Review of research into urban experimentation in the fields of sustainability transitions and environmental governance

ABSTRACT In recent years, scholars from diverse research communities have become greatly interested in the experimental turn within the social sciences and economics. A rapidly emerging field within the literature on experimentation is investigating urban experimentation to promote sustainability transitions. Emphasising the situatedness of experiments, such publications are shedding light on the particularities of places and spaces. Experimentation conducted in real-world settings seeks to combine purposeful intervention with observation and reflection. In the study of urban experimentation, various disciplinary lenses and conceptual frameworks have emerged, especially in the two literatures on sustainability transitions and environmental governance. This has fostered a more complex, differentiated understanding of experimentation. The literature review presented here seeks to elaborate current lines of inquiry on urban experimentation in order to gain insight into how experimentation shapes urban sustainability transitions. By outlining the complementary and contrasting nature of different academic perspectives, it is possible to identify avenues of future research and encourage dialogue between the different research communities.


Introduction
As humankind enters the so-called 'urban age' (United Nations 2019), cities are attracting ever more attention as the sites of societal and environmental challenges, on the one hand, and creative and innovative solutions, on the other (Bulkeley and Castán Broto 2013;Nevens et al. 2013). This is reflected in the growing literature on urban experimentation, which considers the particularities of places and spaces with the aim of fostering sustainability transitions. The experimental turn in the social sciences and economics has drawn the attention of scholars from different research communities, thereby expanding the literature on experimentation to transform society (Kivimaa et al. 2017;Sengers, Wieczorek, and Raven 2019;Weiland et al. 2017). One rapidly emerging field of study within this literature focuses on the urban environment Kivimaa et al. 2017).
Due to the considerable ambiguity in the understanding of experimentation, Karvonen and van Heur seek to provide greater conceptual clarity by emphasising the fact that laboratories and experiments in real-world settings aim to combine purposeful intervention with observation and reflection (2014). Urban experimentation can be defined by three key features: situatedness, change-orientation and contingency (Karvonen 2018;Karvonen and van Heur 2014). More specifically, experiments can be described as situated in that they seek to mediate between the contingency of particular places and the universality of general laws of causality. The knowledge generated by such experiments in real-world laboratories is ideographic local knowledge, offering context-specific lessons and informing middle-range theories, in contrast to nomothetic knowledge, which comes in the form of generic lessons gained in the laboratories of the natural and engineering sciences (Coenen, Raven, and Verbong 2010;Evans 2011;Schäpke et al. 2017;Schneidewind 2014; van den Heiligenberg et al. 2017;von Wirth et al. 2019). The learning in real-world experiments is derived from real-world interventions (Evans 2016). Such forms of experimentation seek to enable collective search and exploration processes that generate actionable knowledge on societal challenges (Ansell and Bartenberger 2016). It is the aim of urban experimenters to foster greater sustainability, including possible disruptive or radical change in incremental steps (Bulkeley et al. 2019). They embrace uncertainty and contingency, acknowledging that cities are continuously 'in the making' (Karvonen 2018;Karvonen, Evans, and van Heur 2013). They intend to redefine failure and learning-by-failing as something productive and constructive Karvonen, Evans, and van Heur 2013). By embracing uncertainty in this way, urban experimentation forms an antidote to modern ideals of rational and comprehensive planning (Karvonen, Evans, and van Heur 2013). The experiments conducted in real-world laboratories have been defined as 'an inclusive, practice-based and challenge-led initiative designed to promote system innovation through social learning under conditions of uncertainty and ambiguity.' (Sengers et al. 2016, 21).
In the study of this real-world experimentation in urban contexts, different disciplinary lenses and conceptual frameworks have emerged, especially in the two research communities of sustainability transitions and environmental governance (Bulkeley 2021;Turnheim, Kivimaa, and Berkhout 2018b;Sengers, Turnheim, and Berkhout 2021). Due to its interdisciplinary nature, there exists a complex and diverse understanding of experimentation (Kivimaa et al. 2017;Sharp and Raven 2021). Previous reviews of experimentation to promote societal transitions have adopted a different scope and focus to that presented here. While these reviews have analysed the study of experimentation in sustainability transitions research and experiments in climate governance (Kivimaa et al. 2017;Sengers, Wieczorek, and Raven 2019;Weiland et al. 2017), they have not directly addressed the meaning of urban settings for experimentation. In contrast to other governance levels, urban settings are characterised by proximity. In their study of favourable environments for urban experimentationa contribution to the debate on the geography of sustainability transitions - Torrens et al. (2019) did, however, focus on the question of place and space.
From an analytical perspective, our understanding of the various concepts of urban experimentation remains under-explored. There is also limited understanding of how urban contexts influence the design, implementation and impacts of real-world experiments. From a prescriptive perspective, the question arises: How can experiments that serve governance innovation actually contribute to sustainability transitions and which pitfalls may emerge? The first objective of the current review is to elaborate the lines of thought and theoretical underpinnings of urban experimentation. In particular, different academic perspectives are compared and contrasted in order to reveal the 'implicit assumptions within experimentation for sustainability' (Weiland et al. 2017, 1). After outlining the conceptual development, the second objective is to identify research gaps for future research and to encourage dialogue between the different research communities. The research questions to be answered can be formulated as follows: . What are the academic perspectives and lines of inquiry on urban experimentation to foster sustainability transitions? Which implicit assumptions underlie the definition and study of urban experimentation? . In which ways do the various academic perspectives and lines of inquiry contrast?
How could they complement each other?
The following section outlines the methodology of the literature review. Section 3 introduces the literature on sustainability transitions and the literature on environmental governance, elaborating how these academic perspectives frame urban experiments. In Section 4 I discuss how these academic perspectives compare with and complement each other. The paper concludes with some critical reflections on the development of the study of urban experimentation.

