Open government and open data are often presented as the Asterix and Obelix of modern government—one cannot discuss one, without involving the other. Modern government, in this narrative, should open itself up, be more transparent, and allow the governed to have a say in their governance. The usage of technologies, and especially the communication of governmental data, is then thought to be one of the crucial instruments helping governments achieving these goals. Much open government data research, hence, focuses on the publication of open government data, their reuse, and re-users. Recent research trends, by contrast, divert from this focus on data and emphasize the importance of studying open government data in practice, in interaction with practitioners, while simultaneously paying attention to their political character. This commentary looks more closely at the implications of emphasizing the practical and political dimensions of open government data. It argues that researchers should explicate how and in what way open government data policies present solutions to what kind of problems. Such explications should be based on a detailed empirical analysis of how different actors do or do not do open data. The key question to be continuously asked and answered when studying and implementing open government data is how the solutions openness present latch onto the problem they aim to solve.
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
© The Author(s), 2023. Published by Cambridge University PressThis commentary argues for the importance of detailed empirical analyses of how open government data policies relate to the problems they pertain to solve. The prevalent tendency to see open government data as instruments to improve “democracy” should be seen as an open question of which all its components should be studied empirically, rather than as the starting point for open government policy. The underlying worry motivating this commentary concerns the tendency to use technologies as solutions for problems that do not necessarily exist (to a sufficient degree), are ill-defined, or demand a different and possibly “non-technical” solution.
Open government and open data are often presented as the Asterix and Obelix of modern government —one cannot discuss one, without involving the other. Footnote 1 Modern government, in this narrative, should open itself up, be more transparent, and allow the governed to have a say in their governance. The usage of technologies, and especially the communication of open government data, is then thought to be the crucial instrument helping governments achieving these goals. Open government and open data policies are closely being monitored by a variety of actors, including non-governmental organizations, consultancy firms, governments themselves, and academics. Footnote 2 Due to the strong conceptual linkage forged between open government and open data, an obvious and popular indicator for the success of open government has been the amount and quality of published governmental data. While the strength of this conceptual linkage has been debated, Footnote 3 much open government research still focuses on the publication of open government data, their characteristics, the number of datasets, their reuse, and re-users. Footnote 4 This commentary limits itself to discussing open government data (policies), though does not rule out the applicability of its conclusions to other types of (open) data. Footnote 5
Popular in this regard is the attempt to identify the barriers that prevent open data from being reused (Janssen et al., Reference Janssen, Charalabidis and Zuiderwijk 2012). While Janssen et al. are critical of all sorts of “myths” about the unlimited positive potential of open data, the identification of barriers suggests that it is good to overcome them and hence reinforces the relevance or legitimacy of openness and transparency as ideals—something that cannot always be taken for granted (Tkacz, Reference Tkacz 2012). A different strand of open data research conceptualizes open data as a component of an “open data ecosystem”. Footnote 6 van Loenen et al. ( Reference van Loenen, Vancauwenberghe, Crompvoets, Dalla Corte, van Loenen, Vancauwenberghe and Crompvoets 2018, p. 5) describe such an ecosystem as “a cyclical, sustainable, demand-driven environment oriented around agents that are mutually interdependent in the creation and delivery of value from open data”. This research aims to optimize such ecosystems to in the end satisfy the needs of the data-user. Instead of optimizing data processes and ecosystems, others prioritized studying the whys and hows of specific open data initiatives in practice. Kornberger et al. ( Reference Kornberger, Meyer, Brandtner and Höllerer 2017) conducted interviews with senior managers in the city of Vienna to find out how they translated the ideal of open government (data) into Vienna’s bureaucratic apparatus. Interviews were also used by both Heimstädt and Currie who combined them with observations and document analysis in their studies of local governments’ attempts to do open data policy (Currie, Reference Currie 2016; Heimstädt, Reference Heimstädt 2017). Researchers working in science and technology studies (STS), next, focused on how open datasets and open data standards were construed in practice (Goëta and Davies, Reference Goëta and Davies 2016; Denis and Goëta, Reference Denis and Goëta 2017). Lastly, a group of public administration scholars from Utrecht University, to which I primarily respond in this commentary, studied open data initiative with “context” sensitive, “activity theory,” and “practice”-oriented methodologies. Footnote 7 Ruijer et al. ( Reference Ruijer, Grimmelikhuijsen, van den Berg and Meijer 2018, p. 4), notably, use a “practice lens” that allows them to focus “(…) on what people actually do rather than what they say they do”. This practice lens is accompanied with a STS-inspired take on the relationship between social context and