Table 1 Distinctiveness of urban experimentation

Urban experimentation relates closely to the notion of ‘reflection in action’, whereby ‘to experiment is to act in order to see what action leads to’ (Schön, 1983, p. 144). Compared to other forms of urban development or policy, urban experimentation comprises a variety of experimental logics that share an inclination for learning from real-world interventions (Evans et al., 2016), with activities situated in real-world places; oriented towards producing changes with an emphasis on improvement or transformation; and an embrace of contingency and uncertainty (Karvonen and van Heur, 2014). Each of these ‘accomplishments’ can be organised in diverse ways (Bulkeley et al., 2019), leading to seemingly contradictory ideas of what constitutes experiments. Crucially, a processual perspective on experimentation is emerging in the literature that moves beyond ‘the context of individual experiments or collections of experiments to conceive of urban experimentation as a process that materially embeds priorities, that seeks to make them durable, through experimentation in particular places’ (Hodson et al., 2018). The current wave of experimentation has a distinctively urban character and is linked to efforts to deal with the complexities, uncertainties and contestations of ongoing urban transformation processes

Reviews mapping the rationales informing experimentalism highlight how diverse experiments can be. For instance, Sengers et al. (2019) identified five conceptualisations within the transitions’ perspective, differing in normative orientation, theoretical foundation, analytical emphasis, and main actors involved. Caniglia et al. (2017), writing from a sustainability science perspective, proposed a six-fold typology, differentiating the focus being either on problems or solutions, and three levels of control: full control, participatory control, and no control. However, that perspective foregrounds the production of evidence for decision-making, contrasting with the idea that experimentation aims to induce changes and improvements in the real world. Ansell and Bartenberger (2016), in turn, highlight distinct logics that can guide experimentation. Those authors distinguish controlled experimentation, based on highly controlled interventions aimed to test particular hypotheses in a deductive manner; Darwinian experimentation, oriented towards generating variety from which best practices can be selected through an inductive approach; and generative experimentation, which applies an abductive logic to interactively redesign and refine a prototype until it meets stakeholders expectations. Crucially, they highlight that learning generated through experiments is not only scientific or technical, but also ‘political learning’, by which ‘stakeholders may alter their preferences, goals, frames, and commitments’ (Ansell and Bartenberger, 2016)

Each of these typologies hints at a very rationalist and neutral process of designing experiments. In practice, however, how exactly experiments are set up depends on a contested process of negotiating priorities, epistemological assumptions, and normative goals while trying to create viable setups. As a result, experiments end up enrolled into varying processes of change. For instance, they could be used to generate ‘single loop’ learning about improving performances or particular artefacts or services that conform to existing rules (e.g. testing a smart energy meter); ‘double loop’ learning that inquires the rules and structures of a given system and prompts reforms (e.g. exploring options for energy retrofit with provision models such as community energy); or ‘triple loop’ learning, seeking to transcend and transform such rules and open up previously unimagined possibilities with new values and principles (e.g. establishing self-governed post-capitalist eco-settlements) (Waddel, 2016, cited in Fazey et al., 2018). However, only some of these various forms are formally recognised and labelled experiments or labs, which privileged certain forms of action and certain forms of knowledge as desirable and legitimate, constituting a political process with biases and normative assumptions (Savini and Bertolini, 2019)