Methodology
To examine the emerging field of research, I conducted a systematic literature review (Petticrew and Roberts 2006), thereby identifying and synthesising the findings according to a transparent, rigorous procedure (Victor 2008). Systematic literature reviews are generally conducted 'to explore and contrast understandings of particular concepts, to enable theoretical refinement and identify areas for further development' (Torrens et al. 2019, 5). Such reviews align with the constructivist perspective on scientific knowledge and the interpretive epistemology prevalent in sustainability transitions research by applying more interactive procedures than the staged protocols prevalent in the social sciences (Sengers, Wieczorek, and Raven 2019;Torrens et al. 2019).
The timeframe of the review is from the 1990s, when this field of study first appeared, up to February 2020. The scope and focus is the literature on experimentation in urban contexts to foster sustainability transitions. The normative objective of urban experimentation is a systemic change from unsustainable to sustainable systems (Loorbach, Frantzeskaki, and Avelino 2017). Accordingly, the criteria for the inclusion or exclusion of publications in the review were defined as the study of urban experimentation with the aim of inducing such systemic change in society. Publications investigating experimentation beyond urban contexts (e.g. rural or national contexts) or with some other objectives than sustainability transitions were excluded (see Appendix A.1).
In analysing the publications, I followed a qualitative, exploratory approach, developing the coding scheme in an abductive manner by combining a deductive and an inductive approach. This analysis entailed an iterative process, i.e. going back and forth between the preliminary concepts generated by the literature review and the further study of this literature and the refinement of the concepts. The academic perspectives on urban experimentation were identified through this process.
Regarding the methodological limitations, the review covers only English-and German-language publications, thereby somewhat limiting the geographical scope. Moreover, any form of content analysis encounters the problem of reliability due to the subjective judgements of human readers (Belur et al. 2018;Mikhaylov, Laver, and Benoit 2012): clearly, two readers may attach different meanings to the same words. While there are virtually no studies comparing multiple coding with single coding to establish the level of reliability of the latter, single coding is still a common research practice due to constraints in funding and staffing (Campbell et al. 2013).

Academic perspectives and lines of inquiry on urban experimentation
Two prominent academic perspectives on urban experimentation have emerged over the past years: the literature on sustainability transitions (transition studies) and the literature on environmental governance (governance and policy studies) (Bulkeley 2021;Sengers, Turnheim, and Berkhout 2021;Turnheim, Kivimaa, and Berkhout 2018a). In the following, I explore how urban experimentation is defined and studied by these literatures. Table 1 3.1. The literature on socio-technical innovations and sustainability transitions Urban experimentation has been explored from the perspective of socio-technical innovations and societal transitions. This literature has its roots in innovation studies, science and technology studies (STS), structuration and evolutionary theory. Research into sustainability transitions considers fundamental changes in provisioning systems such as energy, mobility or agriculture (Loorbach, Frantzeskaki, and Avelino 2017;Markard, Raven, and Truffer 2012). In particular, science and technology studies interrogate the relationship between technological change and social change, viewing technologies as embedded in social settings rather than separate from these. Consequently, transitions in technology and society are assumed to co-evolve. Scholars argue that changes in technology are intertwined with changes in institutions, belief systems, values and lifestyles (Brown and Vergragt 2008). This has given rise to the notions of 'socio-technical systems' and 'socio-technical experiments' (Sengers, Wieczorek, and Raven 2019). Within the literature on sustainability transitions, experiments are framed as 'spatially and temporally circumscribed initiatives that promote new social and technical innovations' (Sengers, Turnheim, and Berkhout 2021, 2). The aim is to replicate and scale them up for a wider impact. Given the origin of transition studies in innovation studies, experimentation is viewed by scholars from the perspective of innovation and the creation of new markets (Bulkeley 2021).
These socio-technical experiments are associated with the concept of niches as introduced by the multi-level perspective (MLP), which depicts change in socio-technical systems as emerging from the re-alignment of the different levels of niches, regimes and landscapes (Geels 2005(Geels , 2002. Experiments are thus framed as 'niche experiments' (Kemp, Schot, and Hoogma 1998). Building on the MLP, the approach of strategic niche management conceives of niches as protective spaces to be developed through the dynamics of shielding, nurturing and empowering (Geels and Schot 2007;Hoogma et al. 2002;Kemp, Schot, and Hoogma 1998;Smith and Raven 2012). Shielding is the protection of the niche by warding off external pressure from the economic and social environment; nurturing refers to processes that support innovation; and empowering means the strengthening of niches vis-à-vis regimes. Geels and Deuten describe the knowledge flows between local and global niches (2006). These can be defined as follows: local niches provide contextualised knowledge and locally applicable lessons, while global niches provide context-independent structures, shared rules as well as decontextualised concepts. Knowledge gained within local niches needs to be abstracted and aggregated to be transferred to global niches. Research on niche experiments has included emerging energy technologies in different local contexts, in which the variation in local energy innovations is contextualised locally by comparing rural and urban settings Schreuer, Ornetzeder, and Rohracher 2010). The socio-technical nature of experiments has been explored in a number of studies (Brown and Vergragt 2008;den Hartog et al. 2018;Peng, Wei, and Bai 2019). In particular, Peng et al. examined the emergence of a new socio-technical configuration in the energy sector in Shanghai, illustrating the realignment of actors, resources and institutions in the process of introducing Energy Performance Contracting (EPC) and Energy Service Companies (ESCOs) (2019). These were realigned by the redefinition of role understandings, a shift to multiple public and private funding sources, the transformation of the planned economy into a market economy, and reforms towards a more service-oriented form of government.
