as akin to biological organisms that have internal logics and order, and, as a result, frame the component parts of social formations as akin to ‘bodily organs’ (p. 8). Such perspectives, he argues, produce a skewed understanding of the relationship between parts and wholes: or, what he terms ‘the micro- and the macro-levels of social reality’ (p. 4).

This ‘micro-macro problem’ (p. 4), DeLanda argues, is plagued by two contrasting forms of reductionism. First, there is a tendency towards micro-reductionism, whereby wholes are framed as ‘a mere aggregate’ (p. 4) of individual component parts. The problem with understanding the whole as simply the sum of its parts is that we obscure the emergent and ‘irreducible properties’ (p. 10) of the whole, that is, properties that exist only as a result of contingent interactions taking place between component parts (i.e., properties that would be different if the components were arranged differently). This strongly mirrors theories of emergence in the field of complexity theory, which highlight the irreducible properties of complex systems, stressing that the nature of systems can only be understood with reference to how its constituent parts relate and generate particular emergent features (e.g., Geyer & Rihani, 2010; Urry, 2003). As DeLanda writes, the properties of a whole ‘cannot be reduced to those of its parts’, because ‘they are the result not of an aggregation of the components’ own properties but of the actual exercise of their capacities’ (DeLanda, 2006, p. 11; emphasis in the original). He adds: ‘These capacities do depend on a component’s properties but cannot be reduced to them since they involve reference to the properties of other interacting entities’ (p. 11). In sum, rather than understanding a policy as coherent thing or as definable as the sum of its constitutive components, an assemblage approach stresses that what is most important is understanding the nature of interactions between components and the capacities such components exhibit when arranged in different ways. This, in turns, draws one’s attention to understanding processes of arrangement and the power relations that make some arrangements possible, and others not. For example, the formation of a national policy typically rests on the novel arrangement of a diverse range of potential component parts (e.g., laws, actors, organizations, technologies of governance, accountability processes, etc.) that must be harnessed and arranged together in a way that encourages the policy to serve its intended functions and operate in specific ways. The particular ways in which components are brought together will determine the properties and effects of any given policy or agenda; and if the very same components were to be arranged differently, or new components were introduced or excluded, then different properties and effects would be produced. As Rabinow (2014) argues, the particular ways that autonomous components are brought together in an assemblage will inevitably make ‘some things and events possible and others improbable’ (p. 206). For this reason, as Li (2007) argues, an assemblage approach invites one to ask how certain policies are ‘made to cohere’ (Li, 2007), while, at the same time, potentials for doing policy otherwise are denuded. As such, this approach opens windows for imagining policy differently (and with potentially normative political aims in mind) in terms of considering how new assemblages might be forged with different and potentially more positive impacts and possibilities (i.e., new and different forms of emergence) (e.g., Tampio, 2009).

The second tendency DeLanda critiques is macro-reductionism, which, converse to micro-reductionism, frames individual components of an assemblage as being ‘mere products’ (DeLanda, 2006, p. 4) of the whole. Here, the reverse problem is created, whereby individual parts of a social formation are seen as primarily determined by the nature of the formation itself. This is a common problem in political science and sociology, whereby scholars might understand specific components of a political system or policy as primarily the result of that system or policy. For example, a researcher might frame one key component of a national system (e.g., a particular law, policy, organization or actor) as primarily being a product of that system and thus as having certain features or operating in certain ways based on views or assumptions about the broader nature of that system. While this might at first seem logical, it is in fact deeply problematic, especially in contemporary contexts in which policy formations are increasingly informed by transnational policy mobilities that cut across political territories in new ways. As Savage and Lewis (2018) argue in relation to national schooling reforms in Australia, while certain policies might ostensibly be national and Australian in terms of their legislative scope and territorialization as implementable technologies of governance, it is misguided to ‘explain’ such policies as simply ‘products’ of Australian schooling policy. Indeed, by adopting an assemblage approach to tracing policy development, they demonstrate how the creation of national teaching standards in Australia was the result of a diverse number of component parts that were strongly informed by transnational flows of policy actors, ideas and practices, which have manifested in place-specific ways. In other words, policy ideas, practices and forms of influence might be strongly informed by transnational flows, but the conditions of possibility for such policies depends largely on local conditions of possibility. As a result, such policies, and the core components which make such policies real, can partially but not fully be understood as artefacts of national political and policy contexts.

The primary implication of rejecting both macro and micro reductionism is that rather than understanding policy assemblages as constituted by relations of interiority, whereby component parts have a necessary or essential relationship to each other, or ‘form a seamless whole’ (DeLanda, 2006, p. 4), assemblages are instead characterized by relations of exteriority, in which ‘a component part of an assemblage may be detached from it and plugged into a different assemblage in which its interactions are different’ (p. 10). As such, rather than relations between component parts being understood as logically necessary to make the whole what it is, relations in an assemblage are instead seen as ‘only contingently obligatory’ (p. 11; emphasis in the original). Again, this strongly mirrors the anti-reductionism and commitment to holism in complexity theory (e.g., Urry, 2003). DeLanda (2016) has furthered this idea in recent work, arguing:

Unlike wholes in which ‘being part of the whole’ is a defining characteristic of the parts, that is, wholes in which the parts cannot subsist independently of the relations they have with each other (relations of interiority), we need to conceive of emergent wholes in which the parts retain their autonomy, so that they can be detached from one whole and plugged into another one, entering into new interactions. (p. 10)

This is also stressed by Rabinow (2014), who argues, an assemblage is ‘not a pre-existing thing of the world with pre-given properties’, but instead ‘brings together entities in the world into a proximity in which they establish relations between an among themselves while remaining external to each other and thereby retaining their original properties’ (p. 206).

