Guest editorial: Integrated policy making as a problem of collaboration

A wicked problem is wicked because different actors see different problems and in turn propose different policy solutions. The research problem we aim to address in this special issue is how shared perspectives emerge, or do not emerge, in a diverse actor constellation. Policy integration is ill-understood (cf. Sahamies and Anttiroiko, 2024). Collaborative governance and policy integration tend to operate as “magic concepts” (Pollitt and Hupe, 2011). We cannot assume that, but instead need to explain how, “different but needed actors” in some way make differences disappear and “gain agreement on what public value is to be co-produced” (Bryson et al., 2017, p. 649).The research question for this special issue is: how, in a world of separate policy formation and implementation, might collaborative governance support integrated policy making?We asked contributors to follow a “decentred” (Bevir and Rhodes, 2016) methodology, directed to the “micro-politics” (Waring et al., 2023) of collaborative governance at the levels of policy definition and implementation, starting from the horizontally and vertically separated perspectives of involved actors, not from the assumed sharedness of the problem approach.In this introductory article we will first discuss our understanding of wicked problems, then discuss characteristics of common approaches to collaborative governance and propose an approach to the governance of wicked problems from a decentred perspective. Such an approach provides new ways forward in theory building about collaborative governance and implementation.In his book The Great Transformation (1947/2001), Polanyi constructed as a central societal problem, the problem of “freedom in a complex society.” Freedom needs to be understood as an individual's capacity to be responsible for the results of the choices he or she makes. The object of one's responsibility is, Polanyi argues, that others affected by your choices are also free. However, when these others are also free, one cannot simply predict what the results of one's choices might be. This, after all, depends on the action of those others: society is “complex” and individuals lack “overview.” Individuals cannot take full responsibility for the results of their actions, because in a “complex society” they cannot know what the results of their actions might be. Therefore, they cannot be free.This conceptualisation strongly resembles concepts of late modernity by authors like Giddens (1990), Beck (1992) and Lash (2003). Individuals develop their own understanding of what their world looks like and become “clever people.” But others become clever people too, developing an alternate understanding of the world around them. We become actors among other actors, pushing forward specific conceptualisations of the world, “reflexively” responding to the conceptualisations of others that we cannot fully grasp or evaluate. “Wicked problems” are related to this complexity.Wicked problems are also related to conflict. Rittel and Webber argued that problem definitions and definitions of the public good are disputed and that, therefore, “it makes no sense to talk about ‘optimal solutions’” (Rittel and Webber, 1973, p. 155). Hoppe (2010, p. 73) builds upon Rittel and Webber's initial conceptualisation and argues that wicked problems are characterised by a lack of certainty on required and available knowledge and a lack of agreement on norms and values at stake. Bannink et al. (2024, p. 303) summarise that “a wicked problem is wicked because actors do not agree on problem definitions and policy preferences.” Wicked problems bring to the table conflicts that cannot be assumed away.COVID-19, the Ukraine war, economic downturns, energy shortages, climate crisis are all exemplary social conditions that strongly suggest that individuals lack the “overview” to determine the results of their actions and conditions that show strong disagreement on the values that are at stake.In this section, we discuss wicked problems first in relation to collaborative governance and then we also include policy implementation.The wide variety of literature on collaborative governance reflects the following core idea: when society is complex, collaborative governance is required and what it does is improve the output of governance processes (supporting collaborative advantages, public value) by improving the input of these same processes (reinvigorated democratic ethos, boundary spanning). In Polanyi's terms: collaborative governance aims at overcoming the “overview problem” in “complex society” by including a broad variety of actors in both policy formation and policy implementation.This core assumption, however, is flawed: in the case of wicked problems, the varying perspectives of a variety of actors are required, but this same variety implies different points of view regarding problem definitions, responses, and success measures (Bannink et al., 2024; Van Duijn, 2022; Waardenburg et al., 2020). The actors needed to tackle the overview problem, lack overview.Bartels and Turnbull made a classification of relational public administration-literature. With respect to collaborative governance, they argued (2020: p. 1335):Bartels and Turnbull mention “boundary spanning,” the creation of “collaborative advantage” and they cite Mandell and Keast (2007, p. 593) who argue that in collaborative governance-networks “the traditional commitment to producing goods and services must give way to a commitment to improved relationships and forming a new whole.” Bartels and Turnbull's analysis enlightens two important characteristics of the collaborative governance-literature:We see these two characteristics indeed in several core publications, including Ansell and Gash (2008), Emerson et al. (2012) or Bryson et al. (2017). Bannink et al. (2024: p. 308) observe that these publications discuss to what extent collaborating actors address “public purposes that could not otherwise be accomplished.” Collaboration operates as a “magic concept” (Pollitt and Hupe, 2011). Collaborative governance is the governance mode required to solve problems for which we lack overview and that we disagree about, in other words: the response to the problem condition (collaboration) is