technology. The authors draw from the older social construction of technology (SCOT) approaches toward socio-technical change that try to explain the functioning of technology with help from the perspectives on, and the meanings attributed to, technology by human actors in their social context (Ruijer et al., Reference Ruijer, Grimmelikhuijsen, van den Berg and Meijer 2018, p. 5). Footnote 8 Next to this methodological move toward open data’s “context,” did members of the “Utrecht school” also focus on their “politics” (Ruijer et al., Reference Ruijer, Détienne, Baker, Groff and Meijer 2020), and the democratic ideals undergirding open data policy (Ruijer et al., Reference Ruijer, Grimmelikhuijsen and Meijer 2017; Ruijer and Martinius, Reference Ruijer and Martinius 2017). While they understand “politics” as the organizational responses to certain pressures, they operationalized “democracy” in the form of various interrelated democratic processes (monitorial, deliberative, and participatory ones). I come back to these evaluative criteria and conceptions below.
By introducing the practice, politics, and democratic processes of open data to the public administration and political science dominated scholarly debate, the Utrecht school presented three methodological starting points for a practice and politics-sensitive study of open government data. I will respond to these by asking three sets of questions. First, what does it mean to do context-sensitive and practice-oriented research on open government data, and to what extent does it allow for the distinguishing between technologies (e.g., open datasets), the social contexts where they are “in,” the users with their perspectives, and the outcomes of the interactions between these elements? (Ruijer et al., Reference Ruijer, Grimmelikhuijsen, van den Berg and Meijer 2018, p. 5)? Second, to what extent can the “political” character of open data be reduced to pressure-induced responses, reasons, and incentives to do open data policy? What, in other words, makes this something “political”? Third, what does it mean to evaluate or be critical of open government data practices, and with what kind of moral-political benchmark—Monitorial democracy? Deliberative democracy? Participatory democracy?—is this to be done? By answering these three questions, this commentary presents a methodological contribution to the recent “turn to practice” in especially public administration research on open government data. This is of importance for two interrelated reasons.
First, reflecting on these questions is important to avoid reproducing dominant and potentially problematic beliefs about how problems and solutions in open government (data) policies relate. Daniel Greene argued convincingly that one of the problems taken for granted in US education policy is the so-called “digital divide.” This refers to the idea that if more children would have better access to technology, then they would also be able to significantly improve their socio-economic position (Greene, Reference Greene 2021). The problem of poverty is here framed as a technological problem. This, according to Greene, completely misrepresents what poverty entails and furthermore transports much of the responsibilities of solving the problem of poverty to those being harmed by it themselves. Much open government discourse still assumes a similar problem to which open data aspire to present a solution. The problem is understood to be an informational gap between governments and citizens, for which the making of more data available aims to present a solution. It is an open question, which I will not answer here, to what extent this formulation of the problem open government should solve, is the most appropriate rendering of the problem. Answers to this question, or so I argue, are best approached by reflecting carefully and continuously on the questions outlined above.
Second, the practice-oriented approach presented in this commentary allows researchers to take the politics of open data into account without reducing it to the actors involved (e.g., civil servants, political representatives), the space where it is being practiced (e.g., the city hall), the institutions or organizations burdened with the related policies (e.g., municipalities), or the effects the practices or policies have on individuals, or society. As I will explain in Section 3, the question to be asked about open data’s politics should be how open data change our understandings and practices of politics and democracy. What types of participation, for instance, are brought into being when opening up particular governmental datasets? How does the municipal decision-making process change when decisions are opened as open data? A practice-oriented perspective on the politics of open data thus emphasizes the need for in-depth empirical analyses of how open data transform the relationships between governments and citizens. In Section 4, I argue that a practice-oriented approach also supplies us with a normative argument that explains why it is important to study open government data as practice without presuming too much about their expected benefits or value. This argument draws from a variety of literatures that conceptualize politics as the capacity of collectives or communities to be in charge of their futures (or be ‘free’). When one understands politics in this (abstract) way, it becomes possible to evaluate to extent to which open government data stimulate or help communities to act “politically.” When, for instance, open government data hinder or are not accompanied by the implementation of policies that stimulate citizen participation and contestation, it can be argued that they do not contribute to the political well-being of the citizenry. The empirical aim to study open government data as practice is thus partly justified by the value attributed to collectives’ capacities to act politically and to co-determine the meanings of the open data practices they are situated in.