Adopting an anti-reductionist position that understands assemblages as defined by relations of exteriority and as generative of emergent properties has a large number of implications for contemporary policy scholars. For example, in addition to the implications already canvassed, by resisting both macro- and micro-reductionism, assemblage theory poses a fundamental challenge to the traditional structure/agency distinction, which has a number of associated implications for doing policy research. For example, accounts that privilege agency over structure tend towards micro-reductionism, whereas accounts that privilege structure over agency favour macro-reductionism. By rejecting both, we are forced not only to question the relevance of the structure/agency binary but also to rethink a range of policy phenomena, such as how individuals are shaped by policies or societies, how we understand the power of policy actors and organizations, how micro-level formations might contribute to broader meso- or macro-level trends, and much more. As Urry (2003) writes in relation to similar dynamics in complexity theory, such approaches to understanding social formations ‘subvert this very distinction between agency and structure’ (p. 111). This also has major implications for how we understand policy impacts and outcomes. For example, an antireductionist approach means we cannot assume any linear or straightforward relationship exists between policies and the impacts that emerge in contexts where such policies are put to work. Put differently, a policy might seek to engender certain effects, but the extent to which it does or does not can rarely (if ever) be attributed solely to the policy itself. Moreover, policies can (and often do) produce impacts that can either not be predicted in advance (Ureta, 2015) or which, at the very least, can only at best be understood by examining how the policy itself interacts with other component parts to produce certain emergent features. Drawing attention to these complexities, and the social worlds that contour processes of policy enactment, vastly complicates core assumptions that continue to underpin the vast majority of research conducted in the fields of policy implementation and evaluation. Indeed, we see that such research suffers deeply from evaluation models that fundamentally rely on forms of reductionism (often through bracketing out from view the social life of policy and/or the complexities of components that a policy intervention intervenes with) in order to isolate the policy for the purposes of understanding impacts and effects (Geyer & Rihani, 2010).

Complicating the interrelationship between policy and impact has significant implications when viewed through the prism of the contemporary debates about evidence-based policy, policy-borrowing and the fascination amongst policy-makers and researchers with identifying ‘what works’ approaches to policy problems (e.g., Lewis, 2017; Parkhurst, 2017; Peck & Theodore, 2015). For example, understanding policies as contingent assemblages, defined by relations of exteriority and rendered place-specific in terms of form and impact, fundamentally challenges assumptions that ‘best practice’ examples of policies might simply be borrowed from one policy context and implemented in another with the same or similar impacts. In other words, just because a crime prevention policy in Finland might contain certain components that are arranged in particular ways and operate with positive impacts in that context, this does not mean these components will have the same positive effects in Singapore or the United States. This is because when such policy components are assembled in a new context, the components themselves will be contoured by numerous context-dependent factors (i.e., specific conditions of possibility) which will render the components place specific and result in new relations being established and maintained between these components and existing components in the new environment. We see, therefore, that policies undergo forms of mutation, translation and re-assemblage as they travel between different policy contexts (McCann & Ward, 2013). This is a defining argument in the emerging field of policy mobilities, and one of the reasons why an assemblage approach has gained such traction in this field, especially over the past decade. Rather than asking ‘what works’, therefore, perhaps the best policy-makers can hope to ask is: what might work here if policy is adopted and adapted to local contexts in a way that remains cognizant of the multiple components and context-specific factors that need to be considered and strategically arranged to render the policy workable? This more modest form of questioning recognizes that ‘much cannot be bottled for export’ (Peck & Theodore, 2015, p. xvii). It also reminds one that ‘while the policymaking imagination may be globalizing, and while transnational circuits of expertise and practice are proliferating, the stubborn reality is that making policies work very often remains a hands-on, messy, and very much “local” affair’ (Peck & Theodore, 2015, p. xvii). Of course, the likelihood of such nuanced questioning and consideration of context gaining widespread traction in the policy making world is low, as the very allure of the ‘what works’ approach is its promise of workable solutions, not the problematization of solutions and introduction of uncertainty.

Heterogeneity, relationality and flux

Assemblages are heterogeneous, comprised of a multiplicity of component parts that have been arranged together towards particular strategic ends. Given the aforementioned commitments to relations of exteriority, anti-reductionism and the rejection of ‘coherent wholes’, the heterogeneous component parts that constitute an assemblage are also understood to have a contingent rather than necessary relationship, brought together into particular relational configurations which have mutable rather than fixed forms. This means an assemblage approach emphasizes the always moving and evolving nature of social formations. The complex relationships between heterogeneity, relationality and flux – and how these features can inform an assemblage approach to policy analysis – require careful unpacking.