the negation of what is missing in the problem condition (collaboration).Bannink et al. (2024: p. 308) argue that in these approaches “problem” is reified “from collaboration”: the collaboration defines the problem. Extant networks of collaborating actors defined a “problem” that allowed their collaboration: “problem” is the issue a network of actors is oriented towards. Numerous alternative problem definitions did not facilitate collaboration. Ansell et al. (2020: p. 571) take an external approach to problem definition: “collaborative processes may be more successful if inclusion is more strategic and selective.” Using this point of view, the authors apparently treat the problem as a given and discuss which level of inclusion might lead to the most successful response approach. However, when seeing “problem” as a reification of the network, one would argue that different patterns of in- and exclusion lead to different problems being addressed. Furthermore, the problem might not represent public value: an actor might share resources to attain an individual goal that requires co-ordination (Vangen and Huxham, 2003). Actors selectively interact, pursuing different, incongruent perspectives (Klijn and Koppenjan, 2016; Lewis, 2011).We assume different actors who have gained differential access to the definition process push forward a specific set of worldviews, problem definitions and response directions, which differentially represent those included on the basis of the influence they have gained in the process, while differentially disregarding or taking into account the worldviews of other actors.Bartels and Turnbull (2020) classify the policy implementation literature as more strongly individualist than the collaborative governance-literature and equal to the collaborative governance-literature as primarily instrumentalist-strategic.This classification implies: the own perspective of actors in implementation is acknowledged in the implementation literature (individualist), but this perspective is (positively or negatively) related to pre-existing ideas of policy objectives (instrumentalist): the action of an actor in policy implementation can be seen as the more or less complete implementation of a policy.Lipsky (1980) is one of the authors who grounded the policy implementation approach, in his case with an emphasis on the street-level bureaucrat. Rowe (2025) re-analyses the work of Lipsky and applies an ethnographic methodology to do so. Rowe argues extensively throughout his analysis that street-level bureaucrats' work cannot only be seen as the more or less extensive use of discretionary space and the more or less complete implementation of policies but instead is also to be understood on its own terms. Street-level bureaucrats engage in their own processes of meaning-making.Street-level bureaucrats' meaning-making is important, because the street-level bureaucrat makes an autonomous contribution to policy implementation that cannot be fully structured through the regulation of his work, as Bannink et al. (2015) argue. Policy formation is translated into policy implementation when concrete cases are selected upon which abstract policy goals are applied. The definition of problems is not only produced in the network of actors involved in policy formation but is produced again in policy implementation. Our understanding of problems as concepts reified from the action of actors requires that policy implementation is also considered.To conclude, we argue that the approaches in the literature on collaborative governance skip the first analytical step: the shared understanding of the wicked problem is already there; indeed, conflict is acknowledged, but a network of actors that may potentially engage in collaborative action is already assumed. Contrary to the collaborative governance-literature, we take a critical-reflexive approach, starting from the own perspectives of actors that form the basis of their definition of problems. We do not assume wicked problems and actors' differentiated orientation towards them. Instead, we assume wicked actor constellations that might, but also might not define a shared problem, might or might not engage in shared problem solving, and if they do might exclude the orientations of others.We argue that “problem” is a reified concept in a similar way Bevir and Rhodes (2016) argued that “structure” is a reified concept. The delineation of the set of actors, delineates the definition of the problem. Problems are not naturally coloured but are constructed and defined by the social actors around them. A decentred approach zooms in on the different perspectives of actors.Decentred theory begins with the idea that actions should be explained with reference to the reasons actors have for those actions, explaining reasons by locating them in the agents' webs of belief, including the ways they understand their position, interests and the norms affecting them. Actions form a way of understanding how patterns of public governance take shape.Tradition, dilemma and narrative account for the situated actions of policy actors. Tradition is understood as the acquired meanings, customs and interests that guide social action in a non-deterministic way. Actors constantly reproduce and adapt their traditions when they encounter novel situations. A dilemma arises for an individual or institution when a new idea stands in opposition to existing beliefs and forces a reconsideration of beliefs and associated tradition. Dilemmas do not have given or correct solutions. When people adopt new ideas they change traditions creatively, because no set of beliefs can fix its own criteria of application. Narratives, in this sense, convey the complex set of assumptions rooted in a tradition that provides a guide for situated action. Diverse and competing narratives that emerge around a given policy issue and policy process are illustrative of how different groups think the world should be. Governance, not necessarily collaborative, takes place across a “complex ecology of self-organizing but interdependent social subsystems” (Jessop, 2011, p. 5).We draw on Scott's (2013) distinction of three pillars of institutions: a cognitive, a normative and a regulative pillar. We argue that the ways in which actors interact and produce definitions