Category: Contribution

Introduction

‘Ayuda Ecuador,’ a deployment of the Ushahidi platform, was launched on April 16, 2016, after the first tremors of the earthquake that killed 660 people and injured 4,605 (USAID 2016). Within a matter of two hours, a group of ‘digital humanitarians’ – who typically comprise bloggers, techies and activists willing to volunteer their time to assist in emergencies remotely (Gutiérrez 2018a) — launched the deployment to ‘generate collectively data relevant to the emergency, threats, logistic needs and response that the affected population was experiencing (…) and effectively channel the efforts by different institutions and agencies’ (Ayuda Ecuador 2016).

The Ushahidi platform – a non-profit tech company from Kenya that allows the visualization of crowdsourced crisis data for humanitarian purposes – is a proactive data activist organization, proactively employing software and data for humanitarian assistance (Gutiérrez 2018a).

Data activism involves a series of practices ‘at the intersection of the social and the technological dimension of human action,’ aiming at either ‘resisting massive data collection,’ in the case of reactive data activists, or ‘actively pursuing the exploitation of available data for social change,’ in the case of proactive data activists (Milan and Gutiérrez 2015, 127). This article deals with the second type of data activism.

Proactive data activism is currently being harnessed to understand, analyze and develop solutions to a range of social problems, from climate change and biodiversity loss, to inequality and human rights abuses (Hogan and Roberts 2015), as well as to assist in humanitarian crises, as in the case of the Ushahidi platform.

Because of its capacity to generate alternative digital public spheres (called here APSs for short), this paper specifically refers to a type of proactive data activism termed ‘geoactivism,’ which employs interactive cartography to communicate and trigger action. In this paper, Geoactivism galvanized to address an emergency is called ‘crisis mapping,’ while geoactivism employed in evidence-gathering for advocacy is called ‘activist mapping’ (Gutiérrez 2018a).

The cases mobilized to illustrate this article include mainly deployments of the Ushahidi platform, based on crowdsourced data, as well as other cases, including InfoAmazonia, which relies on public data, as well as crowdsourcing and sensors to generate data. They are introduced in the analysis because they are typical geoactivist organizations (Gutiérrez 2018a).

To examine how APSs behave, I draw on ‘Why the net is not a public sphere’ (Dean 2003). Particularly useful are Jodi Dean’s reflections on how the net cannot be considered a public sphere, but a set of democratic configurations that she calls ‘neodemocracies’ (ibid., 105). Dean considers that the architecture of the public sphere is based on a set of components – specifically, site, goal and vehicle, as well as means and norms (2003, 96). I take this idea further to highlight facets of the APSs (see Table 1).

The primary purpose of this paper is not to compare the public sphere and Dean’s neodemocracies, but to contribute to a theory of data activism, a relatively unexplored phenomenon. Therefore, the theoretical comparison provided in this paper is a heuristic exercise for enlarging the concept of the APS in data activism.

Discussion

In this section, some of the central concepts and theories mobilized for this study are explored and put in the context of data activism, including Dean’s neodemocracies (2003), Jürgen Habermas’s public sphere and its criticism, James Bohman’s electronic public space (2004), Nancy Fraser’s counterpublics (1990), and the theory of dataveillance (Poell, Kennedy and Dijck 2015). A brief comparison of the public sphere, the neodemocracies and the APSs is offered, which is explored in the next section.

According to Dean, the Habermasian public sphere arises in the nation, is aimed at consensus, abides by the rules of inclusivity, equality, transparency and rationality, and revolves around actors (2003). Meanwhile, Dean’s internet-based neodemocracies aim at contestation, happen on the web via networked conflict occurring within the canons of duration, hegemony, decisiveness and credibility, and emerge around issues (ibid.). The concepts emerging from this taxonomy are employed here probingly to observe the APSs generated by data activism.

I do not entirely subscribe to Dean’s criticism of the Habermasian understanding of the public sphere. Dean’s criticism does not take into consideration how the concept of the public sphere has evolved in Habermas’s thinking. Habermas updated it, suggesting that ‘a cosmopolitan matrix of communication’ could make a ‘world public sphere’ possible (Habermas 1996, 514), decoupling the public sphere from the nation. Therefore, when I uphold the idea that the public sphere is linked to the nation, I am referring to Dean’s critique.

Besides, the Habermasian public sphere is not as placid as Dean portrays it, since it also deals with struggle and is intended for action. In the public sphere, discourse commences once there is a problem and actors can contest each other’s validity claims (Habermas 2006, 413). In fact, the public sphere is the place where differences are settled by reason, and not by war (Negt and Kluge 1993, ix). Likewise, when I mention consensus as the goal of the public sphere, I am using Dean’s views.

One of Dean’s criticisms is that adding an ‘s’ to the concept does not save it, because groups either share the same norms (and therefore are a public in Habermasian terms) or they do not (and therefore are not a public) (Dean 2003, 97).

This distinction does not apply easily to the APSs created by data activism. APSs may share the same norms, underlying values and solidarity, but their evanescence makes them plural and ad hoc. They are similar to Stefania Milan’s karst river movements, which remain dormant, reappearing when conditions require it (2013), while their members feel part of the same movement and principles (Grabowski 2012). They match Donatella della Porta’s and Mario Diani’s description of a network organisation combining formality (i.e. norms) with elements typical of a distributed network, based on the independence of single components which adapt and change as the occasion demands (2006, 159).

Dean’s taxonomy arranges the features of the public sphere and the neodemocracies into boxes whose boundaries blur once you look more closely at them. Conscious that any classification is arbitrary, I employ Dean’s taxonomy as a heuristic tool for reflecting on data activism.

Thus, APSs’ frame of reference is the local and the global concurrently (not the national or the web), since they are ‘transnational,’ in Bohman’s terms (2004). They are directed at action (not solely at consensus, as Dean asserts for the public sphere), and are established around conflicts (not actors or simple issues).

 

Table 1: Comparing the public sphere, neodemocracies and APSs

	Public Sphere	Neo- democracies	APSs
SITE	Nation	Web	Local/global
GOAL	Consensus (legitimization)	Contestation	Action (via consensus)
VEHICLE	Actors	Issues	Crises and problems
MEANS	Procedures (legal, national)	Networked conflict	Networked coordination (via procedures)
NORMS	Inclusivity Equality Transparency Rationality	Duration Hegemony Decisiveness Credibility	Inclusivity and evanescence Equal access Transparency Credibility

 

Source: Elaboration by the author based on Dean (2003, 108).

This new taxonomy allows me to make some observations about how APSs behave.

Other authors’ arguments, compatible or not, are employed later in this paper for thinking about APSs, although the intention is not to subscribe to any theory of the public sphere.

For example, Bohman’s reflection on ‘a new kind of public sphere’ that emerges in the era of ‘electronic democracy’ (2004, 131) is useful. Bohman acknowledges that an ‘electronic public space’ – understood as a computer-mediated, Internet-enabled communication — may have the potential of generating a ‘novel public sphere,’ enlarging the scope of ‘certain features of communicative interaction across space and time,’ especially when applied to ‘deliberative transnationalism’ (2004, 132-151).

Data activism could be considered an enhanced variety of activism in that it is enabled by a cutting-edge form of technology (i.e. the data infrastructure), and data activists often use the Internet as a platform for disseminating information (together with other information and communication technologies or ICTs). Bohman’s viewpoints are therefore interesting for considering how data activism employs technology to generate novel public sphere(s) for transnational advocacy and humanitarianism.

This article draws too on Fraser’s look at marginalized groups excluded from the public sphere, and the formation of their post-bourgeois public spheres as ‘subaltern counterpublics,’ overcoming a ‘hegemonic’ mode of domination (1990, 61-62). These groups generate counterpublics, or ‘parallel discursive arenas,’ to develop and circulate ‘counter-discourses,’ which present alternative ‘interpretations of their identities, interests and needs’ (ibid., 67).

Although there is a difference between Fraser’s bottom-up counterpublics and the spaces generated by proactive data activism, her conceptualization is useful here too. In 2008, four volunteering techies and bloggers, who could hardly be described as ‘marginalized’ in spite of being ‘repressed,’ launched the first deployment of the Ushahidi platform.

This first application was set up to bypass an information shutdown during the bloodbath that arose after the presidential elections in Kenya a year earlier, and to give voice to the anonymous victims. The deployment allowed victims to disseminate alternative narratives about the post-electoral violence. Since then, other applications have been set up by campaigners, organizations and digital humanitarians in every major disaster not only to visualize what is happening but also to assist people affected in real time, offering some ‘expectation of a response’ and opening a ‘space for interaction,’ in Bohman’s words (2004, 135). Namely, the APSs of data activism could be considered subaltern in that they are parallel discursive arenas where people disseminate counter-discourses in opposition to top-down narratives and approaches.

This paper also takes into account the backdrop in which proactive data activism emerges, which includes the datafication process – that is, computing aspects of life that had never been quantified before (Mayer-Schönberger and Cukier 2013, 78) — and the massive collection of data by governments and corporations, giving rise to ‘computational politics’ (Tufekci 2012), what Braman calls the panspectron (2009, 314) or dataveillance (Poell, Kennedy and Dijck 2015). Dataveillance employs ‘the massive flows of data circulating between devices, institutions, industries and users’ for ‘new and troubling practices’ (ibid.). Reactive data activism, a frontrunner of data activism, emerges precisely in answer to this data-gathering and intrusion (Milan and Gutiérrez 2015).