To begin, it is crucial to note that despite being heterogeneous and synthetic in form, assemblages are not simply a random assortment of things. Assemblages, are not, therefore, akin to a junk draw of bits and pieces that just happen to exist together. Assemblages are not anything and everything. Instead, as Ureta (2015) argues, assemblages are the result of heterogeneous elements that are brought together into particular strategic relations and with particular desired impacts. Drawing on Law’s (1994) notion of ‘modes of ordering’, Ureta’s conceptualization of policy assemblage pays close attention to the dynamic processes by which heterogeneous elements come together to transform existing arrangements into something new. Modes of ordering, Ureta (2015) argues, ‘are heterogeneous and variable but always include the search for strategic effects, the aim to transform an existing situation in a certain predetermined way through the establishment of particular sets of relations between new and existing entities’ (p. 12). Rather than being an assortment, therefore, a policy assemblage can be understood as an arrangement of components in particular ways with the aim of governing conduct (see Buchanan, 2017; and Nail, 2017, for a discussion of the term ‘arrangement’ and how it relates to the use of assemblage in Deleuze and Guattari’s work). Therefore, for example, while within a nation there exists multifarious policy and governance components, the mere existence of these components does not make an assemblage. Instead, an assemblage refers to cases where components have been strategically arranged with the view to forming an apparatus for governing. This conceptualization of assemblage reflects Li’s (2007) argument that the forging of an assemblage is the result of ‘hard work required to draw heterogeneous elements together, forge connections between them and sustain these connections in the face of tension’ (p. 264). Again, however, while such processes of arrangement involve the forging of various alignments between heterogeneous parts, such alignments do not mean the assemblage has a coherent essence or singular rationality (p. 265). Instead, an assemblage approach draws attention to the ways in which heterogeneous formations hold together, ‘without actually ceasing to be heterogeneous’ (Allen, 2011, p. 154). Forms of coherency are thus established out of multiplicity. Put differently, a policy assemblage has no definable essence beyond the relations established between its components. This does not mean the assemblage lacks coherency, but it does mean it lacks an essence. This distinction between coherency and essence is fundamental to understanding how a policy assemblage can operate with strategic and observable impacts, but without a singular guiding rationale, which, in turn, has major implications for how we understand power (see next section). As Deleuze (2002) argued, ‘the assemblage’s only unity is that of co-functioning: it is a symbiosis. … It is never filiations that are important, but alliances, alloys’ (p. 52). A focus on alloys, co-functioning and synthetic forms further reinforces the distinction between relations of interiority and exteriority. For example, unlike a bicycle, for which the component parts (e.g., chain, gears, pedals, etc.) serve a strategic function in ways that are logically necessary to make the whole what it is, relations in a policy assemblage also serve strategic functions, but are comprised of contingently obligatory relations that are always evolving in form.

Recognizing the mutability of assemblages requires close analytical attention to processes of flux. At the ontological level, therefore, assemblage thinking reflects a belief that kinesis (i.e., movement of varying forms and paces) is a fundamental to the nature of social reality. Nail (2015) describes the concept of kinesis in relation to what he terms an ontology of movement, which, in short, means understanding society and its various forms as ‘always in motion’ (p. 4). He adds: ‘Societies are not static places with fixed characteristics and persons. Societies are dynamic processes engaged in continuously directing and circulating social life. In a movement-oriented philosophy there is no social stasis, only regimes of social circulation’ (p. 4). We must, therefore, ‘understand society itself according to movement’ (p. 4; emphasis in the original). When related to assemblage and policy analysis, this necessitates an analytical approach that is not only sensitive to how multiple component parts are brought together into coherent and strategically oriented technologies of governance (i.e., assembled) but is also attentive to the many ways that policies are subject to forms of disruption and change (disassembled or reassembled). Indeed, a number of policy scholars have drawn attention to such processes of re/dis/assembly (e.g., Savage & Lewis, 2018; Youdell & McGimpsey, 2015). In doing so, attention has been drawn to the complex processes through which policies come into being but are also (often simultaneously) dismantled and reassembled into new forms. Savage and Lewis (2018), for example, use the terms ‘assembly’, ‘disassembly’ and ‘reassembly’ in ways that build directly on Deleuze and Guattari’s concepts of territorialization, deterritorialization and reterritorialization. As they note, Deleuze and Guattari used these terms to describe how assemblages come together (territorialization), come undone (deterritorialization), or to understand cases where existing or disrupted assemblages are reassembled (reterritorialization) (e.g., Deleuze & Guattari, 1987). Assemblages always include, therefore, a complex interplay between territorializing, deterritorializing and reterritorializing tendencies, with some components working to stabilize the assemblage and others working to transform it (Ureta, 2015, p. 12). As McCann and Ward (2012) put it, ‘an assemblage is always in the process of coming together  …  just as it is always also potentially pulling apart’ (p. 328, emphasis in original).