Recently, various fields and disciplines underwent a “turn to practice”. Footnote 9 Though this might suggest some common trend, practice theorist Davide Nicolini argues it is better to understand these various contributions as having “family resemblances” with their own histories, problems, and presuppositions (Nicolini, Reference Nicolini 2012, p. 9). Without thus pretending to be comprehensive, I want to highlight two characteristics of practice-oriented approaches that clarify in what way a focus on practices could help analyzing open government data: the importance of understanding practices as results of interactions rather than as “input”, and the prioritization of practices over preferences. Footnote 10
First, practice-oriented approaches emphasize that all the seemingly stable and durable entities and objects in the world—from human beings, meanings, technologies, organizations, facts, identities, institutions, objects to values—should be seen as results or products, rather than as starting points of interactions (Nicolini, Reference Nicolini 2012, p. 6). Footnote 11 The processes where such entities, objects, and values are produced, however exactly defined, are the practices the practice-oriented scholar (or “praxiographer”) studies. Consider, for instance, the use and production of a wide variety of documents and data by the city council. A project part of the Dutch open government policy tries to make as much of these documents and data more widely available for the public by building a municipal-transcending open city council data portal. Footnote 12 The production of these documents and data in municipalities, however, does not always allow for easy querying through the local municipal’s (digital) archive, and subsequently, the municipality-transcending open data portal. Illustrative here is the practice of hand-written vote registration (Keulen et al., Reference Keulen, Korthagen and Diederen 2019). The councilor’s secretariats still sometimes count and register the results of voting procedures on paper, by hand. For both secretariat and councilors, this way of registering and making votes is something they are familiar with, and which they might characterize as “efficient,” “fast,” or “easy.” For the practice-oriented open data scholar, this way of registering votes can be understood as a practice in which votes and a set of related norms are produced on a particular moment through complicated interactions between humans, technologies, at and on a specific site or situation and moment. Studying such a practice, in practice, could be an important component of a study of open data policy at a municipality from a practice-oriented approach.
For praxiographers, the study of the production of votes equals the production of order. The analysis of how actors structure and stabilize reality in practice, often through collaborating and interacting with other techno-mediated practices, has priority over the question of whether that order or stabilization is legitimate or not (the “evaluative” question) (Lynch, Reference Lynch, Schatzki, Cetina and von Savigny 2001, p. 140). The praxiographer is interested in the practice of voting, and only less or secondly in whether and from what perspective the voting procedure is (legally) valid. Because the question of how the production of order is being done in practice is prioritized over the question of whether it is legitimate, praxiographers are also more attentive to the fragility of order, and the amount of work required to establish it. Imagine the amount of paper city councils need to be able to make their decisions! Footnote 13 Think as well, in light of researching open government data, about the work required to gather, clean, standardize, publish, and manage a data portal or open dataset. Footnote 14
Second, practice-oriented scholars are critical of the study of individual opinions or perspectives on technologies like open data. Annemarie Mol, for instance, criticizes methodologies that focus on whether things are “true” and from what kind of “perspective” this then is the case. In her work of “empirical philosophy” she followed a disease through a Dutch hospital and tried to not focus on how from different perspectives—patients, doctors—the disease was experienced, analyzed, and known. For her “[t]he driving question is no longer “how to find the truth?” but “how are objects handled in practice?””. And “[w]ith this shift”, she continues, “the philosophy of knowledge acquires an ethnographic interest in knowledge practices” (Mol, Reference Mol 2014, p. 5). The implications of this move from knowledge to objects is the need to study how stuff gets done by different actors in practice, and together, rather than focusing on how different actors see objects and activities differently (Latour, Reference Latour 2005, p. 116; Mol, Reference Mol 2014, pp. 20–21). The (different) meanings attributed to open data by civil servants and citizens are less relevant than the forms the open data take at the various places and situations they can be encountered. Footnote 15 This for the reason that focusing on such perspectives overemphasizes the divides (experienced) between actors, and underestimate that what makes them (having to) stick together—the open data practice they are involved in.