Labels employed in this paper are derived from Gutiérrez (2018a), where eleven attributes observed in forty data initiatives (i.e. how collaborative a concrete organization is), thirty semi-structured interviews with data activists, practitioners and experts, and four in-depth interviews with Ushahidi developers and deployers were the basis for a non-normative taxonomy employed in examining proactive data activism (ibid.). According to this classification, proactive data activists can be categorized as: skills transferrers, focused on transferring data or social science skills, forging opportunities for alliances and generating digital platforms and tools for data activism; ‘catalysts’ who chiefly fund data projects; activists who produce data journalism, when news media organizations are not willing or capable of doing so; and the actual data activists and geoactivists (ibid.).

The ability of some proactive data activist initiatives to generate APSs is the pivotal idea in this paper. Happening both locally and globally across borders (site), aiming at action (goal) and gathered as a temporary network around the crisis caused by the earthquake (vehicle), ‘Ayuda Ecuador’ is an example of how data activism can generate temporary spaces for dialogue and action.

The ‘deployers’ at ‘Ayuda Ecuador’ included several organizations from different countries, supported by four experts: Luis Hernando Aguilar, on knowledge management; Byron Herrera, on software development; Estela Navarrete, on mobile software development; and Angela Oduor, from Ushahidi (Ayuda Ecuador 2016). The application connected them and other digital humanitarians with victims (reporters) generating data via the crowdsourcing platform, and traditional humanitarian agencies, using the resulting information on the ground. For almost three weeks, these three communities – that is, deployers, reporters and humanitarian workers — ‘met’ in the public space generated by the platform to agree, make decisions and coordinate actions.

The theory of the public sphere is employed here to analyze these spaces. It was Habermas who described the public sphere’s emergence in the 17th and 18th centuries as space where a communicative action leads to mutual understanding (Habermas 1984, 17–18). However, Habermas was criticized for having created too abstract a model, and for having bypassed some challenges, such as the exclusion of women and men from the lower classes, and the heterogeneity of the public (Webster 2006, 163–68; Calhoun 1992).

Nevertheless, the idea of the public sphere has been revived with the emergence of ICTs, the ‘networked information economy’ (Benkler 2006, 212), and the data infrastructure — including databases, algorithms, servers and systems needed to obtain, curate, analyze and visualize data (Gutiérrez and Milan 2017) —, due to the fact that these technologies appear to augment opportunities for interaction and debate (Rheingold 2002; Papacharissi 2002; Benkler 2006). Habermas himself reviewed the concept, suggesting that ‘a world public sphere’ could become possible (1996, 514).

However, in the ‘network society’ (Castells 2010b), the public sphere does not remain intact. The rise of ‘digital contention’ generates a ‘structural transformation of the public sphere,’ where access, participation and communication ‘are constantly redefined and renegotiated’ in ‘multiple arenas‘ (Celikates 2015, 14).

This revival, again, met with more criticism. Distinctions between the bourgeois and proletarian public sphere (Negt and Kluge 1993), national and translational publics (Bohman 2004), the Habermasian public sphere and the ‘networked public sphere’ (Friedland, Hove, and Rojas 2006), as well as discussions about whether the concept of the public sphere could be applied to the net (Dean 2003) and social media (Fuchs 2013) ensued, leading the debate towards ‘opposing assessments’ (Bohman 2004, 132).

For Lewis A. Friedland, Thomas Hove and Hernando Rojas, the ‘centrality of networks’ demands revisions of the public sphere (2006, 5). These authors argue that the ‘networked public sphere’ presents challenges to the public sphere in that it raises questions about the structure of communicative action, as the institutions of socialization are now networked, transforming ‘core forms of intersubjective communication and sociation’ (ibid., 24).

Echoing Oskar Negt’s and Alexander Kluge’s proletarian public sphere, Christian Fuchs critiques ‘the contemporary web’ and social media as false emancipatory public spheres because they are dominated by capitalist interests (2013). He adopts the notion of ‘proletarian counter-public sphere’ as a ‘radically different’ public sphere (Fuchs 2010, 176).

Concurring, Dean notes that the net is ‘the architecture for communicative capitalism, both as order establishing itself and as an order being resisted’ (2003, 105). The online presence of activists supports her arguments that resistance and discord are the goals of the web (ibid.). There is a different way to look at it. The data infrastructure embeds several forms of bias, discrimination and asymmetry, which do not disappear when activists use it for their causes (Gutiérrez 2018a, 2018b). However, I do not think that these contradictions annul the APSs’ power to foster democratic participation; they simply reflect the complexity of our world, where simultaneous, often clashing, phenomena happen.

Having framed the discussion, what follows is an analysis of the comparison between the public sphere, the neodemocracies, and the APSs.

The Tensions of the Site, the Goal and the Vehicle

Data activism imposes tensions on the public sphere. The following section dwells on them, paying particular attention to three strains (site, goal and vehicle), which are mainly significant because they present the most noteworthy differences between the public sphere, the neodemocracies, and the new APSs.

The Tension of the Site

Several communities typically convene when launching a Ushahidi deployment tackling a crisis: the group of expert ‘digital humanitarians’ (Meier 2015a) setting up and supporting the deployment remotely; the people affected by the disaster or violence, who contribute their data and use the information; and the humanitarian workers, responding to the crisis.

The first group can be based on different geographic locations, taking advantage of the ‘space of flows,’ which allows real-time social practices to happen simultaneously (Castells 1992, 2010a). The second and third groups operate locally, on the ground, and globally, when they communicate in the APSs generated by the platform. ‘Ayuda Ecuador’ offers an example: ‘locals’ (i.e. on-the-ground humanitarian personnel and reporters) and experts from different parts of the world (i.e. the digital humanitarians) create a space where they can discuss and coordinate actions.

According to Dean, the public sphere occurs in the nation, while the neodemocracies are situated on the web (ibid., 108); meanwhile, the new APSs happen both locally and globally, all at once. During the Enlightenment, the public sphere surfaced in the salons of England, France and Germany, a neutral arena of critique against the absolute power of royals (Habermas 1991, 33-34). The conceptions that Habermas mobilized initially to examine the bourgeois public sphere (i.e. ‘State,’ ‘Church,’ ‘aristocracy’ and ‘royalty,’ among others) cannot be fully appreciated outside the framework of the nation.

Meanwhile, Dean states that neodemocracies – understood as democratic configurations which are organized through contestation (2003, 108) — happen on the web, a place of conflict and antagonism, where differences ‘emerge, mutate, link up into and through networks’ (ibid., 105-106).

Instead, the communities that engage in initiating an Ushahidi deployment interact in a space where the local and the global intersect via the web or other communication systems (i.e. mobile phones).

As with neodemocracies, the concept of the nation limits the way proactive data activism can be theorized. In proactive data activism, a global interaction seeks local effects. The nation can be an element in the offline aspects of proactive data activism (i.e. the level of freedom of speech of the country where the crisis happens), but it is not determining. For instance, the first Ushahidi deployment in Kenya managed to bypass national restrictions and to impose an alternative narrative thanks to a synchronized, collaborative effort.

The Tension of the Goal

APSs are created for tangible action. The interactive mapping employed by Ushahidi, for example, depicts different layers of information in diverse formats (i.e. images, graphs, text). These strata are provided by reporters and deployers (i.e. where and when ‘water purification and sanitation’ services are offered on the ‘Ayuda Ecuador’ map) so that others can use them. Namely, verified reports about where water purification and sanitation services are available, and for how long, are offered so people can act on them.

Another example is the geoactivist initiative ‘Rede InfoAmazonia,’ which offers actionable information on the quality and quantity of the Amazon river’s water and issues warnings, so the communities in the West of Pará, Brazil, can make decisions in their daily lives (Rede InfoAmazonia 2016). ‘Rede InfoAmazonia’ is based on a system of connected sensors that capture data based on physical and chemical parameters, which produces cartography, data and notifications via mobile technologies when the water indicators exceed or go below boundaries of normality (ibid.).

Ushahidi largely engages in crisis mapping, increasing the efficiency in response to humanitarian emergencies, while InfoAmazonia often deals with activist mapping, generating data and empowering communities for participating in decision-making processes. Occasionally, data activism engages in contest (which places it closer to Dean’s neodemocracies). That is, crisis mapping is the art of visualizing citizen reports on a map during a crisis to enhance humanitarianism, whereas activist mapping is more focused on galvanizing communities and gathering bodies of evidence that can be used for advocacy (Gutiérrez 2018a).

Consequently, the idea of proactive data activism should be regarded here as a broad grouping that embraces cases with multiple goals related to action, but not all of them are considered in this article.

Habermas regarded the public sphere initially as a locus of rational discussion aimed at understanding (Habermas 1991). In the public sphere ‘everybody’ gets a chance to talk and to be listened to. That is why Dean considers that the Habermasian public sphere’s goal is principally consensus; in contrast, contestation is what neodemocracies aim at (Dean 2003, 109). A democratic theory built around the notion of issue networks aiming at struggle recognizes that antagonism is what gives democracy ‘its political strength’ (ibid., 111).

Meanwhile, the APSs generated by proactive data activism exhibit a propositive teleological pragmatism, as seen in crisis and activist mapping. This makes proactive data activist initiatives ephemeral: they are created only as a response to a crisis or to an issue perceived as problematic, which they seek to highlight, elucidate and even solve. When the difficulty disappears (or the deployment fails), these initiatives tend to fade as well, in the same way as collective action sometimes remains concealed and resurfaces only when the conditions are auspicious, mimicking a karst river (Milan 2013, 137–68).