Again, when understood against the backdrop of an ontology of motion, this means an assemblage approach rejects the proposition that policies or political systems (or any other social formation) can ever be wholly static in form. Emphasizing perpetual flow and simultaneity of re/dis/assembly, however, does not mean policies are never formed (e.g., laws, standards, accountability systems, etc. do obviously ‘exist’) or that policies cannot maintain periods of stability (i.e., temporary fixity). Instead, it means that: (1) from the moment any policy is formed, it is already subject to forms of disruption, challenge and multiple interpretations; (2) when enacted, a policy often takes on highly varied forms, with impacts that are both partially predictable and unpredictable; and (3) all policies will eventually come undone (i.e., disappear or change form). Policies never exist, therefore, as one ‘complete’ thing, time immemorial, but instead are always subject to multiple and evolving interpretations, enactments and (very often) reforms or discontinuation. To borrow from Deleuze, therefore, we see, ‘the current is not what we are but rather what we are in the process of becoming’ (Deleuze, 1992, p. 164). Temporal considerations are thus crucial. As Savage and Lewis (2018) argue, while some policies ‘might give an illusion of stability, a wider view ultimately reveals constant motion and new beginnings’ (p. 124, emphasis in original). Even when a policy might seem all-encompassing in its forms and effects, therefore, a wider historical view always reveals that policy to be an artefact of the times, which was always going to change. A number of other policy scholars have emphasized such temporal flux in adopting an assemblage approach. Ureta (2015), for example, argues that policies are not ‘solid or stable’, but are ‘temporary concatenations of heterogeneous entities, always on the verge of becoming something completely different’ (p. 12). Or, as Tampio (2009) suggests, an assemblage ‘perpetually transforms itself, like a cloud that pulls together and loses water molecules’ (p. 394). Policies are, in this sense, ‘always in-the-making, always on-the-move’ (Kingfisher, 2016, p. 14). Indeed, the holding together of a policy in a coherent form and for an extended period is often the exception that proves the rule, given the consistent pressures and contestations to which policies are inevitably subject (Shore et al., 2011, pp. 1–3); especially in contemporary contexts in which reform fever is rife.

The theoretical foundations established so far, especially concerning relations of exteriority, heterogeneity and flux, bring one to another core foundation of assemblage theory: a commitment to understand forms of relationality. Indeed, centrally connected to (and part of) an ontology of movement is a relational ontology: which, in the context of policy research, means seeing the relations established between policy components as just as (if not more) fundamental to understanding policies as the components themselves. At the very least, an assemblage approach requires a commitment to some form of relational thinking and both analytical and methodological approaches that give primacy to understanding relations between components as a core part of policy analysis work. An approach to policy analysis that claims to adopt an assemblage approach, but which does not reflect relational thinking or associated analytical approaches, can hardly be understood as an assemblage approach, as such an approach would either grate uncomfortably against (or be entirely incompatible with) core tenets of assemblage theory. As Bueger argues, central to an assemblage approach is, ‘a relationalist understanding of reality’ (Bueger, 2014, p. 62, citing Hayden, 1995), which understands ‘relations are not fixed and stable’, but instead ‘are emergent and enacted’ (p. 62). Relations, he argues ‘are made and re-made in practices’, which means what is required of researchers is, ‘study of the practical work needed to generate relations between the elements of an assemblage’ (p. 62). Ong (2014) makes a similar argument, stressing the benefits of assemblage thinking in terms of eschewing forms of abstraction and reification in favour of focusing on the actual material practices, things and relations through which social formations come into being (p. 24). An assemblage approach to policy analysis thus ‘directs our attention away from theoretical abstractions and ideal types, which are rife in political science and public policy studies, towards more materialist, relational, and bottom-up orientations that seek to understand the tangible stuff of policies’ (Savage, 2018, p. 310; emphasis in the original).

Attention to forms of relationality is a core reason why assemblage approaches have gained such traction in the policy mobilities field, especially among critical geographers and sociologists working with a topological lens (e.g., Allen, 2011; Allen & Cochrane, 2010; Amin, 2002; Hartong, 2018; Lewis & Lingard, 2015; Prince, 2017). Topological accounts seek to disrupt and re-render dominant narratives about scale, local–global relations, the exercise of power and other dimensions central to theorizing the new spatialities of globalization. By rejecting the Euclidean notion of space as a set of a priori ‘fixed coordinates’ upon which political, power and power relations play out, topological accounts instead stress the importance of relations established across space, that is, the creation of relational spaces that bring together the near and the far into new assemblages that cannot be understood in term of fixed notions of territorial scale (Allen & Cochrane, 2010). As Prince (2017) argues, a topological approach ‘emphasizes relationality rather than proximity’, which means ‘elements can be topologically close, even if they are topographically distant’ (pp. 337–338). Topological assemblages are thus constituted by ‘overlapping near–far relations and organisational connections that are not reducible to scalar spaces’ (Amin, 2002, p. 386); or, as Thompson and Cook (2015) put it: ‘A topology’s character is given by the qualities of the connections binding its elements and not its position in abstract external space–time’ (p. 734). New topological relations, which according to Lury, Parisi, and Terranova (2012) define contemporary culture and globalization, are serving powerfully to remake the conditions of possibility for policy, power and governance. Allen and Cochrane (2010) describe this in terms of new ‘power-topologies’, which Lewis and Lingard (2015) argue are central to the assemblage of ‘new geographies of power and possibilities for action’ (p. 624).