While socio-technical experiments explore the development of technologies, scholars have introduced the notion of social niches or grassroots experiments to develop an alternative view on innovation. These highlight change through experimentation by civil society from below (Seyfang and Haxeltine 2012;Seyfang and Smith 2007;Smith et al. 2016). Grassroots experiments are evidence of the democratisation of innovation through the engagement of civil actors (Pesch, Spekkink, and Quist 2019). In particular, the Transition Towns movement defines local communities as primary agents of change (Feola and Nunes 2014;Seyfang and Haxeltine 2012;Smith 2011). Such communities challenge the accepted belief in the power of ecological modernisation and technological progress to transform society, instead promoting the development pathways of degrowth and self-sufficiency (Feola 2020). They warn that technological progress may merely reinforce rather than transform growth-oriented practices of production and consumption, arguing that a shift in values from efficiency to sufficiency is essential for societal change. The study of grassroots experiments builds on the literature of social activism and social movement theory.
In a comparative study of community energy initiatives as a form of grassroots innovation, Smith et al. investigated the relationship between grassroots experiments and policymaking (2016). In particular, they applied three analytical frames to explore the question of how these grassroots experiments shape policy: strategic niche management (SNM) (aimed at developing solutions for sustainability), niche policy advocacy (to make sustainability solutions matter), and critical niches (to provoke debate on sustainability). Their findings show how grassroots experiments can redirect policy discourses by developing alternative solutions for energy provision and by providing proof-of-concept through demonstration (ibid. 426). Beyond the techno-centric view of SNM, the perspectives of policy advocacy and critical niches highlight how community energy initiatives can reframe 'ways of thinking about and acting upon energy questions' (ibid. 429).
Transition studies conceive experiments as spatially and temporally circumscribed interventions. Clearly, these have to be scaled-up from niches (as protective spaces) to the regime and other contexts in order to generate fundamental change. This development and diffusion of experiments is captured by the transition mechanisms of deepening, broadening and scaling up (Bai, Roberts, and Chen 2010;Bos and Brown 2012;Dijk, de Kraker, and Hommels 2018;Farrelly and Brown 2011; van den Bosch and Rotmans 2008; van den Bosch-Ohlenschlager 2010;Williams 2016), as well as embedding, translating and scaling . Embedding describes the integration of experiments in local structures and communities (Turnheim, Kivimaa, and Berkhout 2018b). Translating is the horizontal diffusion of experiments when they are replicated elsewhere; this might be a spatial, organisational/institutional or sectoral context, transferred across sectors such as education or energy. Scaling describes the internal development and growth of an experiment. Scholars in transition studies aim to explore the potential of urban experiments to transform the incumbent regime. They identify the dilemma of localism, namely the potential disparity between context-specific knowledge generated in local settings and generic lessons that are transferable across time and space (Bai, Roberts, and Chen 2010;von Wirth et al. 2019;Williams 2016). Such local lessons may in fact be incompatible with other local contexts.
In order to scale and embed these local lessons, intermediates are required to act as mediators, facilitators and translators (Bos and Brown 2012;Matschoss and Heiskanen 2017). This implies that the role of agency in urban experimentation is studied primarily through the concept of intermediary actors. The roles and functions of these intermediaries will depend on the particular perspective. Under the multi-level perspective, they bridge niches and regimes. Under a governance perspective, they act as intermediaries between multiple governance levels (e.g. local, national or international). Combining both perspectives, Matschoss and Heiskanen (2017) summarise the functions of intermediaries as follows: . To aggregate lessons by integrating local and non-local knowledge as well as expert and local knowledge; . To disrupt existing practices by introducing new practices; . To scale local innovations for impact beyond the experimental context.
Experimentation is a core feature of transition management (TM), viewed as a prescriptive approach to reflexive governance. Combining a descriptive analytical perspective to understand transition dynamics with a prescriptive perspective to inform governance and promote societal change, TM builds on the intersection between literatures on complex systems and evolutionary theory, governance and transdisciplinary research (Loorbach 2010;Rotmans, Kemp, and van Asselt 2001). It follows an iterative cycle of system analysis, problem structuring and envisioning (strategic), backcasting to determine major pathways and agenda setting (tactical), experimenting (operational) and, finally, monitoring and evaluating (reflexive) (Frantzeskaki and Kabisch 2016;Nevens et al. 2013;Nevens and Roorda 2014). In particular, frontrunners are defined as critical actors who explore alternative pathways, conduct experiments and draw lessons. Within the approach of TM, so-called Urban Transition Labs are institutional sites designed to facilitate processes of social learning (Nevens et al. 2013;Nevens and Roorda 2014). They combine long-term visioning with on-site experimentation through transition experiments.
This reflexive mode of governance is exemplified by a study of changes to urban water planning in the Cooks River Catchment in Sydney, Australia, which emerged as an ongoing transition experiment (Bos and Brown 2012). The development followed three phases of deepening, broadening and scaling up ( van den Bosch and Rotmans 2008; van den Bosch-Ohlenschlager 2010). More precisely, the deepening phase was characterised by lessons learnt about alternative planning approaches towards water management within the local context. Frontrunners proposed a vision and an agenda for novel forms of urban water management. During the broadening phase, these planning approaches were refined and replicated within the river catchment area through multiple collaborations. Again, frontrunners helped build new partnerships and replicate the lessons learnt. In the final phase of scaling up, frontrunners promoted an agenda for institutional change to embed the governance innovations into the established arrangements of urban water management.
Transition management builds on the diverse literature of transition studies, governance studies and transdisciplinary research. In a similar vein, scholars seek to bridge transition studies and governance studies by integrating neo-institutional theory. They frame sustainability transitions as processes of institutional change, whereby established sociotechnical systems are de-institutionalised and new ones institutionalised Geels 2004). These authors illuminate the co-constitutive relationship between experiments and institutions in which institutional settings shape experimentation and experimentation redefines institutions Raven et al. 2019).