The Tension of the Vehicle

The crisis or problem that the data-activist endeavor is tackling is the vehicle, if ‘vehicle’ is to be thought as the trigger for the formation of the APS. However, actors are as significant in the APSs as in the public sphere, giving good reason why they deserve some attention here too.

As with citizens’ media, proactive data activism places people at the forefront of content production (Rodríguez 2001, 20). In ‘Rede InfoAmazonia,’ local communities generate data via sensors, contribute to their interpretation and use the resulting information to make decisions. Their data agency is enhanced when the datafication process – typically the monopoly of governments and corporations — is reversed. Thus, these communities’ ‘routine association’ with automatized data collection is transformed (Milan and Gutiérrez 2015, 130).

In the case of crisis mapping, actors include deployers, reporting victims and humanitarian workers. These three communities are fundamental. Useful crisis mapping cannot occur without credible deployers, eager reporting witnesses (reporters) and collaborating humanitarian workers. The challenge in crowdsourcing data does not come from its technical sophistication, but from the convening power of the project. For instance, hundreds of Ushahidi deployments have failed due to the absence of people submitting reports from the ground, transforming maps into ‘dead maps’ (Vota 2012). Likewise, the failure of some Ushahidi deployments to engage humanitarian forces on the ground has led to hiccups as well (Keim 2012).

Actors in APSs are significant in more ways because they can include marginalized people, making APSs close to Fraser’s counterpublics (1990, 61). For example, the Ushahidi deployment in Haiti in 2010 offered visibility and decision-making power to the victims affected by the earthquake. Another example is InfoAmazonia’s ‘Annual cycles of the indigenous peoples of the Rio Toque’ project, which crowdsources and visualizes information from indigenous people (Cardoso 2015). However, APSs can be considered counterpublics, up to a point. Even if some deployers are victims themselves, there are apparent power asymmetries between the tech-savvy deployers, building and managing the platform, and the data reporters, merely using it under predetermined conditions.

Nonetheless, actors are not the vehicle for APSs. Without the emergency or conflict, the APS do not make any sense. The vehicle for proactive data activism is, then, the crisis or problem at hand, and the nature of data activist projects is teleological.

Proactive data activism can be seen as an instance of collective action triggered by conflicts and problems, which then deploys unconventional means – specifically, the data infrastructure used critically – to deal with them.

In contrast, in the public sphere actors are the vehicle. They are conceptualized as rational beings that make coherent choices, based on reason (Habermas 1984, 89-93). As said, the public sphere has been challenged from many fronts, one of them questioning who can be an actor because the bourgeois public sphere was an elitist island for educated, propertied, white men. Its character was ‘equal’ only for the privileged few, but ‘exclusionary’ for others (Calhoun 1992, 223).

Dean states that a democratic theory built around the notion of issue networks can avoid this ‘fantasy of unity’ (2003, 107). In fact, the neodemocracies are focused on issues because the web is structured by networks formed around issues (ibid., 110-111). Although actors are ‘always embedded in networks,’ they are replaceable because networks’ configuration continuously changes as players enter or leave them (Dean 2003, 108), realizing forms of autopoiesis, ‘or network-based self-organization’ (Friedland, Hove, and Rojas 2006). In a Ushahidi deployment, for example, roles are static while people can change responsibilities or move in and out of the APS.

Looking at the element of the vehicle, APSs seem closer to neodemocracies than to the public sphere in that they are characteristically triggered by (problematic) issues and not actors. However, while any neodemocratic network can be gathered around any issue, any problem will not trigger a proactive data endeavor.

Whether an issue is catastrophic or problematic enough (or perceived as such) depends on different factors. Droughts, for example, are amongst the most deadly disasters regarding the loss of lives (Live Science 2013), but because they often occur over long periods in far-away developing countries, most droughts get a fraction of the coverage that any hurricane receives in rich countries (Tzvetkova 2017). Meanwhile, public attention, media reach and emotion are essential in any successful Ushahidi deployment, since they require a high degree of buy-in among actors.

Other Tensions

The new APSs stand in contrast to both the bourgeois public sphere and Dean’s neodemocracies; but the lines which separate them become more imprecise when looking at their means and norms.

The means mobilized by APSs is networked coordination, while the public sphere is only possible through procedures, and neodemocracies ‘open up opportunities for (networked) conflict’ (Dean 2003, 108).

Networked coordination works within the APS through the formation of temporary networks for action, triggered by the conflict or problem and glued together by solidarity and values, enacting a ‘logic of aggregation’ as they assemble individuals from diverse backgrounds (Juris 2012). While the deployment lasts, the deployers become digital humanitarians, a collective category that groups people of different backgrounds, ranks and nationalities.

However, the ‘diversity of individualized actors’ imposes stress on the sustainability of the effort (Juris 2012, 261). In fact, the need for an unremitting exertion inflicts a strain within the APSs of crisis mapping. Volunteer burnout is, for example, a problem in humanitarian emergencies, since they can last for months, demanding a sustained determination and long hours. It is not the individual that guarantees results but the network, which is resilient because their elements are disposable. In that way, data activist organizations resemble the flexible networks described by della Porta and Diani (2006).

From the perspective of norms, the APSs are ruled by inclusivity and evanescence, equal access, transparency and credibility. This places them halfway between the public sphere, whose rules are inclusivity, equality, transparency and rationality, and the neodemocracies, governed by duration, hegemony, decisiveness and credibility (Dean 2003, 108).

Where the public sphere is regulated by ‘inclusivity’ in Habermasian terms, the neodemocracies value duration (Dean 2003, 109). But for APSs duration makes no sense, since they exist only while the conflict or problem persists. APSs get constantly activated and discontinued, and a conditioned ‘inclusivity’ is essential for generating the crowds, networks and communities that keep them alive.

Paradoxically, ‘inclusion’ in crisis mapping is delimited by the rules determining that actors must belong to one of the three communities necessary to launch a deployment. But for Bohman, ‘any social exclusion undermines the existence of a public sphere’ (2004, 134). To think about the inclusion of the APSs, this concept must be qualified.

The APSs are inclusive, but they establish boundaries for inclusion. Ushahidi deployments create spaces for equal participation with concrete rules: anybody can become a digital humanitarian, but they need to register, declare their abilities and potential contributions to the crisis, and then follow guidelines to produce as much verified, actionable information as quickly as possible. These guidelines are not imposed from external actors in a higher hierarchical position, though, but as a result of internal, horizontal processes of evaluation (Gutiérrez 2018a). For the duration of the deployment, the space is open to experts and non-experts, as long as they share the same objectives, can contribute time and comply with the rules. Then, it shuts down.

Volunteering experts are privileged people with access to technology; however, the APSs also include victims and vulnerable people. The only barrier to the victims’ mediated participation is their access to ICTs, especially mobile phones, which in the past few years has been growing [e.g. in 2008, with a population of over 39 million people, Kenya had experienced a sharp growth in mobile technology penetration, with more than 16 million subscribers (Oteri, Kibet, and Ndung’u 2015)]. In the case of InfoAmazonia’s ‘Annual cycles of the indigenous peoples of the Rio Toque’ project, participation is even more mediated.

Therefore, APSs are ruled by impermanence and qualified inclusion.

In the context of antagonism, ‘neodemocratic politics are struggles for hegemony’ (ibid. 110), and the public sphere is governed by the norm of equality. Meanwhile, the APSs are ruled by equal access.

In Ushahidi deployments, for instance, all actors can participate in the decision-making process: everyone has a parallel capacity of generating and accessing information.

However, equal access does not mean equality here. There are three primary power asymmetries that can be observed in crisis mapping. First, by verifying the information reported by witnesses, deployers turn into gatekeepers of what is public in an APS. Namely, they are placed in a higher echelon. Second, although victims can include deployers, being a deployer is not as dangerous as being a witness or a conventional humanitarian worker, and there is an abyss of difference between working on a disaster remotely and confronting the physical conditions of an emergency on the ground. Third, the reports that are published by the deployment can expose, and have exposed, the victims’ and the human rights defenders’ whereabouts, activities and networks, endangering them in situations of repression and violence (Hankey and Clunaigh 2013; Slater 2016).

Another rule of the public sphere is transparency. But according to Dean, nowadays power is not hidden, so transparency does not need to be a goal. Echoing the concept of infoxication (Toffler 1984), Dean means here that, in the net, there is not a lack of information, but the opposite: an overload. ‘Fully aware that there is always more information and that this availability is ultimately depoliticizing, neodemocratic politics prioritizes decisiveness’ (ibid.).

But contrary to what Dean says, information is not always available, even when it is most needed such as in the case of disasters or emergencies. Snowden revelations in 2013 put an end to any illusions of transparency on the part of governments too (Castells 2015).

APSs seek transparency when information is hidden, not as a final goal, but as a tool for action. Ushahidi’s first deployment was put in place precisely to break the news blockade by the Kenyan government and media.

However, too much transparency can render victims vulnerable to being targeted. The use of the data infrastructure enables digital humanitarianism, but it also produces new risks for both humanitarian workers and witnesses, since their reports can disclose their location, actions and networks to the dataveillance machinery (Hankey and Clunaigh 2013, 536; Stottlemyre and Stottlemyre 2012, 12). Namely, data and metadata produced in data activism can be harvested by people in power to control and locate activists and witnesses (Burns 2014). That is why Ushahidi deployments do not usually open their data.