When used together, assemblage and topology are highly generative concepts for doing policy mobility research (Prince, 2017). For this reason, both concepts have been harnessed in recent education policy research, which has focused on the role of the Organisation for Economic Co-operation and Development (OECD) in generating a new global space of educational measurement and governance (Hartong, 2018; Lewis & Lingard, 2015; Lingard, Sellar, & Savage, 2014; Savage & Lewis, 2018). Through a vast array of new metrics, standardized assessments, league tables and evidence repositories that claim to offer evidence about ‘what works’ to improve education systems globally, the OECD has brought geographically distant and diverse political systems into a new transnational field of commensurability; thus creating a new closeness between education systems both near and far in ways that are having significant implications for how national and subnational education systems are governed, and how core policy definitions are understood and debated (e.g., including guiding concepts such as student achievement, equity, quality, and the relationship between education and economic growth). As a result, we can now ‘speak of local locals and non-local locals’ (Lingard et al., 2014, p. 721) in respect of the topological relations established through comparative measures. The OECD, therefore, serves as an exemplar of Prince’s (2017) argument that the ‘technocracy’ – that is, ‘the technical experts who produce ostensibly neutral and objective knowledge of objects like the economy in the form of universal measures of economic performance’ (p. 338) – is an increasingly powerful force in bringing nations into new topological assemblages, ‘often in the form of a ladder with the “best” at the top and the “worst” at the bottom’ (p. 339). Laddering nations in this way allows for new forms of global benchmarking, furthers the allure of evidence-based policy making and the ‘what works’ approach, and makes possible new transnational policy networks as policy-makers seek to borrow and learn from nations ‘further up’ the ladder. Topological assemblages, therefore, are facilitating the production of new global imaginations and, as a result, new modes of governmentality (Ruppert, 2012). Again, returning to relations of exteriority, when seeking to understand topological assemblages, what matters most is seeking to understand the nature and structure of connections between component parts. In other words, attention is need to how connections are made, what these connections look like, what is connected to what, and what these connections do (i.e., make possible, or not). As such, how a component of an assemblage is positioned in relation to other component parts of the assemblage (and what its role/potentiality is) becomes potentially more important than ‘where’ it is located in a topographical sense. Doing such work requires keen attention to issues of power, politics and agency, to which the paper now turns.

Attention to power, politics and agency

To focus on policy assemblage is to examine how multiple heterogeneous components are arranged to create governable forms. Through strategically harnessing the relational capacities of multiple component parts, assemblages represent a gathering together of political imaginations, rationalities, technologies, infrastructures and agents towards steering individuals and groups in particular directions. Yet, given the heterogeneous and emergent form of assemblages, and the complexities of understanding and analysing relations of exteriority, assemblage theory offers both highly complex yet potentially very productive ways of understanding power, politics and agency; and the context-dependent ways these forces result from, and contribute to, the making of policy.

With roots in Deleuzian theory, which was in close conversation with Foucauldian theory (Legg, 2011; Tampio, 2009), assemblage thinking presents a view of power as immanent, capillary-like and relational in nature. Power, therefore, is not seen to exist ‘somewhere’ in particular (i.e., in one fixed place), but instead is everywhere, always flowing through things, albeit in disjunctive and uneven ways. In contemporary contexts of intensified mobility, policy is often made possible by establishing new conduits and topological relations through which power can flow and realize its potential, both within and across political territories. A core implication of this is that power must be understood not only as topologically rendered but also as de-centred and polycentric in nature. In other words, power is extended across space in new ways and also lacks a singular ruling centre or nervous system from which it might extend its forces. Power thus has multiple nodes and centres through which its forces coalesce, interconnect and are transformed and redirected. As Allen (2009) argues, when we think in terms of assemblages, networks and topological relations, we see power ‘is not so much exercised over space or transmitted across it’ but rather is ‘composed relationally through the interactions of the different actors involved’ (p. 207).