The question of places and spaces is particularly significant in the study of experimentation in urban contexts. Not only are actors embedded in social contexts and relationsactors and contexts are assumed to co-constitute one another (Giddens 1984). Informed by the spatial turn in sustainability transitions research, scholars focus on the geography and place-specificity of contexts (Carvalho, Mingardo, and van Haaren 2012;Coenen, Raven, and Verbong 2010;Dignum et al. 2020;Håkansson 2019;Hodson, Geels, and McMeekin 2017;Späth and Rohracher 2012;Torrens et al. 2019Torrens et al. , 2018van den Heiligenberg et al. 2018van den Heiligenberg et al. , 2017van Steenbergen and Frantzeskaki 2018;von Wirth and Levin-Keitel 2020). They strive to promote dialogue between the literature on innovation studies and socio-technical transitions (primarily SNM), and the literature on economic geography and the geography of innovation (especially regional innovation systems and clusters). Within the spatial sciences, two contrasting conceptions of space can be identified: (a) the absolute conception of space, which defines physical-material space as a neutral container and where space and social processes are separate; and (b) the relational conception of space, which defines space in relation to society, and where space is viewed as constituted by society and social processes (von Wirth and Levin-Keitel 2020). Scholars of the geography of transitions have shown how local environments and experiments co-evolve. In particular, while experiments evolve and align with local contexts in a pathdependent manner, they also move beyond and reshape these contexts to generate pathbreaking dynamics (Carvalho, Mingardo, and van Haaren 2012;Torrens et al. 2019Torrens et al. , 2018; van den Heiligenberg et al. 2018van den Heiligenberg et al. , 2017. Reviewing the interpretations of place and space in the literature on urban experimentation, Torrens et al. identify three lenses or metaphors: seedbeds, harbours and battlegrounds (2019). Rooted in SNM, seedbeds represent protective spaces for innovation. They are analysed by exploring the dynamics of shielding, nurturing and empowering. The metaphor of the harbour highlights the connectivity and exposure of cities embedded in transnational and cosmopolitan contexts, connected through multiple flows and mobilities. This strand of literature foregrounds the transnational linkages and knowledge flows between cities, which give rise to innovation journeys across cities (Blok 2014;Blok and Tschötschel 2016;Wieczorek, Raven, and Berkhout 2015). In line with governance studies on experimentation, the metaphor of the battleground portrays cities as sites of political contestation between divergent interests and imaginaries of the future. Drawing on political ecology and social movement theory, this perspective focuses on actor constellations and processes of collective sensemaking.
Following the perspective of seedbeds, Coenen et al. seek to ground SNM in a spatial context (2010) by adopting a conception of proximity which originates in the literature on economic geography and emphasises the proximity of regional innovation clusters (Boschma 2005). Cognitive proximity refers to knowledge and competence, organisational proximity to the structuring of relationships in formal organisations, social proximity to ties of trust and friendship, institutional proximity to institutional thickness through shared values and rules, and geographical proximity to the physical-material distance between actors. Coenen et al. relate these dimensions of proximity to the three dynamics of SNM: the building of social networks, the articulation of expectations and second-order learning (2010). In line with Boschma, they argue that an optimal environment for innovation emerges if actors, institutions and places are at a medium level of proximity, i.e. neither too close nor too distant.
Building on the geography of sustainability transitions, Hodson et al. combine sustainability transitions studies with urban geography to explore the spatial dynamics of transitions (2017). They seek to advance a contextual understanding of transitions, elucidating how institutional contexts shape socio-technical processes in particular places. Moreover, they introduce the notions of multiplicity and reconfiguration to move beyond an understanding of experiments as single interventions, bound in time and space, towards an understanding of experimentation as a fluid and reflexive process (Hodson, Geels, and McMeekin 2017). This aligns with the understanding of experimentation as a governance process in governance studies (see Section 3.2). The notion of multiplicity is constituted by three elements: multiple experiments with innovations, multiple forms of governance and multiple conceptions of sustainability. Hodson et al. argue that urban transitions are not bound to particular innovations but are about 'how multiple innovations are experimented with, combined and reconfigured in existing urban contexts' (ibid., 1). The argument that multiple institutional arrangements influence urban experimentation underlines the embeddedness of urban experiments in multi-level governance contexts, and illustrates how urban governance institutions and experimental processes are framed by national institutions. In the political struggle between diverse societal interests, various conceptions of sustainability are mobilised. This points to the politics of experimentation as discussed by governance scholars (see Section 3.2). The debate on the multiplicity of experimental processes and contextual reconfiguration is evidence of an emerging dialogue between scholars of sustainability transitions and of environmental governance.

The literature on environmental and climate governance
In the literature on environmental governance, experimentation is studied as a novel mode of governing the city in the face of uncertainty. This literature originates in research into environmental governance, climate urbanism, urban political economy and ecology (Sengers et al. 2016;Turnheim, Kivimaa, and Berkhout 2018b). Local experimentation reflects the renewed emphasis on polycentric governance and the local level. The notion of 'climate change experiments' (Bulkeley and Castán Broto 2013) or governance experiments describes the shift to new techniques of urban governance which embrace uncertainty and contingency (Karvonen 2018;Karvonen and van Heur 2014). If experiments are viewed as a mode of governing, this represents a conceptual shift. Rather than functioning as a kind of bounded intervention, experimentation is 'an ongoing effort across multiple projects, domains, and longer time-frames' (Sharp and Raven 2021, 202). Experimentation transforms urban governance itself as it realigns old and new forms of governance. This implies that experiments do not unfold their impact through scaling and diffusion, but through processes of reconfiguration and embedding (Bulkeley 2021;Castán Broto and Bulkeley 2018).