Proactive data activism can also offer alternative narratives when the mainstream versions of reality are manipulated. An example is the ‘counter-mapping’ project set up by several communities in Indonesia in resistance to governmental initiatives to remap land and surrender it to companies (Radjawali and Pye 2015). Indigenous counter-mapping challenges the state power over maps, and its categorization of land uses (ibid., 4). This movement is currently testing ‘grassroots UAVs’ (Unmanned Aerial Vehicles) to generate data, and strengthen people’s datasets and political actions (Meier 2015b). Again, the idea in data activism is not to produce transparency for transparency’s sake or to create awareness, but is aimed at action.

Finally, issue networks are ruled by credibility over rationality; in contrast, rationality governs the public sphere (Dean 2003). The APSs look more like neodemocratic networks since credibility and trust are vital to sustaining the crowd and the activist effort (Gutiérrez 2018a).

The APSs are not ideal havens of cooperation and equality: power asymmetries exist within actors, and some glitches and oversights in real-life cases have created ‘dead Ushahidi maps.’ However, the rules of the APSs imply that they have to be guided by: conditioned inclusivity and temporality, within a set of procedures; equal access, so actors can communicate and act; transparency, so unconventional data-based narratives are produced; and credibility, so buy-in and adhesion on the part of the actors involved in tackling a crisis are generated.

Framing the New APSs

Based on the observations made earlier, this section justifies why the sphere of communication and action created by data activism are called here ‘alternative digital public spheres,’ and also provides a broader context for the rise of data activism in the big data society.

The APSs are ‘alternative’ because they engage alternative actors (i.e. non-experts, victims and vulnerable people) in producing counter-narratives, using technologies in unconventional ways (i.e. the data infrastructure employed critically), and sometimes altering power dynamics by reversing the monopoly of maps and the datafication process. Namely, they propose formats, actors and subjects outside the established norms.

They are ‘digital’ because they are enabled and constrained by the data infrastructure, that is, a digital infrastructure that allows the managing, sharing, storage and use of data. That is, they are digital because they are enabled by information, communication and data technologies.

The APSs conform to the rule of ‘publicness’ as a condition of the public sphere, defined by Bohman, although he talks about face-to-face interaction (2006, 134). In the APSs, there is little physical contact, but the ‘conversation’ is public not purely because it could be seen online by others but to the extent that it could be taken to address anyone, paraphrasing Bohman, (ibid.). That is, they are public because they are ‘the result of the social activities made by individuals sharing symbolic representations and common emotions in publicness’ (Mateus 2011, 275).

Data activism happens in the context of other social uses of the data infrastructure. It can be employed in two opposing ways: it can facilitate governmental and corporate intrusion, manipulation and control, but it can also be a powerful tool for interaction, activism and social change.

The unyielding datafication of all human activity has profound impacts. With big data and computational techniques, massive surveillance and manufacturing consent is easier: it has moved from the ‘magnifying glasses and baseball bats’ of the 20th century into ‘telescopes, microscopes and scalpels’ of the 21st century (Tufekci 2012, 6).

To be exact, employed as a scalpel, data infrastructure allows manipulators to engage people on an individual basis. Before data infrastructure, social scientists analyzed relevant samples of the universe they wanted to study and relied on surveys in which people said what they thought they did. With data infrastructure, social scientists now can examine what people do (Stephens-Davidowitz 2017). Data infrastructure enhances and quickens surveillance, makes it personal and renders it ubiquitous and surgical. Surveillance today involves ‘the ability to go beyond what is offered to the unaided senses and minds or what is voluntarily reported’ (Ball, Haggerty, and Lyon 2012, 26).

But the ICTs and, lately, data have transformed activism too. The rise of the data infrastructure and massive surveillance prompted ‘reactive data activism,’ which employs the data infrastructure to shelter online interactions and communities from massive surveillance (Milan and Gutiérrez 2015, 127). With reactive data activism, the infrastructure that makes it possible has become more accessible to the average user, ‘rapidly evolving from a peripheral, elitist form of activism to a diffused one, whereby also non-skilled users take part in the game’ (ibid.).

Data infrastructure allows a zooming in and out of processes and places and can offer both the high-resolution landscape of a situation and a vision of the life of one single person at a given time (i.e. a request for help posted at ‘Ayuda Ecuador’). These capabilities are incorporated into the new APSs.

The realization of the APSs is not free of contradictions. Differing from that which reactive data activists usually do, proactive data activists often resort to capitalist technologies, whose values they do not seem to share. Some of these technologies are based on business models that require the low production costs that only semi-slave conditions and child labor afford (Amnesty International 2016, 29), and create environmental damage (Hogan and Roberts 2015). Proactive data activists typically turn to whatever tool is at hand – including Google Maps, mobile technology and other corporate inventions — to respond to a crisis (Gutiérrez 2018a). Other problematic issues include the power asymmetries and the failures that real cases of data activism accumulate.

Conclusions

The spaces for debate, consensus and networked action that are generated in proactive data activism can be theorized as public spheres, because they share some of their attributes: although proactive data activism’s final aim is action, it is articulated by actors who interact and organize following procedures framed by ‘inclusivity,’ equal access and transparency. However, they do not resemble the bourgeois public sphere entirely. I have called them ‘alternative digital public spheres’ or APSs for short. Table 2 summarizes their qualities, which emerge in comparison with the public sphere and the neodemocracies. 

SITE: Local/global	Global interaction in the space of flows with local effects (local dimension also in on-the-ground action).
GOAL: Action	Ø  Consensus (which embeds an expectation of response) is a tool for action; Ø  crisis mapping’s aim is tackling emergencies and activist mapping’s is to enhance advocacy and empower communities.
VEHICLE: Crises and problems	Ø  APSs can bypass information blockades, generate alternative narratives and counter-maps, and revert datafication; Ø  emergencies can trigger crisis mapping and social problems, activist mapping; Ø  although not the vehicle, actors are important, citizens are placed at the forefront of production, marginalized people can get empowered and some roles are fundamental in crisis mapping.
MEANS: Networked coordination	Ø  APSs behave like decentralized network organizations, where roles are static (i.e. crisis mapping), while concrete individuals are expendable; Ø  a sustained effort over time strains the endeavor, but networks are resilient.
NORMS: Inclusivity, evanescence, equal access, transparency, credibility	Ø  Norms come from the network, not imposed in top-down processes; Ø  APSs emerge with an built-in obsolescence; Ø  they are ruled by conditioned inclusivity; Ø  their equal access (not equality) is framed by asymmetries (in the roles and rules, in the level of danger and because victims can be exposed); Ø  transparency is a tool (like consensus), which can generate disruptions, but too much transparency can be dangerous for victims/reporters; Ø  without credibility, data activist initiatives fail.

Table 2: The alternative digital public spheres. Elaboration by the author.

What follows is an explanation of Table 2. APSs generated by data activism have lost the nation as a reference because they are located both in the local and the global, taking advantage of the space of flows. The action coordinated globally in the APSs has local consequences. In crisis mapping, for instance, the local determines the conditions in which victims can report, and humanitarian workers can use the information on the ground.

Actors in the APSs seek consensus, but this accord embeds an expectation of a response and is used as an instrument for action. The two main varieties of geoactivism, crisis mapping and activist mapping, are targeted at action, although the former’s aim is tackling emergencies (short-termed action,) and the latter is deployed to generate evidence for advocacy and empower communities (medium- and long-termed action). In APSs there is a difference between the local and the national. For example, disasters may affect vulnerable people in remote rural areas more than others living in cities of the same country. Also, catastrophes do not respect national frontiers.

APSs are formed around conflicts and problems they seek to address, and sometimes they bypass information blockades and generate alternative narratives and counter-discourses. They can allow ordinary people to generate their own data on their own terms, reversing the datafication process, which is typically the monopoly of governments and corporations. Emergencies can trigger crisis mapping, which is the geolocation of citizen data to assist humanitarian efforts, while activist mapping can be activated by social causes. Although not the vehicle, actors are essential in APSs: citizens are placed at the forefront of content production; disenfranchised people are at the decision-making table, which empowers them and enhances their agency; and some roles are fundamental in crisis mapping, for example.

APSs are managed by decentralized, networked coordination in which roles are static, while individuals are expendable. Although a sustained effort over time can impose strains on the actors supporting the APSs, it can be overcome by the resilience of the network.

The norms that rule APSs come from the network, that is, they are not imposed in top-down processes by external actors. APSs in data-activist initiatives are fluid and short-lived, embedding a built-in obsolescence because they arise in the face of finite emergencies and problems. Another rule governing APSs is conditioned inclusivity, which allows the participation of anyone, as long as he or she follows the guidelines, for example, to produce verified reports during crises. However, this equal access (not equality) embeds some asymmetries. In the case of crisis mapping, for example, deployers act as gatekeepers of verified reports, victims and humanitarian workers often face dangerous situations, and victims can be exposed by the APS. Transparency, another rule, is employed in the APS as a tool and can generate disruptions when it becomes evident that, for example, news media organizations are not doing their jobs. However, too much transparency can be dangerous for victims reporting data about their situations. Finally, credibility, needed to generate buy-in, ensures the success of the data-activist initiative.

But exactly as is the public sphere, the APSs are a construct. They are offered here as a heuristic concept for understanding how proactive data activism raises networked communities of individuals from a diverse background, sometimes based in remote locations, for coordinated action. APSs, in which actors work in perfect harmony towards results, are never to be found in real life, as actual-life implementations discussed in this paper demonstrate.

What real cases show is that, with enhanced data-based capabilities, activism can unleash processes of real communication conducive to action. APSs also provide another turn of the screw in the theoretical debate about the validity of the public sphere as a concept today.