This means we need to think in distinct ways about where power comes from and how it is put to work to conduct the conduct of individuals (Foucault, 2007; Miller & Rose, 2008). For example, if power is an immanent force, flowing in disjunctive ways through polycentric networks, then it also needs to be seen as both plural and volatile: that is, not as one solid or stable thing, but as an always temporary and contingent arrangement of forces that can splinter off in different directions, have different impacts in different contexts, and can be directed towards particular ends, but can never be fully contained. Moreover, far from being unidirectional or smooth flowing, power is made possible through ongoing contestation and resistance. As Li (2005) argues, policy (and the power relations that undergird it) are ‘the outcome of agency and struggle rather than a master plan’, adding that policy is not something ‘emerging fully formed from a single source’, but instead is an ‘assemblage of objectives, knowledges, techniques, and practices of diverse provenance’ (p. 386). Moreover, policy is, ‘always subject to contestation and reformulation by a range of pressures and forces it cannot contain’ (p. 386). Again, this is because power is everywhere in an assemblage, which, if we follow Michel Foucault’s lead, means resistance is also always present. As Foucault (1980) argued, ‘Where there is power, there is resistance’ and, as a result, ‘resistance is never a position of exteriority in relation to power’ (p. 95). This does not mean all actors and organizations have equal capacities to exert agency or impact upon change – power is not, therefore, ‘distributed equally’ (Anderson & MacFarlane, 2011, p. 125) – but it does mean resistance potential is always present. Resistance, therefore, is not something speaking back to power from ‘the outside’ but is always embedded and woven into power relations. As Li (2005) suggests, there are no ‘pristine spaces outside power’ or ‘pure sites of resistance’ outside of power relations (p. 385). Again, this view of power reflects strong connections to complexity theory, where systems are portrayed as lacking a single ruling centre, defined by relations of exteriority, made possible by power relations flowing through networks in non-linear ways, and always subject to resistance and forms of re/decomposition (Urry, 2003).

Building on the discussion of macro- and micro-reductionism above, an assemblage view of power not only has wide-reaching implications for how we understand and research policy development, enactment and impact, but also has major implications for how we understand established concepts such as ‘the state’, ‘the nation’ and, indeed, the hyphen in between the nation-state. Again, this is because a commitment to understanding assemblages as comprised of heterogeneous parts, brought together into relations of exteriority, means the various component parts that make a nation-state (e.g., organizations, political parties, actors, territories, etc.) must be understood as just part (but not all) of the policy assemblage process, and just part (but not all) of what creates the conditions of possibility for policy emergence in the first place. Indeed, as Tampio (2009) argues, assemblage thinking radically disrupts many dominant claims about power and the state, making the point that both Foucault and Deleuze reject the common notion that power is primarily ‘located in the machinery of the state’ (p. 390). No longer, therefore, might we assume that the state is the primary holder of power, from which force is extended in linear or top-down ways, or even that power might be extended smoothly across spaces of governance. Instead, the state is akin to one player in a dynamic game of power, with power flowing in and out of the state, and across political territories, in complex and non-linear ways. As Li (2005) argues, there exists no ‘up there’ or ‘all-seeing state operating as a preformed repository of power spread progressively and unproblematically across national terrain’ (p. 384). This is not to suggest the state lacks capacity to wield power (as such a claim would clearly be absurd), but instead that the state does not ‘own’ or ‘hold onto’ power in an absolute fashion, nor does it have absolute power to extend its reach across space. Rather than being an exclusive proprietor of power, the state can thus be seen as a primary force in directing power: harnessing, channelling and experimenting with it in the hope of steering the conduct of individuals both within and sometimes beyond the territories over which it claims control.

This clearly has implications for how forms of resistance are understood in relation to the state. Rather than seeing resistance as something that primarily acts on or against the state, with the state seen as the possessor of powers that might somehow be won by or transferred to successful resistive agents (Tampio, 2009), an assemblage approach invites one to see power and resistance as not only intimately connected but as existing simultaneously within and outside of the assemblage of components that act in the name of government. As Li (2005) argues, powerful forms of resistance often lie within political assemblages such as bureaucracies: that is, among actors and individuals engaged with the central work of the state (p. 385). The same goes for policy production, as while the machinery of the state might be uniquely placed to assemble policy, acting authoritatively when doing so, policies are formed through interactions with components located both within and beyond the state, and potential for both forging and resisting certain policy designs is distributed throughout all components implicated in the policy process. Put differently, the state does not simply make policy which might then be resisted by those outside the state, but instead power/resistance is embedded into the entire policy production and enactment process. Again, therefore, we see that policies not only lack an essence or singular guiding rationale, but the same holds for the power and political relations that determine the nature of their production.

This line of assemblage thinking does not sit well with the use of reified categories, not only concerning the nation-state but also other terms such as ‘the market’, ‘class’ and more. Indeed, it is difficult for theorists to think with assemblage while at the same time assuming the existence of such reified forms. While such categories might serve useful heuristic purposes, they ultimately fail to capture the complexity of the very relations they seek to represent. Sassen (2014), for example, sees the challenge that assemblage thinking poses to reification as one of its most potent weapons. In relation to the nation-state, for example, she argues that assemblage helps, ‘make visible how territory cannot be reduced to either national territory or state territory’, and also allows one ‘to expand the category of ‘territory’ to a measure of conceptual autonomy from the nation-state’ (p. 22). As Sassen argues, this does not mean that we throw established categories ‘out the window’, but, instead, allows one to ‘actively destabilize them’ (p. 18). Again, this demonstrates strong linkages to Foucault’s work, who also argued for a methodological approach that did not focus on pre-given ‘universals’, but instead started from a position of examining how these categories are assembled through ‘concrete practices’ (Foucault, 2008, p. 3). Put differently, rather than assuming from the outset that nations or states (or even national policies) are pre-existing things to be studied, we should instead begin by questioning how it is that these categories have come to be assembled in different ways in different spaces and times, and with what effects. National policy spaces, therefore, should not be understood, ‘in terms of a priori coordinates – a fixed stage upon which events occur – but instead as something formed by relations between heterogeneous parts’ (Savage & Lewis, 2018, p. 137, emphasis in original), relations that typically cut across and go beyond national policy spaces in complex ways (Temenos & McCann, 2013). Any national policy, therefore, might be better understood as more of a claim than a fact, and cannot be understood outside of the components that actually constitute it; with such components always contingent and subject to change. As Anderson and MacFarlane (2011) argue:

Assemblages always ‘claim’ a territory as heterogeneous parts are gathered together and hold together. But this can only ever be a provisional process: relations may change, new elements may enter, alliances may be broken, new conjunctions may be fostered. Assemblages are constantly opening up to new lines of flight, new becomings. (p. 126)

With all this in mind, it makes sense that an assemblage approach invites researchers to pay strong attention to politics and the relative capacities of individuals and organizations to exercise agency in relation to both the creation of policy and its enactments. For example, the very notion of policy assemblage carries with it a focus on various forms of process, that is, the many acts of arrangement and gathering together that occur in order to create forms of coherency out of multiplicity with the hope of steering individuals and groups towards particular ends. To understand processes of policy assemblage, therefore, it is difficult not to focus on the role of various actors and agents in creating the conditions of possibility for certain policies to emerge, while at the same time obscuring possibilities for doing policy otherwise. Li (2007), for example, places a strong focus on agency and the role policy actors and organizations play in highly contested processes of putting together and enacting policies. Li suggests assemblage implicitly ‘flags agency’ (p. 264) and the complex work involved in forging and maintaining connections between components in the hope of maintaining the coherency required to govern. With regards to processes of policy development, Li suggests a central practice that needs to be understood by researchers is that of problematization: that is, ‘how problems come to be defined as problems in relation to particular schemes of thought, diagnoses of deficiency and promises of improvement’ (p. 264). Crucial here is how ‘certain kinds of problems and solutions become thinkable whereas others are submerged’ (p. 386). In this way, an assemblage approach speaks productively to other work on problematization that draws upon Foucauldian perspectives (e.g., Bacchi, 2012; Webb, 2014). Savage and Lewis (2018) argue that problematization also requires close attention to how different political and policy contexts provide different conditions of possibility for policy to emerge in different ways, arguing: ‘certain policy contexts provide conditions amenable to certain policy ideas and practices, but not others, and the agency of policy actors and organisations, and the fidelity of policy proposals, is thus always ‘situated’ and context dependent' (p. 125). Different spaces and times, therefore, create different potentialities (McFarlane, 2011).

At the same time, attention is needed to the enactment phase of policy, especially to how actors interact with policy, sometimes enabling and sometimes subverting policy aims. As Baker and McGuirk (2017) argue, ‘assemblage methodologies’ are not only ‘committed to revealing the labours that produce and maintain assemblages: the labours of assembling’ (p. 431); but equally to how assemblages are then either held together or subject to forms of disruption and change. This is especially important in the field of policy mobilities research. As Temenos and McCann (2013) argue, policies do not just ‘move around in some abstract sense’, but instead ‘people move them around for particular purposes’ (p. 344; emphasis in the original); which means close attention is needed to the role of such actors in contouring the shape of policy movements and subsequent enactments. Ureta’s (2015) analysis of the development and enactment of the Transantiago public transport system in Santiago is illustrative of a commitment to tracing the role of variously placed humans – or ‘human devices’ (p. 4) – in the making and doing of policy. Ureta examines the development of the infrastructure reform from its initial proposal in 2000 to its launch in 2007, and its resulting (often negative) impacts up to 2009. In doing so, he pays specific attention to the roles of various actors, ranging from the graphic designers commissioned to promote the new policy and the academics and citizens harnessed to sell the reform’s merits, right through to the public servants tasked with managing and repairing failed policy enactments and the transport users who engaged with and often resisted the reform. Ureta positions humans as always imbued with agency, with ‘the capacity to alter the current state of affairs in one way or another’ (p. 7). Using insights from Foucault, Nikolas Rose and other governmentality theorists, Ureta frames humans as both as governed and self-governing subjects: that is, subjects of policy but also often active resistive agents. In doing so, he introduces the term ‘strange things’ (p. 10), which he adapts from the work of Marres (2005) to capture the often unexpected ‘overflowings’ (p. 8) (i.e., unexpected effects) that result from the tension between subjects being governed and individuals seeking to self-govern and exert agency. Ureta also draws centrally upon insights from the field of actor–network theory, which has been central to developing policy assemblage approaches, and which maintains a strong focus on human and non-human agents (see also Gorur, 2011; Koyama, 2015; Latour, 2005).