For scholars drawing from Mol, to illustrate, it makes sense to say that there is not one open data portal being used in different ways by different people at different locations. The various and different ways in which the open data portal is used and interpreted can better be understood as illustrative of the co-existence of different and multiple open data portals (or practices) in different situations. The practice-oriented open data scholar is thus interested in how open data practices are brought into being at different places and in different situations while still “hanging together” in some way, and for that reason also favors research methods that allow them to track and follow the activities that make up such practices. Footnote 16 Surveys, accordingly, are replaced or at least accompanied with in-depth interviews and long-term participant observation. Within the context of our example, the challenge would be to locate and study the practices in which the manually registered votes are transformed into the (scanned) documents or machine-readable datasets to be found on the city council’s data portal. Who is doing what where and when? What does it do to the data? What disappears in this transformation, and what is being added, or produced? Whose and what kind of order is produced, and how does it affect other practices in the vicinity? These last questions also point at the second issue raised by the Utrecht school: the politics of open data. What would this mean, from a practice-oriented perspective?
It has become a platitude to argue that numbers, data portals, or quantifications have their politics, and that their usage is best to be understood as a form of political action. Footnote 17 The production of datasets, the choices made about what should be digitized and made transparent or not, and the complex battles fought over the inclusion and exclusion of datasets suggest that these issues are also relevant to the study of public and open data. Footnote 18 It is therefore no surprise that scholars of open government have been trying to explicate the political character of the datasets and policies they studied. Politics, to recall, was defined there as organizational responses to external pressures. In this section, I put forward a way of understanding politics that is less focused on organizations and more in line with the practical understanding of open governmental data presented in Section 2. Understanding the politics of open data practically is important firstly for analytic reasons: it motivates the specification of how open data change what we consider to be “politics” or “democracy” (Section 3). It, secondly, has an evaluative importance which is premised on an understanding of politics as the practical activities undertaken by communities or collectives to foster their freedom (Section 4). I, in other words, shall argue there that this seemingly semantic (or “analytic”) discussion on whether practices of quantification like open data are “political” or not, has normative importance. When a politics of numbers stops being political, freedom decreases, and democracy is in trouble—hence the need to study the politics of open data practices.
But let us start with a basic clarification: politics refers to more than what happens in parliament, in our city councils, and on our screens, where political representatives debate about what kind of policies should be implemented or not, and their underlying reasons, incentives, and ideologies. Though these activities and ideas are an important part of what, for instance, Chantal Mouffe describes as “politics,” by only focusing on what politicians do, we lose sight of how in other places activities take place that have an effect on where we as political collectives head toward (Mouffe, Reference Mouffe 2005). When studying the politics of open government, we thus move beyond the analysis of the state and its formal constitutive elements.
A second way in which one could understand the politics of open data is by making reference to the effects open government data have on society. Footnote 19 Think here, for instance, about the discriminatory character of an algorithmically made decision that is interpreted as reflecting a particular political agenda. Because technologies and objects more generally have effects on society, they are political, the argument then goes. One implication of this argument is the extension of “politics” to practically everywhere (De Vries, Reference De Vries 2007, pp. 800–802; Marres, Reference Marres 2015, pp. 138–144). From citizens, to non-human animals, and to trees and things: everything can become political, or have its politics, if it can be shown that it effects others (Maanen, Reference Maanen 2021). This warrants the worry that that the concept is transformed into a rather meaningless adjective to be inserted everywhere where change occurs (Latour, Reference Latour 2007). If everything is performative, and everything becomes political, how does this then help us studying the politics of open government data?