Governance through experimentation moves beyond traditional forms of hierarchical, top-down governance (vertical logic) towards novel forms of network, participatory and bottom-up governance (horizontal logic) (Healey 2004). This has broadened the scope of those participating in urban governance from public actors to encompass a broad range of private agents across civil society. Governance experiments are a form of institutional innovation that seeks to strengthen participatory forms of governance and create new partnerships, blur the boundaries between public and private authority, and reform traditional modes of urban planning by introducing collaborative planning and integrating resilience and adaptive co-management in planning processes ( Experimentation is not applied as a specific research method, but rather as a governance approach to learn and innovate ). It is seen as a mode of reflexive governance, enabling learning from real-life interventions to provide situated feedback and facilitate evidence-based policymaking Karvonen 2018). This emphasis on learning aligns with the notion of 'the city as a machine for learning' as depicted in the literature on reflexive governance (McFarlane 2011). Learning is described as an incremental and iterative process of learning-by-doing and doingby-learning (Ansell and Bartenberger 2016;Karvonen, Evans, and van Heur 2013). Learning in sustainability transitions is defined as a reflexive process, which can be broken down into three types: . technical or single-loop learning, i.e. adopting alternative technological solutions; . collaborative/social or double-loop learning, i.e. building new networks and collaborations; . conceptual learning or triple-loop learning, i.e. developing new dialogues and discourses on norms and values (Argyris and Schön 1978;Glasbergen 1996;Pahl-Wostl 2009).
However, the origins and impacts of learning from experimentation on urban governance and sustainability transitions remain under-explored .
Scholars of governance call for closer engagement with the politics of experimentation so as to shed light on the underlying interests, ideologies and imaginaries (Bulkeley, Castán Broto, and Edwards 2015;Caprotti and Cowley 2017;Karvonen, Evans, and van Heur 2013). This governance perspective defines cities not merely in terms of the spatial scale they represent, but as actual social and political entities. They are the places where the questions of 'who gets what, when, how' (Lasswell 1936) are deliberated and materialise in the everyday lives of citizens. Experiments within cities and communities are of an inherently political nature. On the one hand, they frequently spark conflict and controversy by seeking to challenge and possibly dismantle existing socio-technical configurations (Bulkeley, Castán Broto, and Maassen 2014a;Bulkeley and Castán Broto 2013;Caprotti and Cowley 2017;Crowe, Foley, and Collier 2016;Edwards and Bulkeley 2018;Evans 2016;Hodson, Evans, and Schliwa 2018;Karvonen, Evans, and van Heur 2013;Savini and Bertolini 2019). This is also described as the 'politics of niches' (Savini and Bertolini 2019). On the other hand, urban experiments have been described as constituting the 'politics of determining solutions' (Evans et al. 2018, 422). Operating through demonstration, they make alternatives visible and tangible while offering practical insights into the implementation of innovations. In so doing, experiments offer a way of creating political legitimacy, along the lines of 'seeing is believing' (ibid. 422).
Here it is important to bear in mind that 'experiments [themselves] are often vested with particular interests and strategic purpose in the governing of the city' (Bulkeley and Castán Broto 2013, 373). Economic and technical rationales of ecological modernisation, technological progress and sustained economic growth may be hidden behind experimentation, as exemplified by the 'twin narrative of climate risks and techno-economic opportunities' (Blok and Tschötschel 2016, 12, emphasis in the original). This suggests a shift from a techno-managerial paradigm towards one of political experimentation, which elucidates 'the politics through which experiments are defined, framed and constituted' (Savini and Bertolini 2019, 845).
Drawing on Foucault's (1986) concept of heterotopia, Edwards and Bulkeley explore these politics of experimentation (2018). Defined by their otherness, heterotopias create spaces for experimenting with utopias. Rooted in enactment, experimentation is a performative mode of practice which enables us to compare utopian and dystopian images of the future. In particular, top-down and often technocratic visions of 'smart urbanism' can be compared with bottom-up and community-focused visions of Transitions Towns (ibid. 5). However, Edwards and Bulkeley also caution against post-political thinking, namely when the fear of some crisis is invoked as an argument for consensus-based politics, and the suggestion that there is no alternative forecloses dissenting voices and any debate on real alternatives (Edwards and Bulkeley 2018;Swyngedouw 2011).
Experimentation may reinforce existing modes of urban governance rather than transform them (Castán Broto and Bulkeley 2013b;Levenda 2018). Change may in some cases constitute a shift from public provision to the liberalisation and privatisation of urban infrastructures (Bulkeley, Castán Broto, and Maassen 2014a). The politics of experimentation are further revealed by the tension between short-term, tangible economic benefits and long-term objectives of equitable development (Chu 2016). Experiments bound in time and space are preferred over long-term and large-scale development projects as they are associated with accountable implementation and manageable costs. Yet this orientation towards economic benefits might sideline questions of social justice and welfare (Chu 2016). Especially in the Global South, the securitisation of resources such as water, energy or land is a major concern to be addressed by experiments (Castán Broto and Bulkeley 2013b;Chu 2016). Unfortunately, path dependencies have shaped the implementation of experiments and hindered the emergence of a culture of experimentation (Levenda 2018;Parks 2019). These have their roots in established power relations, institutional configurations, administrative compartmentalisation and risk aversion, societal norms or the obduracy of material infrastructures. While experiments thus create novelty, they struggle to dismantle unsustainable socio-technical configurations (Bulkeley, Castán Broto, and Maassen 2014a;Parks 2019).
The politics of experimentation also provokes normative questions of inequality and justice, inclusion and exclusion, which are prominent in the debate on environmental and climate justice Edwards 2015, 2014b;Castán Broto and Bulkeley 2013a;Karvonen, Evans, and van Heur 2013). Under this normative perspective, we can ask: Who stands to gain and lose through sustainability transitions? In this context, Bulkeley et al. outline different conceptions of justice (2014b), namely: . international justice, where rights and responsibilities accrue to nation states; . intergenerational justice, where rights and responsibilities accrue to generations; . and cosmopolitan justice, where rights and responsibilities accrue to people.