Introduction

Data are often framed as tools of organized, entrenched power. The etymology of the word “statistics” reveals its origins as an aspect of the operation of the state, and it has traditionally been states, bureaucracies, and large capitalist enterprises that accumulate and manage through data, and especially through digital information-processing technologies. Indeed, since the 1960s, the dominant models of organizational action have been information-processing models (Simon 1997). Data, then, have largely lived within these centers of power and indeed have been thought of largely as consolidating that power.

To some extent, though, recent shifts have challenged this status quo. While most data sets remain located exactly where they were—in centers that have the capacity to produce, analyze, and make use of data—technical developments, such as the burgeoning capacity for information processing in the hands of everyday users, and social developments, such as transparency initiatives and movements advocating for more public access to data and data literacy, have begun to put similar kinds of big data and data processing tools in the hands of individual citizens, grassroots organizations, activist groups, and others engaged in what we might broadly class as “data activism.” This offers the potential for new forms of activism and a reconfiguration of power relations – the opportunity for forms of “sousveillance” to counter the “surveillance” at the heart of data-driven governance, for example.

We ask, what happens when these new opportunities must be enacted in the richer textures of practical organizations, and how do actors and activists resolve the tensions and frictions that shape the contexts of their action? In this paper, we examine the practical realities of everyday data activism in urban governance, with a particular focus on the ways in which people balance the activist ideals and institutional mechanisms of their daily work. In so doing, we seek to unpack notions of “activism” itself and identify within it multiple different moments, functions, and positions individuals may adopt for achieving activist goals and for opening up opportunities for others. Our argument draws on material from an ongoing research investigation into civic data work. For the last ten months, the first author has been conducting ethnographic fieldwork with an official urban data team. This team sits in the office of the mayor of one of the largest cities on the West coast of the United States. Our ongoing study looks at the practical, day-to-day work of the data team, including advocacy work (articulating and communicating the value of open data and extending data capabilities, both internally to departments and externally to the public), organizational work (including formulating, running, and publicizing specific data-driven initiatives, and developing and maintaining collaborative relationships with other parts of city government), and technical work (working with the city’s legacy systems, and cleaning, regularizing, publishing, and maintaining data sets). Through participant observation, it becomes apparent that many of the grand notions surrounding the revolutionary power of big data for both social justice aims, or for the reinvention of the bureaucracy, occur incrementally, collectively, and non-linearly, in the quotidian moments of data work. Here, the ambivalences and contested dynamics of fields of activist action become more prominent.

The data team itself comprises a range of diversely skilled individuals who evince (in casual conversations, presentations, and formal mission statements alike) a strong individual and collective concern with open government and the value of data. They see open access to data as a site for accessible and equitable city services, enhanced quality of life for all citizens, and deeper forms of citizen engagement across the range of city government functions. Some members of the data team consider themselves data activists, or at the very least more on the “radical” side of the civil servant ethos, as stated by the chief data officer in conversation with a social media platform research team in the planning of a forthcoming partnership. The actors leverage a variety of resources—such as county and state, or private, sources of data, as in the case of this collaboration—to achieve these “open” aims even as they work inside city government and within its practical limits.

Drawing on Meyerson and Scully’s notion of “tempered radicals” (Meyerson and Scully 1995), we consider how the staff of municipal data offices, and other informal data workers, enact their own “more radical” agendas and forms of data activism while negotiating the complex dynamics and organizational structures of city governments. This formulation encourages us to adopt a more nuanced approach to questions of power, activism, and resistance, one that focuses on the variety of configurations of “insider” and “outsider” perspectives, the complexities of meshing politics and practical action, the diversity of positions within the landscape of resistance, and the ongoing daily encounter with the ambiguities of positionality. Our ethnographic work, then, provides a starting-point for a broader reconsideration of how politics and activism are enacted in practice.

Data Activism, Organizing, and Ambivalence

Exploring the political consequences of the contemporary spread of data-driven analytics, Milan and van der Velden (2016) describe data activism efforts as supporting “the emergence of novel epistemic cultures within the realm of civil society, making sense of data as a way of knowing the world and turning it into a point of intervention and generation of data countercultures.” They suggest that data activism might even incite new populations of tech activists, stating:

because datafication is such a prominent feature in public life, data activism, as a mode of dealing with it, might progressively appeal to more diverse communities of concerned citizens, beyond the expert niche of previous incarnations of tech activist engagement (Milan and van der Velden 2016, 4).

Milan and van der Velden describe the opportunity for activists to appropriate the technical tools of data analytics typically associated with sites of corporate power and municipal governance, and the opportunities that those tools provide both for holding such entities to account and for offering alternative accounts of matters of concern. Central to the authors’ concern is a notion of resistance, counter-hegemony, or alterity – that is, the idea that the core of any activist position is its opposition to mainstream thought or to traditional power.

However, as writers such as Turner (2006, 2013), amongst others, have noted, stark dividing lines and oppositional stances are not always so clear. Particularly in the domain of digital technology and its development, a complex interplay can be seen between mainstream culture and counter-culture “resistance.” This interplay results in complicated, hybrid contexts, such as former Merry Prankster Stewart Brand operating the camera for Douglas Englebart’s famous demonstration of his military-funded NLS project. Turner traces a complex lineage, reaching back to American anti-fascist movements, in which the distinctions between “conventional” and “counter-cultural” positions are not always easy to discern.

Noting that “politics always takes place in a field criss-crossed by antagonisms,” Chantal Mouffe (2013) has advanced a political philosophy focused on disagreements, local tensions, and radical plurality. Perhaps most important for us here is the intricate texture of the forms of agonistic political relations that she explores, in which, again, no global orientations are easily sustainable in the face of a multi-dimensional and multi-faceted series of local tensions and distinctions of position. In light of Mouffe’s thinking, any activism must always be read locally, as a relational matter, and as a matter of degree, allowing again for a complex internal polyvocality. Sometimes, this polyvocality reflects intersectional concerns and the work involved in engaging in, and maintaining positions of, strategic essentialism in order to achieve particular political ends. Sometimes it speaks purely to the pragmatics of effecting action.

In a recent episode of the Politically Reactive podcast series (Bell and Kondabolu 2017), guest Patrisse Kahn-Cullors—one of the founders of the Black Lives Matter movement, and a decorated social organizer—spoke briefly on her role as an organizer. When one of the hosts queried why she identifies as an organizer instead of an activist, Kahn-Cullors answered that she works to make distinctions between the various critical roles in resistance movements, namely the activist and the organizer. The activist, to paraphrase, is the one who goes out and does the action, one such example being attending a protest. Conversely, the organizer lays the groundwork for there to be a protest in the first place, through planning, organizing, and networking. Kahn-Cullors states, that for her, activists are “super important” as “they sign the petition, they show up to the rally, they show up to the board meeting. But the organizer is the one who develops the petition, they are the ones who get the rally together, the ones that bring communities together so that activists can show up.” Both roles are critical in order for social change to take place.

Within the revolutionary rhetoric of the digital, these pragmatic considerations and organizing efforts often go unrecognized. Tufekci (2017) draws attention to this oversight when discussing the role of digital technologies in supporting and organizing protest movements in sites such as Istanbul’s Gezi Park and Cairo’s Tahrir Square. While celebrating the power of social media to mobilize and connect large numbers of people in the absence of prior large-scale social infrastructure, she cautions that traditional protest movements build infrastructure slowly as they grow, and so are more likely, when decisive moments come, to already have in place the support structures necessary to, say, provide food, clothing, material assistance, and medical care to protesters. In enabling mass assembly without passing through the phase of building a community, movements grounded in social media risk focusing on the “activism” component of Kahn-Cullors’ construction without proper attention to the “organizer” role.

Building on these analyses, we want to take a further step, moving from the organizational requirements of activism towards a consideration of the potential configurations of activism in organizations more broadly. We see “activism” here as a complex set of practices and acts of positionality, encompassing many different kinds of work. Importantly, we aim to underscore a need to look at activism and social change at the scale of the micro negotiations of the everyday, while understanding their affects are neither discrete nor direct. Our field site is one where the data team members find themselves navigating and leveraging the uncertainty in large formal institutional structures undergoing change, working to effect social change, but charged too with the need to acknowledge limits and conditions upon that change. The questions that this paper attempts to open up is just how radicalism is “tempered,” in Meyerson and Scully’s (1995) terms, and how that shapes encounters with new understandings of data, particularly in terms of what new forms of data and data technologies can mean for social justice and urban governance.

The budding forms of data collection and expression that we see in the data team (making data open, automated, and interoperable), provide new ways of knowing, or new knowledge, and along with them alternative methods for challenging hegemonies. We approach activism in this domain as a collective process. In reflecting on our own experiences at the study field site, we ask: how might those with more radical aims incorporate activist modes and ideals into the everyday operations of a large urban bureaucracy? What does it mean to do so at a politically volatile time? We focus on data activism as it is expressed in organizing roles, embodied by the ambivalent data activist. Here, we consider the productive ambivalence in everyday data work for understanding how uncertainty is leveraged to push activist ideals in a bureaucratic setting. The characteristics of this setting, along with the motivations of the individuals who inhabit it, allow for expressions of activist ideals and of “making sense of data as a way of knowing the world and turning it into a point of intervention” (Milan and van der Velden 2016, 1).