Ultimately, by dedicating attention to the complex channels of power, politics and agency that contour the possibilities for both policy creation and enactment, an assemblage approach signals rich opportunities for critical research driven by normative visions about how we might do policy better in the future. For example, by engaging in a forensic analysis of how different component parts make certain politics and policy possible, researchers are not only well placed to forge sophisticated forms of resistance, but also to imagine how we might do politics and policy otherwise: that is, to consider how new and different assemblages might be forged with potentially more positive impacts and possibilities (i.e., new forms of emergence). Building on Peck (2011), therefore, an assemblage approach not only helps one see how policy itself is ‘saturated by power relations’ (Peck, 2011, p. 791), but shows how any assemblage always carries within it the seeds of potential change (see also Tampio, 2009). As McFarlane (2011) argues, by ‘emphasising potential through its orientation to assembly, reassembly and constitution, assemblage focuses on the disjunctures between the actual and the possible’, thus positioning one to consider ‘how relations might be assembled otherwise’ (p. 210). Acuto and Curtis (2014) make a similar point in arguing that assemblage thinking brings with it a clear political orientation, suggesting ‘the commitment to critique found in assemblage work is itself a political orientation’ (p. 12), adding: ‘assemblage is charged with critical and political possibilities’ and raises ‘new questions about the nature of power’ (p. 13). Of course, this does not mean that we should always see assemblages as negative forces that must invariably be resisted. After all, as Legg (2011) argues, the forging of assemblages can very often be generative and positive, noting:

There is, of course, a need for ordering, security and stratifications, and these powerful processes need not be negative. A trade union movement, a family, a partnership, a migration or a waist-line all need some degree of control, but this can be productive not deductive. (p. 129, emphasis in original)

In this way, therefore, assemblages should not be seen as simply capturing or determining forces, but instead always as ripe spaces of potentiality and change, always opening new windows and lines of flight towards imagining and assembling something better in line with some normative preferred vision of the world.

TAKING POLICY ASSEMBLAGE FORWARD

Anderson and MacFarlane (2011) argue that ‘part of the reason assemblage is being increasingly used across a wide range of contexts is its very manipulability’, suggesting it is variously used as a descriptor, a concept and an ethos in ways that often differ in form and effect, which in turn, produces complex ‘differences and tensions’ (p. 126) in how the term is articulated and put to work. This manipulability of assemblage carries clear risks. For example, when a term is chameleon-like, disposed to meaning too many things, it can end up meaning everything and anything, and thus nothing at the same time. Just as concerning is Allen’s (2011) argument that assemblage research often fails to extend beyond ‘thin’ tracing and description of various component parts. If assemblage continues to serve in this way, as little more than a justification for an elaborate mapping of various bits and pieces of policy, then it will ultimately fail to generate explanatory power (Brenner et al., 2011) and will be incapable of addressing deeper questions about politics, power and agency which assemblage thinking offers such rich potential to explore; and which have the potential to lead to the formulation of visions towards a better future. It is important, therefore, to avoid assemblage slipping into a conceptual labyrinth.

This paper has sought to respond to such risks. By synthesizing core threads from existing literature, three theoretical and conceptual foundations central to a ‘policy assemblage’ approach were identified: (1) relations of exteriority and emergence; (2) heterogeneity, relationality and flux; and (3) attention to power, politics and agency. Together, these foundations are seen here as signalling a coherency to assemblage thinking, which suggests an assemblage approach has powerful potential, allowing researchers to see and explain things in ways that many established traditions in policy research do not. By identifying foundations and offering examples of how each might be mobilized, the paper has sought to provide the beginnings of a framework for policy assemblage research not previously articulated in a systematic form, thus inviting further discussion about what it means to do policy assemblage research. In doing so, it has not been the intention to suggest assemblage should have one fixed meaning or be anchored exclusively in the work of particular scholars (Allen, 2011; Anderson & MacFarlane, 2011). Nor is it the intention to suggest that the three foundations articulated above should constitute any kind of final word on the core features of an assemblage approach. Instead, the paper serves as an invitation to engage in further discussion about an approach to researching policy that can be seen as being rich with possibilities.

Of course, while this paper has engaged in a detailed unpacking of the theoretical and analytical foundations of an assemblage approach to policy analysis, significant further work is needed to articulate what the methodological foundations of such an approach might be. As Baker and McGuirk (2017) argue, while there are ‘many accounts using assemblage-inflected methodologies of various sorts as analytical tools for revealing, interpreting, and representing the worlds of policy-making’, it is also the case that ‘few are explicit about their methodological practice’ (p. 425). The path forward, therefore, not only lies in forging further clarity over the kinds of methodological practices we associate with assemblage, but in bringing these together in a coherent and convincing way with the kinds of theoretical and conceptual foundations the paper has sought to articulate above. Only through these synthesizing practices will assemblage thinking be able to ensure the level of ‘thoughtful conceptualisation’ (Allen, 2011, p. 156) it requires if scholars are to productively defend and extend its use in policy research.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author.

Additional information

Funding

This work was supported by the Australian Research Council [grant number DE160100197].