The notion of cosmopolitan justice highlights the distribution of environmental vulnerabilities and responsibilities within cities and communities, which often corresponds to previous patterns of inequality (Satterthwaite 2008). Against this background, the social implications of experimentation deserve particular attention. In particular, experimentation may reinforce patterns of inequality by strengthening the affluent few and aggravating the risks for the most vulnerable (Bulkeley, Castán Broto, and Edwards 2015;Hodson and Marvin 2010;Smeds and Acuto 2018).
While the shift from governance by hierarchy to governance through networks has broadened the scope of actors and blurred the boundaries between public and private authority, research into the role of municipalities in urban experimentation has shown how these remain key agents. This is due to their public authority based on democratic legitimacy, their accountability to the citizenry through safeguarded transparency, and the power they hold through access to and distribution of resources such as public revenue and public infrastructure ownership (Castán Broto and Bulkeley 2013a;Kronsell and Mukhtar-Landgren 2018;Madsen and Hansen 2019;Mukhtar-Landgren et al. 2019). This democratic legitimacy constitutes a fundamental difference between public and private actors.
Research into the role of municipalities in experimentation is informed by literature on cities in environmental and climate governance as well as governance and institutional theory (Kronsell and Mukhtar-Landgren 2018). With reference to debates on the politics of enabling or 'governance through enabling' (Bulkeley and Kern 2006;Kern and Alber 2008), municipalities are portrayed as enablers of local sustainability governance (Kronsell and Mukhtar-Landgren 2018;McGuirk et al. 2015;Mukhtar-Landgren et al. 2019). As governance by hierarchy has shifted to governance through networks, their roles have transformed from representatives and civil servants to intermediaries, facilitators and networkers (Mukhtar-Landgren et al. 2019).
Combining the logics of hierarchical governance through formal steering and authority, and network governance through participation, Kronsell and Mukthar-Landgren developed a typology of four municipal roles: promoter and enabler (governance by hierarchy), and partner and a non-role (governance through networks) (2018). By comparing Urban Living Labs (ULL) in countries across Europe, they show that municipalities most commonly adopt the partner role in ULLs, where partnership is based on shared leadership and participation on equal terms. Despite the highly diverse institutional settings of the countries they consider, similarities are evident as all four roles are found to be represented in each country. This suggests that institutional settings have little influence on the roles adopted by municipalities in experimental governance (Kronsell and Mukhtar-Landgren 2018;Mukhtar-Landgren et al. 2019).
Governance scholars also illustrate the embeddedness of urban experimentation in multi-level governance contexts by showing how experiments are contingent uponand may be impeded bynational policy contexts, creating a funding competition between cities and promoting an orientation towards economic growth (Hodson, Evans, and Schliwa 2018). Moreover, scholars highlight the importance of networking within and between cities for creating and sharing innovations. By studying networked forms of governance, they explore the processes of diffusion in multi-level contexts, namely the vertical scaling between different governance levels and the horizontal replication of experiments between cities (Kern 2019; Smeds and Acuto 2018). The interlinkages between the global and the local level are further illustrated by a study of the piloting of the Sustainable Development Goal 11 'Sustainable Cities and Communities' in Cape Town, South Africa, helping to inform the formulation of the SDGs by the United Nations (Patel et al. 2017).
Another line of inquiry on urban experimentation is the building of local resilience. Here scholars explore how local experimentation can foster the integration of resilience in urban planning practices (Crowe, Foley, and Collier 2016;Evans 2011). Rooted in the literature on urban political ecology, this line of research applies ecological principles to cities, viewed as adaptive socio-ecological systems. When ecological theory is applied to resilience and experimentation in cities, this serves to create a socio-ecological governance framework (Evans 2011). There are two main definitions of resilience: first, as the ability to return to a stable equilibrium after a disruption of the status quo; and second, as a non-equilibrium or socio-ecological change following a dynamic and evolutionary paradigm, thus describing the ability of a system to adjust to constant change and disturbance (Folke 2006;Holling 1973;Pickett, Cadenasso, and Grove 2004). Adopting the perspective of resilience as socio-ecological change, scholars of urban experimentation promote an emancipatory concept of resilience as the ability not only to adapt but also to renew and transform cities (Crowe, Foley, and Collier 2016). At the same time, scholars of ecological resilience warn that the concept of resilience can be instrumentalised to normalise cycles of crisis and recovery, thereby preventing any political dialogue on the deeper causes of crises and how to address them (Evans 2011).

Discussion
The review of these academic perspectives and lines of enquiry on urban experimentation to advance sustainability transitions provokes the questions: In which ways do the literature on sustainability transitions and the literature on environmental governance contrast? And how could they complement each other? By reflecting on developments in the study of urban experimentation, it is possible to outline avenues for future research and potential areas of dialogue between the different research communities.
In order to better understand how experimentation in urban context differs from previous approaches to experimentation (Sharp and Raven 2021), it is necessary to explore the meaning of places and spaces in shaping experimentation and transition dynamics. In particular, the study of the geography of urban transitions aims to reveal this meaning of places and spaces. In this regard, however, Levin-Keitel et al. argue for a better conceptual grounding of the notion of space (2018). In a similar vein, scholars of urban political ecology seek to elaborate both the socio-technical and socio-spatial dimensions of urban transitions by elucidating the flows of power and materials (Castán Broto and Bulkeley 2018;Evans et al. 2018). Analysing the change of urban infrastructure networks, they detail how these networksunderstood as socio-technical regimesare embedded socio-spatially and structured by political economies and ecologies (Bulkeley, Castán Broto, and Maassen 2014a). Further, they propose a conception of agency that is both socially and materially constituted (Bulkeley, Castán Broto, and Edwards 2015). This calls for a dialogue between both strands of literature to study the geography of urban transitions with their specific histories and path dependencies, thereby illuminating the relationship between the social and the spatial in urban experimentation. Future research could elaborate how the proximity of local settings affects experimentation, especially at the different scales of cities, precincts and neighbourhoods (Sengers et al. 2016;Sharp and Raven 2021; van den Heiligenberg et al. 2018).