Possibilities in Being Data-driven

The broader domain of this inquiry is the increasing turn to data-driven approaches in urban governance. Promising greater transparency, quality of life, and civic participation, these data-driven approaches avow to more effectively and efficiently address complex social problems. While information objects have long been a key resource in the creation of systems, products, and policies that shape the world in which we live, this current movement—what many term a “data revolution”—marks the advent of a new and forceful volume, variety, and velocity of data (Kitchin 2014). Derived from sources such as citywide sensor systems and more pervasive online tracking through a myriad of personal devices, these new data assets engender new forms of data curation and processing, along with a greater anticipation of new possibilities.

Milan and van der Velden characterize the big data movement writ large as evoking “a broad set of socio-technical phenomena enveloped in quasi-mythological narratives that univocally emphasize possibility and magnitude” (Milan and van der Velden 2016, 2). The “datification” of everything, from sleep habits to stock-brokering to municipal sanitation services, emphasizes in particular the possibilities of better, faster, and more accurate decision making; these “better decisions” become the foundation for a more effective problem solving. This is owed in part to the seeming objectivity and representativeness located in big data’s near constant accumulation and administering, as compared to earlier forms of quantification. This data-as-truth account is leveraged by data activists and the civic realm (citizens, civic technologists, and city staff) alike; each hope for data to more objectively expose entrenched injustices and enable equitable futures.

The movement is not without its critics. Many are quick to point out that these new forms of data-driven decision-making may—much like extant forms of social rationalization—end up reinforcing both existing inequities and hegemonic interests (Kitchin 2014). There are certainly concerns that these novel tools will make and remake unjust futures and do so with greater potency. But while many have viewed the burgeoning enthusiasm around the possibilities of big data with skepticism, there exists a greater lot of researchers, corporations, technologists, and other technology users and data workers who encourage and normalize the potentiality of big data. Such stakeholders have lent a legitimacy to this burgeoning movement, as exhibited by the increasing number of formal data positions and identities (such as the data scientist, the data janitor, the chief data officer, the data journalist, official city data teams, and so on); in fact, a current project of this urban data team is to expand the formal position titles and training tracks within the bureaucracy to “modernize the workforce,” encompassing more of such roles and skills.

Context: Everyday Data Work

The data workers in our study sit in the mayor’s office in one of the largest cities in the United States, with a population of well over one million and a city government that employs about 40,000 people. Established in 2013—when many cities around the world were beginning to “open” data— this team is responsible for managing data assets and related technologies within and beyond city hall, for making data visualizations and data dashboarding tools available to departments and citizens, and, most expressly, for motivating the city to be open and interoperable with their data practices for the making of data-driven decisions. This latter charge involves encouraging different departments, initiatives, and teams to work across otherwise siloed operations in order to utilize each other’s data, as well as other sources of public (county) and private data (via data warehousing services and tech companies). The aim here is to optimize data operations and engender a level of transparency whereby departments and citizens alike know how (and what) the city is performing.

The team itself is comprised of a small but diversely skilled set of data enthusiasts who collect, clean, analyze, visualize, package, and promote data assets, techniques, and tools; much of this work occurs behind the face of a given initiative (e.g. recently the team cleaned, visualized, and trained a group of city staffers working on domestic violence awareness and prevention). The professional and educational backgrounds of these individuals run the gamut, spanning from software engineering and business analytics, to public policy and communications; together these skillsets forge the necessary work of accumulating, organizing, and communicating the city’s data. Whether it is a focus on equity, transparency, or environmental sustainability, each team member is invested in data-driven governance as a means to enhance the quality of life for city staff and the citizens they work for, to make the city more “livable” for all.

The data with which they work are generated from a range of government functions, including city services (e.g. tracking speed and distribution of graffiti cleanup or waste management), urban planning (e.g. measuring traffic, tracking affordable housing stock or the quality of paved lane-miles and sidewalks), and economic development (e.g. understanding local business needs, or the landscape of employment opportunities in the growing tech industry). In the best cases, this means that data from numeric tables in static spreadsheets will be moved to a system (via sensing or scraping) that automatically collects and updates standardized data recurrently, whereby the data sources have defined “owners” and all necessary metadata, and in some cases universal IDs (categories that work across all operations in the city).

Opening and activating new streams of data, then—especially all-encompassing, continuously accumulating data—requires a lot of organizing. Organizing toward transparency can be a challenge in a bureaucratic institution for a variety of reasons, not least of which is the fact that making governing processes transparent creates new accountabilities that departments may not yet have the bandwidth to address. While there was indeed an executive mandate to open data and become both more open and more data driven, there is little direction on how this should work in practice, or how to define success. Thus, there is much uncertainty surrounding the open data movement at the sites of actual instantiation, especially with regard to authority, accountability, and even the very definition of data work.

Uncertainty, Bureaucracy, and a Tempered Approach

A bureaucratic setting may not be the first that comes to mind when looking for examples of data (or any form of) activism or radical thinking. However, through leveraging contingency and ambiguity in day-to-day workflows, and by embodying the position of the “tempered radical” within those areas whereby a given bureaucratic staff member maintains both an authentic activist self as well as a professional position self, opportunities to exercise activist aims arise.

In The Bureaucratic Phenomenon, Michael Crozier (1964) outlines a theory of power and uncertainty, particularly the important role of the latter in enabling agency as well as stabilizing conflict (if one is to bend rules or exploit uncertainty, it is important that organizational homeostasis remains a possibility). Crozier offers a broad study of a French monopoly where automation is not able to fully rationalize work or action, and he characterizes the ways in which there remain undefinable, human-dependent actions within an organizational system. These un-rationalized moments create a wellspring of wiggle room against authority.

Crozier finds that “even those with the least amount of status in an organization” will push “to the extreme their initial advantage—their control over the last source of uncertainty remaining in a completely routinized organizational system” (Crozier 1964, 154). The reason uncertainty becomes important is because it removes predictability. Where behavior is determined, it will be obvious and anticipated. With full rationalization, there would be little need for interpersonal negotiation, leaving no room for bargaining or ingratiating, no room to expand or push upon the boundaries of roles or duties. Crozier argues that in a completely determined and predictable bureaucracy, power relationships could not develop because staff would exist in a context where no one person’s actions would have any potential to impact the behavior of another’s.

In being a sanctioned office of the mayor, this data team certainly owns some leverage for making data demands of other departments. However, there are many ideological and infrastructural constraints to fostering engagement and transparency through open data. Due in part to this being both a newer office, concept, and indeed a new organizational function, there is considerable uncertainty as to what these data projects produce, or how they should be structured. The data team attempts to take advantage of this lack of determinacy in defining their roles in relation to data as a means of procuring it from less than willing departments, (those who, perhaps, do not see the value in increasing their workload to clean and “open” their data, or do not have the capacity to do so, nor the capacity to address what the data might reveal). However, these departmental operations are already rather opaque, and departments are likewise able to contest or expand upon the areas of uncertainty in turn, offering up reasons of bandwidth, capability, political, or infrastructural complications that prevent them from submitting to the data team’s requests. While many leverage this uncertainty against open data advocacy (in the very form of resistance against authority Crozier hints at), ambiguity can be operationalized in favor of activist-leaning goals with the new tools of big data.

Although full rationalization and prediction is often the goal of big data, it is still a novel and unfolding practice, especially in these civic spaces, and the data team can and does define the meaning and structure of this data work, iteratively and collectively, in a variety of creative manners, to expand their own domain of authority. This may translate into a series of more informal programs, meetings, and trainings to build mutual understanding in micro-moments and discussions over time, or it may mean sacrificing the visibility of their own work in order to build bridges and partnerships on matters that they care about. It may also mean leveraging external networks such as the neighborhood, academic, and civic tech communities—in the form of events, public talks, trainings, or dedicated projects. In leveraging others’ time, motivation, and expertise, the team is able to push projects forward that may otherwise raise hackles inside the bureaucracy, or simply become stalled due to lack of proper economic or personnel capacity.

In contrast, the overarching structure of bureaucratic rules and hierarchy create just enough unity to avoid conflict; organizations are buttressed by the fact that employees need to coexist, which often includes a mutual dependency, and a mutual desire for efficiency, ease, or efficacy in getting the work done. In the context of urban governance, the mayor’s influence, in tandem with an overall desire for recognition, may become a structuring order, but as each department holds domain over their given expertise, or in this example, their own information and data, departments and individuals within them can exploit the unpredictable variables of their territory to strike a balance.

Crozier also shows the tendency, with the advent of new technologies, to move toward “technical controls instead of human leadership.” This desire to displace the human is rhetorically framed as a move away from the “affective” and mistake-ridden to the “rational” action of machines, and such a logic can be found again in the big data or the data-driven governance paradigm. In this movement, there is perhaps an aim to rationalize even more invasively, and more rapidly. This is in some ways the entire premise of data-driven governance: a move not only to rationalize but also to automate (and predict) decision-making so that action is based on information unsullied by human affect. How can these same rationalizing techniques instead be leveraged for a more positive and inclusive form of social change?

Exploiting Uncertainty: Data Activism as Tempered Radicalism

Milan and van der Velden (2016, 7) note that “[m]any contentious actions in data activism are performed at the individual level: think, for example, of engaging in programming or inserting data into a spreadsheet.” Bureaucratic employees similarly—especially in zones of uncertainty—have the agency to exploit ambiguity in their discrete, daily work practices and “perform activism at the individual level” even in the rote aspects of data work. The concept of the “tempered radical” becomes important here, which is a theoretical and feminist position of productive ambivalence (Meyerson and Scully 1995).