of problems are based on their individually constructed, but partially overlapping:Actors' individual constructions might overlap to a greater or lesser extent. The challenge in collaborative governance, and what we seek to explore in this special issue, is to what extent and how actors make sense of, and reconcile, the differences between them.The articles in this special issue reveal how the meanings, motives, and materials of actors involved in collaborative governance affect its construction.Van der Woerd and Van der Scheer (2026) outline how managers engaged in a transformation towards networked child welfare policy implementation try to work towards “shared leadership.” The article emphasises the “processual, provisional, and political nature” of navigating through emerging tensions. This provisionality is characteristic of “working on a plot in the making.” It is the actors, not the network, who work on the plot in the making. Their sense-making is important to understand how – if at all – shared leadership or collaborative governance unfold.Siemers' (2026) point of departure is more prescriptive. Siemers' (2026) emphasises the role of policy entrepreneurs in complex adaptive systems (CAS). Policy entrepreneurs might be able to induce change in a CAS, when they display “social acuity,” which means “to understand the ideas, motives, and concerns of others in their local policy context” and are able to define problems and build teams based on that acuity.This requires “institutional work,” Van der Heijden et al. (2026) argue. The authors studied many cases and showed that “outcomes in integrated service delivery did not happen by itself, but required active work by local officials,” while at other times policy goals met “ongoing resistance from service providers.” Furthermore, they show how conceptual work was needed to inform relational activities.Van Duijn et al. (2026) discuss the implementation of a “virtual ward,” a digital communication platform aimed at improving communication between care professionals. The authors show how the virtual ward operated as a “boundary object,” a system that is “intended to facilitate interaction between different organizational worlds but is not necessarily used as such by the collaborating actors.” The various actors approached the material object, the virtual ward system, starting from their own meaning and motives.Crompton and Waring (2026) show that the disjunction between collective and actor-level perspectives causes actors to employ “uncollaborative” strategies. Crompton and Waring (2026) described three modes of “un-collaborative governance,” underscoring that also when a collaborative approach is widely supported by involved actors, these actors build upon their own capacities, e.g. by limiting access to shared decision-making by others, excluding dissenting views or by controlling the flow of information, setting the agenda around the own understanding of the policy problem.In this section, we translate the analysis of “integrated policy making as a problem of collaboration” into implications for the understanding of collaborative policy-formation and implementation.When an analysis of collaborative mechanisms shifts from network to actor, it becomes an analysis of mechanisms of inter-actor interaction about problem definitions and the positioning of actors vis-à-vis these definitions. Research does not start from a problem or a given network, but from actors that might potentially (but are not necessarily actually) involved in debating a mode of action. Different actors with their own meanings, motives and materials have different problems in mind. Not the “problem” is wicked, but instead the “actor constellation.” Dependent on the action actors take, an actor constellation might produce various, different problem definitions. The concept of “problem” is a reified concept.Mouffe's (2005) emphasis on “contestation” helps us understand why an analysis of governance needs to take the differences between actors as a starting point. Contestation clarifies what these actors consider important. The connections between actors involved in policymaking yield tensions that these actors need to “tinker” with (van Duijn, 2022). Crompton and Waring (2026) argue that “collaborative governance is commonly presented as a progressive and democratic model.” We do not fully agree and would like to use the final part of this guest editorial to explain why. Conflict resolution is agentical in nature. It starts from the position of the actor in conflict. Analysing the “micro-politics” of conflict allows us to recognise structural or ideological tensions.It is important to define what collaboration means (Bannink et al., 2024). It refers to a shared orientation to an at least partially shared problem definition, the intention to share resources in order to address this problem (policy formation) and possibly even the realisation of this intention (policy implementation); or, earlier in the process, the shared intention of actors to produce these (cf. Emerson et al., 2012). Instead, the concept of “empathic positioning,” as opposed to “sympathetic positioning” (La Grouw, 2025; Bannink et al., 2024) better describes what actors do: they reflect upon their own meanings, motives, materials and try to factually understand (not necessarily positively value, which is “sympathetic positioning”) those of others. Wicked actor constellations might produce shared problem definitions without resource sharing, resource sharing without shared problem definitions, shared policy intentions, without a shared problem definition and/or a realised sharing of resource, etc. In other words, collaborative governance is a very specific type of governance. Interactions between actors to address a variety of defined problems occur in a much broader set of governance modes.This leads us to make two statements:We argue that our analysis is more “democratic” than a collaborative governance-approach, because it takes seriously the positions of actors in a constellation. It analyses these actors from their own perspective, instead of analysing their position vis-à-vis a problem that is defined elsewhere, either by the researcher studying wicked problems or by the sub-set of the wicked actor constellation successfully able to define it.