While there have been calls to address the politics and power dynamics of sustainability transitions in order to reveal underlying interests and controversies (Avelino et al. 2016;Avelino and Wittmayer 2015;Meadowcroft 2011;Patterson et al. 2017), this largely remains a blind spot within transition studies. Under the multi-level perspective, transitions are framed as regime shifts, whereby innovative niches challenge and ultimately transform existing regimes. By introducing the notion of multiplicity, transition scholars point to the contested notions of sustainability within cities, analysing how and why these are mobilised by different stakeholders (Hodson, Geels, and McMeekin 2017). In so doing, they seek to develop an integrative perspective that bridges transition studies and governance studies.
By contrast, the politics of experimentation are integral to the study of urban experimentation by governance scholars. Indeed, experimentation is seen as a process rooted in the politics of sustainability transitions. Adopting the concept of heterotopias, scholars have examined how competing notions of sustainability and urban development are trialled and deliberated through experimentation (Edwards and Bulkeley 2018). They have critically interrogated the practice of experimentation and highlighted the dynamics of empowerment and disempowerment through experimentation. By broadening the range of actors and blurring the boundaries between public and private spheres, experimentation raises questions of authority, accountability and legitimacy Voß and Schroth 2018). Governance scholars have entered into the debate on climate justice to show how local experimentation may reinforce existing inequalities instead of alleviating them, arguing that experiments can be co-opted by established interests and thus strengthen rather than transform neoliberal modes of governance (Bulkeley, Castán Broto, and Maassen 2014a;Evans and Karvonen 2014;Evans et al. 2018;Karvonen 2018;Karvonen, Evans, and van Heur 2013;Karvonen and van Heur 2014;Marvin et al. 2018). Local experiments can overshadow issues of profound systemic change such as a paradigm shift in economic thinking, whereby economic ideals are reoriented from efficiency and growth to sufficiency and post-growth (Feola 2020;Kivimaa et al. 2017). This critical reading implies a shift from a technical-functionalist understanding of experimentation towards one which highlights its deeply political nature. It calls for transition scholars to engage more deeply with the politics and governance of experimentation and to trace empirically how experiments affect democratic accountability and social equity.
Beyond the local scale, urban experimentation is embedded in contexts of multi-level governance. Transition scholars underscore this embeddedness by defining it as an element of the multiplicity of experiments and by studying how local experimentation is influenced by multi-level governance contexts (Ehnert et al. 2018;Hodson, Geels, and McMeekin 2017). This again seeks to integrate both academic perspectives. Experimentation with different approaches is an idea central to polycentric governance (Ostrom 2010). Building on the study of multi-level and polycentric governance, scholars analyse the interactions between multiple governance levels to explore the scaling up of local experiments (Kern 2019) and the sharing of knowledge from local experiments through networked forms of governance such as city networks (Madsen and Hansen 2019;Smeds and Acuto 2018). This implies that concepts and insights from the literature on multi-level and polycentric governance could be integrated in future research on urban experimentation to better understand how the interactions and dynamics in governance settings mediate local experiments (Bache and Flinders 2004;Enderlein, Walti, and Zurn 2012;Jordan et al. 2018;Kern and Bulkeley 2009). Such scholarship could examine how novel forms of governance such as transnational networks contribute to knowledge generation and transfer, orunder a more critical perspectiveillustrate how national contexts may constrain the possibilities for local experimentation and innovation, and how city networks can privilege pioneers instead of integrating diverse communities (van der Heijden 2018).
Processes of diffusion are related to this embeddedness of local experiments in multilevel/polycentric governance contexts, even though the two bodies of literature are often treated separately (Tosun 2018). In order to capture such diffusion processes, transition scholars have defined the transition mechanisms of scaling up, broadening and replicating ( van den Bosch and Rotmans 2008; van den Bosch-Ohlenschlager 2010;von Wirth et al. 2019). Scaling up is understood as the impact of niche innovations in transforming socio-technical regimes; broadening and replicating are the transfer of these innovations to other contexts. Some studies have highlighted the dynamics of de-contextualisation, transfer and re-contextualisation as insights and lessons are replicated elsewhere (Geels and Deuten 2006). Governance scholars have studied these processes of knowledge transfer in the context of multi-level governance and transnational city networks as networked forms of governance (Kern 2019;Madsen and Hansen 2019;Smeds and Acuto 2018). They have further framed world cities as cosmopolitan risk communities to show how these share not only knowledge (cognitive dimension), but also values and identities (normative dimension) (Blok 2014;Blok and Tschötschel 2016). In contrast to a systemic conception of mechanisms of change, a socio-institutional perspective frames policy diffusion and transfer as governance processes. Future research could consider studies of policy diffusion and knowledge transfer in multi-level governance systems to examine the transferability of lessons learnt from local experiments to other cities and communities (horizontal direction), and across local, national and international governance levels (vertical direction) Marsh 2000, 1996;Hakelberg 2014;Marsh and Sharman 2009;Stone 2012). This could improve our understanding of the processes of de-contextualisation, transfer and re-contextualisation, as well as how local lessons are transformed as they are transferred (Stone 2012).