Tempered radicals are individuals, often located in a professional organization, “who identify with and are committed to their organizations, and are also committed to a cause, community, or ideology that is fundamentally different from, and possibly at odds with the dominant culture of their organization” (Meyerson and Scully 1995, 585). Frequently, these are people whose work focuses on aspects of organizational change or evolution. While this mode of boundary-pushing and change is situated in the domain of organization science, their concept is in conversation with both feminist and activist inquiry. Here the authors seek to expand the frame of “being radical” most commonly featured in literature, and move toward a “collaborative division of labor among activists” which will help to buffer against what they term “the counterproductive tendency […] to judge who is being the best and most true advocate for change.”

Adhering to radical ideals, while working professionally in a bureaucracy of slow and opaque data practices, is one such example. Importantly, the tempered radical is a position of purposeful ambivalence, and it is a challenging role to occupy as it requires holding multiple intensities. The authors of the concept state that:

individuals can remain ambivalent and quite clear about their attachments and identities. In contrast to compromise, ambivalence involves pure expression of both sides of a dualism; compromise seeks a middle ground which may lose the flavor of both sides (Meyerson and Scully 1995, 588).

Far from the typically negative connotations of the word ambivalence, Meyerson and Scully show both the challenges and the constructive nature of maintaining a dualism. Simply by existing in an organization while not fitting into the organization’s overarching, hegemonic structures gives rise to owning ambivalence; this certainly holds true for the data activist enlisted in bureaucratic processes, but may also characterize anyone working through or with data who maintains skepticism about its promises. Alongside the ambivalence exists a complimentary embracing of uncertainty, or an embracing of the inability to completely control. While this may fly in the face of what using data-driven practices hope to achieve, the pragmatic response to precariousness can be used productively in the spaces of big data administering.

Strategies of the Tempered Radical

Meyerson and Scully describe four important strategies of effecting, affecting, and surviving via the role of tempered radical. The first strategy employs Karl Weick’s concept of small wins whereby the tempered radical can undertake smaller or trial projects to encourage larger moments of change, and test or expand existing and entrenched cultural boundaries (Weick 1984). Local boundary-pushing schemes can build, creating future opportunities and ultimately turning into larger projects over time. As does the human body in reaction to something perceived as invasive, so too does the organization when faced with dangerously different and potentially destabilizing ideas. A small wins approach helps to avoid any such reaction while still effecting real change.

The second strategy may seem reductive, but it is that of essentially being true to oneself, or what the authors call authentic action. Specifically, they refer to being oneself continuously and spontaneously. Being true to oneself when one does not neatly fit into predetermined parameters of organizational culture creates opportunities, through exposure, for new ways of doing things by providing examples of alternative modes. The third strategy is that of language styles, or speaking or being fluent in multiple languages (i.e. the languages of each facet of the ambivalent self, or the language of each space the tempered radical inhabits). The important work of this language strategy involves communicating a collage of these languages (i.e. knitting the radical or activist language together with that of the urban bureaucrat so as to be legible to both domains) instead of employing only one language across domains, or only the local language. Strategically leveraging language styles allows for discourse deconstruction, where the tempered radical can dismantle hegemonic ideas and assumptions in either domain to “make room for alternative conceptions of organizing and management” (Meyerson and Scully 1995, 597). Lastly, owing to the cumbersome nature of juggling these dualities, a final and important tempered radical approach involves actively garnering support. The authors cite an affiliations strategy—particularly maintaining affiliations outside of the organization for emotional, ideological, and informational support—as a means to express rather than suppress emotions and ideas, and avoid burnout.

With this in mind, then, we explore productive ambivalence in everyday data work to understand how uncertainty is leveraged to push activist ideals in a bureaucratic setting. We ask: how might civil servants leverage zones of uncertainty in daily data work to push personal or “radical” agendas using these strategies? How do orientations towards “data” in particular allow them to present a view on new organizational opportunities within “legacy” organizational frameworks? Our ethnographic work begins to provide us with a view of how these concerns figure in the work of the data team.

Civic Data Organizing

The constellation of the aforementioned big data enthusiasms—alongside streams of private sector support, civic-minded investment, and a growing political interest in the administrating potentiality of big data—has engendered formal data teams in urban governance across the globe. On taking office, the mayor of the city of our study sought to lead the way for these global smart cities to certify that the city uses data to solve problems and promote innovation, as stated in the open data mandate. The city thus institutionalized data work in an official mayoral data team (a common configuration for these official data teams), which has been the site of our inquiry.

Despite the data team’s institutionalization, the onus is on the scrappy data team to define what data-driven governance looks like by encouraging personnel in the more than forty departments to switch over from legacy systems and make their work “transparent.” Many share the excitement of potential and possibility through data, but enthusiasm is dampened by tensions surrounding the previously opaque becoming open to scrutiny; once a performance issue is identified, it must be addressed lest it be subject to media attention or citizens’ demands. What “tempered radical” strategies might be enacted by the data team staff members to encourage this leap of faith into data-driven governance? How can they provide public access to the city’s data assets and the more activist ideals of civic participation, justice, and equity?

Small Wins

As Crozier explains, being in a position where one does not have full authority to direct occupational efforts engenders a situation where “each group fights to preserve and enlarge the area upon which it has some discretion” (Crozier 1964, 156). This is certainly true of the tempered radical caught in the crossfire of opposing definitions of data work and civic responsibility. Where the data team makes attempts to define and fortify their open data program throughout the city, other departments are likewise doing the same (i.e. holding onto their old and opaque ways, or even coopting and claiming new forms of data work as their dominion). Nevertheless, the streamlining and opening of city data is precisely the type of organizing work that will allow for other forms of data activism to take place: data resources can be used to advocate for various community needs where, for instance, one is able to see the landscape and distribution of city services. Of course, revealing that one neighborhood is not receiving timely sanitation or security services, for instance, is the sort of visibility that some city departments may be keen to avoid.

Meyerson and Scully (1995) show how the tempered radical strategy of obtaining small wins (instead of, say, a top-down overhaul of civic data work) can be used as trials to “test the boundaries of an organization’s capacity for change.” The city data team members employ such incremental moves to “test the boundaries” of a given department’s ability and willingness to partner in the open data paradigm. A recent instance of this boundary-testing at our field site involved the need to release (via a mandate and pressure from citizen and media data requests), and the desire to automate (so as to avoid manually uploading data every month), the city’s safety data. One data staff member in particular, whom we refer to here as “Dina” (a pseudonym), dedicated herself to procuring this data so as to allow it to, in her words, “speak” about crime patterns across the city with the hope of fostering both discourse and action. Dina met with the data-owners—the city’s police department data team—in person in a series of more casual office visits, coffees, and lunches to explain aims and intentions, and to learn what she could do to facilitate the process. She offered up extra time to jump-start the project with hands-on training, and encouraged regular problem-solving check-ins throughout (a process which later became institutionalized across the city).

The experimental nature of small wins helps to “uncover resources, information, allies, sources of resistance, and additional opportunities for change” (Meyerson and Scully 1995, 595). Rather than insisting the data be fully and immediately released per the top-down mandate, Dina labored continuously to ease concerns, educate, and enhance overall capacities, eradicating at first the infrastructural barriers and later the ideological ones. In the course of these nudging efforts, she also came to learn that prior resistance was enacted for a constellation of less-obvious reasons, namely, legitimate concerns surrounding citizen privacy, technical and personnel bandwidth, and the complicated nature in which data are gathered and stored in the field. This process, then, garnered unlikely allies within the police department itself, data workers who align with her ideals despite the appearance of opposition; these new allies in turn are able to leverage their domain expertise to create space for opening data within broader systems of actual departmental resistance.

Weick suggests that “small wins reduce large problems to a manageable size. Big, unwieldy problems produce anxiety, which limits people’s capacities to think and act creatively” (Meyerson and Scully 1995, 595). Through persistent, smaller projects of familiarization, narrativization, accommodation, and automation, Dina was able to secure the release of the contentious data (contentious as the data points to marginalized neighborhoods with fixed or increasing crime while other neighborhoods are experiencing a decline), paving the way for other highly-desired but potentially problematic datasets to follow suit. This day-to-day organizing work is certainly not part of the city’s open mandate, nor is it the way many other data workers have interpreted what it means to be data-driven; this labor often is invisible and goes unrewarded. However, small wins such as these are a crucial foundation for other efforts of data journalism, data activism, and even internal advocacy, that aims to use the city’s data for political action. For example, owing in particular to the scale and granularity of new forms of data, advocates can discern whether the city is indeed becoming safer for all.

Authentic Action and Language Styles

Dina achieved a cultural shift by “acting in a way that was simply authentic” (Meyerson and Scully 1995, 596). Her earnest interest, in both the release of safety data and in being supportive to those who owned the data, demonstrated an alternative interpretation of the mayor’s open data directive (collective vs. authoritative), and city work writ large which tends to be guarded and competitive. Rolled into the small wins strategy are two other tempered radical approaches: authentic action and language styles. In the conversations regarding safety data, Dina was able to leverage the language of the mandate to open data, and also define the data work as a service and as mutually useful, by couching it in both bureaucratic and progressive frameworks. Her authenticity was perceived and appreciated by others, and even allowed her to forge new partnerships for the open data efforts to come.

However, Dina later stated that she wishes that all of these laborious negotiations (e.g. constant follow-ups, couching transparency in terms of automated workflow efficiency, or framing what the data reveals in a more palatable light) were not necessary; she wishes the data could instead “speak for itself.” This individual considers herself a data activist. She works in the data team less than full time so as to perform other forms of data work in her spare time to achieve activist pursuits. When engaging with the various audiences, she is able to stitch together bureaucratic languages with the vernacular of activists, data scientists, and tech evangelists. She finds that holding these multiple identities and maintaining ambivalence is useful, that “there’s value being both an insider and an outsider.”

Currently, she believes that a lot of the more legible data activism is occurring outside of official offices, but she finds the organizing work done within the data team to be very important for achieving activist ends. When asked what work, if any, is done to ensure external data activists are able to leverage the work done inside city hall, she stated:

“You don’t have to do that much, all you need to do is to put the data out there. Like a lot of people say it [the work that needs to be done to promote social change] is not just open data, like you can’t just do that on its own, but what I’m finding is that you can, as long as you have a data dictionary and that there’s enough data where it’s interesting, people immediately do something with it.”

Multiple Languages and External Affiliations

Meyerson and Scully discuss the importance of external affiliations for tempered radicals (with “people who represent both sides of their identity”), not only to strategize and accumulate useful information, but also for maintaining the ability to remain “fluent in multiple languages.” In another recent instance, the data team’s membership fluency allowed them to take special advantage of an event in the informal civic data-hacking sphere to affect change inside the institution. The advent of new information-processing tools, and broader data literacy in everyday users, has engendered new data cultures such as those located in the volunteer-based civic tech community. Civic data hacking, one such example, “can be framed as a form of data activism and advocacy: requesting, digesting, contributing to, modeling, and contesting data” (Shrock 2016, 4). Many cities utilize such civic tech groups and efforts to enhance the function of their city, or to address social concerns and injustices more directly. In fact, once off the clock, the city employees themselves may also participate in such endeavors to build bridges or exercise their own interests.

One such informal civic data group recently organized around a controversial topic—that of racial discrimination in police procedures—using the “Veil of Darkness” hypothesis. The Veil of Darkness hypothesis “asserts that police are less likely to know the race of a motorist before making a [traffic violation] stop after dark than they are during daylight” (Grogger 2006, 878). Thus, if there is indeed a pattern of racial discrimination, the pattern of traffic stops will differ during the daytime from those conducted at night, where the race of the driver is (supposedly) less readily visible. On the event’s online description, however, the organizer remarked that this more contentious open dataset was recently made inaccessible (password-protected), and those attending the event would now need to collect the data from the organizer who had already downloaded it. This otherwise unexceptional caveat within the event description quickly traveled to city hall.

Employing “multiple languages” (activist language surrounding concerns about racial profiling, together with the language of data science and inclusive of bureaucratic undertones), the team was able “to speak to multiple constituencies” (Meyerson and Scully 1995, 590) about this issue, and leverage the novelty of the situation for their benefit. In the context of this city’s open data efforts, all departments retain control of the datasets they share. That the local civic tech community (and, thus, citizens) noticed and publicized the locking of a particular department’s data became cause for general concern, but there was much uncertainty with regard to what the occasion meant or how it should be dealt with. The data team converted this event into a lever for imploring transparency objectives. They leveraged uncertainty here by highlighting this open data breach to remind stakeholders that these assets (and especially their withdrawal) have the capacity to garner positive or negative citizen and media attention. Thus, the event became an opportunity to apply pressure on appropriate parties to (re)release data, not only in the vein of the open data mandate, but as a means to serve their constituents’ needs and interests. In this sense, the data team was able to use the more familiar language of civil service to encourage the opening of data.

Leveraging external networks like the civic tech community as such exemplifies the affiliations strategy, where those holding dual modes garner the necessary support (emotional, professional, and otherwise) from like-minded peers to buttress against “difficult emotions” and the possibility of burnout (Meyerson and Scully 1995). Members of the data team connect with formal and informal networks that seek to use data for social good to extend their bandwidth, but also to attain support within and outside of the walls of city hall. In a recent interview, Dina reflected on leveraging such a network: civic volunteers (often data scientists or technologists) in the city. She remarked that there appeared to be a boost in interest from citizens (especially those in the civic tech domain), stating that there are: “just a lot of people looking to help the city” via socially-minded data projects. She credited the increased interest to the current political climate; many see this liberal sanctuary city as “a pocket of hope in the nation.”

Local concerns, then, play out against a broader backdrop. The current US presidential administration’s value system is markedly different to predecessors, with differing data policies and demands, notably with regard to immigration and citizenship. The support and enthusiasm shown by citizens and other city data teams helps the team remember why their values are important, providing emotional, technical, and ideological support to buffer against the fatigue incurred from the lack of internal support or recognition, especially at a time of nation-wide consternation.

Tempered Data Organizers

Organizing in the era of big data—and especially big and open data—allows new forms of activism to be layered on top of such efforts. There is a copious amount of work that goes into standardizing and contextualizing data (with meta data, data dictionaries, attributed ownership available for questioning datasets, and so on), as there is with “making it interesting.” Members of the data team believe that this work of making data relevant is their most challenging charge, as it involves procuring vast amounts of interoperable data across city functions and over years before anyone will be able to derive useful meanings from these data resources. In desiring the data to “speak for itself,” the data team is referring to its opposite: the invisible and tedious labor entailed in training, standardization, and meaning-making with data. Tempering data projects with useful interpretations, quality control, and capacity building, allows the data to travel with more ease, making it more accessible and less likely to engender resistance.

Milan and van der Velden discuss how data activism is “polyfunctional” because “it can be read through diverse disciplinary lenses, and can be domesticated to investigate different dynamics and relations, between and within people, information, technology, and the state/industry complex.” Allowing the data to speak for itself does not mean that the data team believes data are wholly objective. Instead, they believe the data are tools to reveal patterns which might help various city workers, civic technologists, non-profits, and data activists speak to matters of concern which might not otherwise be legible.

Although exhausting, organizing from the inside allows for invaluable insight into how and why decisions are made; this has the effect of productively humanizing the opposition, allowing space to introduce activist ideals that may indeed only be impeded by structural ones. By leveraging insider knowledge and strategically pushing zones of uncertainty to engender change, along with being their authentic selves, using variegated language styles, and leveraging external networks, tempered data activists can normalize the exposure and scrutiny involved in releasing massive streams of data to the public, push their own social justice ideals into every day open data work, and allow others outside of centers of power to advocate using these resources.

Conclusion

Open data initiatives in urban governance have been advocated by many in and outside of city government. Values embedded in these programs attract those with more radical ideals about transparency, technology, and civic participation into bureaucracies. However, there are practical realities, and even contradictions, in carrying out such an agenda in the vein of more traditional forms of civic work. We have turned to Crozier’s discussion of the productive uses of uncertainty, and to Meyerson and Scully’s concept of tempered radicalism, to find conceptual resources to account for the practical work of a municipal government data team involved not just in making data work for the city and its citizens, but also charged with figuring out what that might even mean.

Of course, it is easy, particularly for those more actively and visibly radical, to criticize a tempered approach for not going far enough, for not doing all that is possible with data assets, or to accuse those within positions of power of not advocating social change. The authors of the tempered radical concept also take up these concerns, explaining that while judgements of conservativism and hypocrisy abound, the ambivalent position of the tempered radical can play a role in a larger movement outside of themselves and outside of their organizations. They conclude that “the labor of resistance may be divided among those who push for change from the inside, from the outside, and from the margin, each effort being essential to the others and to an overall movement of change” (Meyerson and Scully 1995, 598). Perhaps the world of the tempered data activist in a bureaucratic context can only go so far as that of the data organizer, working tirelessly to systematize institutional data practices to make as much data as accessible as possible, and to normalize ideas of data transparency (whatever that may mean). This role or duty aims to produce a framework and resources for others who own more radical or visibly activist positions, or more fully dedicate their time to activist causes (such as organizations that explicitly work to do so, or data and civic technologists who have more resources at hand in conjunction with less entrenched hierarchies to answer to). Thus, in the broader context of the work that data (and non-data) activists must perform, the work of a tempered radical in this domain, or an activist-as-organizer, is only one, though important, node in the larger data activism assemblage.

The tempered data activist will face additional skepticism—coming from outside the big data movement—as to how well data capture any meaningful truths or experiences of urban citizens, particularly the experiences of those most in need of services and social change. As we noted at the outset, big data itself can only go so far, and it is important to consider what conversations spoken through the language of data might restrict or preclude. In considering tempered radicalism, Meyerson and Scully deliberate on the role of language and its tendency to circumscribe. The authors reveal in one interview that “the power of language was located not in the ability to communicate technically, but rather in its capacity to rule out other forms of talk, thought, and identity.” What possibilities do the current expressions of data assets, technology, and the still unfolding practices of data activism rule out?

Tempered radicals have the advantage of being “outsiders within,” and although this position entails a lot of unaccounted for and uncredited labor, of both a technical and emotional nature, they are also positions of power (fringe identities operating in more traditional power structures are less predictable, and as Crozier shows, there is power in unpredictability). Although she speaks to a more challenging and entrenched position on the margin, bell hooks holds that ambivalent identities offer “the possibility of a radical perspective from which to see and create, to imagine alternative new worlds” (Meyerson and Scully 1995, 589; hooks 1984). The very nature of holding such an identity allows for a vantage point not afforded those well within the margins, though maintaining this ambivalent post certainly has its challenges. What practices—located in everyday negotiations—might foster appreciation and collaboration in and outside of the margin? How can we encourage and better appreciate those embodying these exhausting roles to stick with the discomfort of ambivalence? Given the uncertainty surrounding the possibilities of new data technologies and their application, and their emphasis on vastness and speed, it is crucial to examine the more micro moments of potential as a means to indeed “imagine alternative new worlds,” and to do so incrementally, and collectively.