In regard to the impact of urban experiments, the question arises: What follows? And more particularly: How can these interventions be embedded in the mid-and long-term? Transition scholars describe embedding as a mechanism of anchoring innovations in established structures, cultures and practices. Some reconceive embedding as a form of institutional change that entails processes of de-institutionalisation and re-institutionalisation, thereby integrating insights from neo-institutional theory into the study of sustainability transitions Geels 2004). Governance scholars argue that experimentation is an innovation in governance and urban modes of governance need to be transformed themselves to enable transformative change within society (Evans 2016;Karvonen, Evans, and van Heur 2013). This implies some convergence between the two research communities in the adoption of institutional conceptions. Future research could explore how novel modes of experimental governance and traditional modes of urban planning and development could be realigned as well as elaborating how experiments could be embedded in existing governance arrangements Voytenko et al. 2016). To better understand these processes of institutional change, it is important to engage with the literature on neo-institutionalism (Davies and Trounstine 2012;Peters 2005;Rhodes, Binder, and Rockman 2008;Schmidt 2008). Moreover, governance scholars point out that local experiments should not be analysed solely as single units, but rather the focus must be on the collective effort of experimentation and the relationships between multiple experiments (Madsen and Hansen 2019). Transition scholars also highlight the challenge of projectification and projectified governance settings. The imposition of a project logic on experiments could reinforce technical thinking and depoliticise experimentation instead of encouraging reflexivity and systems orientation (Torrens and von Wirth 2021). This raises the question of how to move from short-term interventions to the long-term governing and reconfiguration of transition pathways through experimental governance Karvonen, Evans, and van Heur 2013). The transition trajectories which emerge from linked experiments could be studied to illustrate how they are shaped by and reshape urban contexts.
One persistent research gap in the study of urban experimentation is the role of agency. While transition scholars have examined the role of intermediaries in aggregating lessons and upscaling local innovations (Matschoss and Heiskanen 2017), there is limited understanding of micro-level agency. The role of these intermediaries should be explored at greater depth, as they can be important mediators and translators between the old and the new (Torrens et al. 2019). Future studies could build on conceptions of agency in environmental governance (Betsill, Benney, and Gerlak 2020;Liefferink and Wurzel 2017;Wurzel, Liefferink, and Torney 2020), conceptions of actor roles, institutional entrepreneurs and transformative agency in sustainability transitions (Avelino and Wittmayer 2015;Schot, Kanger, and Verbong 2016;Westley et al. 2013;Wittmayer et al. 2017) and especially of transition intermediaries (Kivimaa et al. 2019a(Kivimaa et al. , 2019b to analyse how the micro-level of agency interacts with meso-and macrolevel change. In particular, scholars could analyse how the question of who is included or excluded shapes the form and impact of experiments (Bulkeley et al. 2019;Voytenko et al. 2016).
Learning is a defining feature of experimentation, which is meant to be a form of reflexive governance that encourages learning-by-doing. However, the literature on urban experimentationboth within transition studies and governance studieshas barely engaged with the concepts of learning processes (Ansell and Bartenberger 2016;Bellinson and Chu 2019;Gerlak et al. 2018;Singer-Brodowski, Beecroft, and Parodi 2018;Wolfram et al. 2019). This mirrors a gap between learning theory and sustainability transitions studies (van Mierlo et al. 2020). Future research could adopt concepts of learning theories to elucidate how learning emerges from and refines the practice of experimentation. This should also account for failure and learning-by-failing. Scholars could explore unlearning or learning to resist change, and study the unknown relationship between superficial learning (learning related to everyday practices) and deep learning ( van Mierlo and Beers 2020). Further, they could explore learning by incumbents to show how innovation can emerge within the regime. This implies moving beyond the niche-regime dichotomy (Avelino et al. 2016) to generate insights on how experimentation creates intermediary spaces for bridging the niche and the regime, and how to foster learning within regimes. It could also provide insights on how learning and reflection on experimental governanceso-called meta-learning (Wolfram et al. 2019)could be institutionalised. However, scholars also caution against an instrumental perspective on learning, which too easily equates learning with the solution of social and political challenges (van Poeck, Östman, and Block 2020).

Conclusion
Urban experimentation to promote sustainability transitions is a newly emerging field of study, and one which has provoked great interest in different research communities. This has led to much complexity and ambiguity in the understanding of experimentation. The literature review presented above has elaborated different academic perspectives and lines of enquiry on urban experimentation while discussing how scholars of sustainability transitions and environmental governance conceive such experiments. In particular, transition scholars are found to view experiments as single interventions to develop innovative niches. Adopting a systemic understanding of the mechanisms of change, they highlight the importance of the scaling up and diffusion of these niche innovations to transform established regimes. By contrast, governance and policy scholars frame experimentation as a mode of governing cities in contexts of contingency and uncertainty. They propose an understanding of experimental governance as trajectories of linked experiments and deeply political processes. This is rooted in the assumption that change does not result from scaling-up or diffusion but rather from the reconfiguration of socio-technical and governance systems.
A comparison of these different lines of enquiry allows us to reflect on the study of urban experimentation, revealing blind spots and encouraging dialogue. The review elaborates the potential of the experimental turn while also representing a critical reading. It cautions against overestimating the transformative potential and calls for an examination of the politics of experimentation. On a critical note, local experiments are often seen to focus on single elements within sectors, while systemic change relies on cross-sectoral and co-evolving dynamics. Local experiments thus need to be seen in the context of urban sustainability transitions, where actors and processes other than experimentation influence transition trajectories. In line with the notion of multiplicity, experimentation should be conceptualised as one process within more complex transition dynamics. This has implications for experimental governance, underlining the importance of moving beyond short-term interventions and individual experiments towards continuity and the long-term governing of sustainability transitions. Experimental governance should be integrated into a holistic framework of transition governance, one which links experimental governance with traditional forms of urban planning and development. Scholars could explore this in the future by engaging in dialogue and by treating different academic perspectives as complementary.