This question is taken up by Noortje Marres who argues for a more “experimental” approach toward the politics of technologies and objects (Marres, Reference Marres 2013). What Marres argues for is in-depth attention to how objects not merely have an effect on our conventional political practices, but how in practice, objects and the practices in which they are situated (like open data) constitute our conceptions of politics and democracy. The challenge for the researcher is to redescribe and respecify practices of tech governance in such a way that technology is neither one of the factors whose influence on political processes needs to be explained, nor that politics becomes the explanans of technology, but how and in what way technologies and objects co-constitute the “political” or “democratic.” Helpful here is to once return to our Dutch city councils and reflect on the kind of openness produced through the production of open city council data. Footnote 20 As hinted at above, the votes registered by the secretariat and later transformed in either documents or data allow interested users of the city council’s portal to reconstruct a limited aspect of the council’s decision-making process. The opportunity to have some influence on the outcome of this process would by then be made undone. The kind of politics or democracy being “opened” here is thus one focused on decisions made in the past, and as such follows and reflects the conventional municipal decision-making procedure. The kind of participation—one of the other often listed open government values— such type of open government allows for is (at best) reconstructive, monitory, “ocular,” and vigilant in nature (Rosanvallon, Reference Rosanvallon 2008; Green, Reference Green 2011; Dijstelbloem, Reference Dijstelbloem 2016). Regardless of whether my brief redescription of the openness constituted through open city council data convinces, it does show how Marres’ call to redescribe objects into political terms could help us better grasp how politics in all its multiplicity could be done, and undone. Because of this, such attempts to “ontologize” politics have added analytic value over theories that state that everything (all technology! all methods!) is political, or approaches that limit open data’s politics to the types of actors (e.g., organizations) involved.
Within the context of open government data, the task thus becomes to redescribe in detail the ways in which the often-suggested positive contributions to “democracy” and “politics” are realized. Not only should one study how the opening of datasets has effects on the conventional political decision-making process, but also how this activity could change our understandings of democracy, politics, and participation. Exemplary here is the argument made by Mikkel Flyverbom that transparency should be seen as a “prism” rather than a “window” (Flyverbom, Reference Flyverbom 2019). Here, this redescription of transparency draws attention to the idea that promoting transparency amounts to a form of “visibility management” rather than a form of unmediated communication. Transparency both shows and veils, and what is shown does not necessarily denote a process taking place somewhere in a municipality or parliament.
But why, in the very end, should we care about this act of redescription of practices into political lingo? While Marres and others argue for the importance of paying attention to politics and ontology while studying technology and objects, they seem to take for granted that this is a (morally? politically?) important activity, while simultaneously also try not to put forward too much normative judgment. To put that differently: where does the normativity come in when arguing for the importance of analyzing and scrutinizing open data politics in practice, especially from a practice-oriented approach that primarily understands norms and values as the results of the to be studied practices? And why and how does it matter that some of these practices are to be redescribed into political, and others into other-than-political terms?
While in the previous section “politics” was approached empirically through asking how open data constitute key political and democratic categories in practice, this section explicates an implicit normative commitment undergirding such a practice-oriented take on politics: a conception of politics as a collective activity aimed at freedom. I suggest that this rough conception of politics helps analyzing and evaluating the politics of open government data, explains what is at stake when doing research on open government data, and that it can be distilled from a common but unspecified characteristic in much STS and praxiographic literature.
Many scholars working in these fields emphasize the value of forms of indeterminacy, complexity, ambiguity, multiplicity for one’s research practices. Footnote 21 The praxiographer of open data should try to avoid reducing our realities and futures to hands full of explanatory factors, structures, agents, or values. A careful scrutiny of how data practices change institutions and societies is to be preferred over the formulation of substantive moral-political judgments. While this tendency to not judge too quickly is something to be appreciated and is also present in disciplines such as political theory and ethnography, it does not really explain why we need to be attentive to various forms of (de)politizations, and why values like ambiguity and indeterminacy are important—methodological openness and flexibility are not self-explanatory. Though rarely explicated as such, I read these arguments politically, in the sense that politics has to do with whether and how communities have the capacity to transform themselves, or be free. Politics, then, is reconceptualized as a practice with a specific teleology (freedom).
Such an abstract rendering of politics as practical activity bears resonance with scholars such as Tully and Dean who draw from Foucault (Dean, Reference Dean 2014; Tully, Reference Tully 2008), philosophers like Gerard de Vries who make reference to Aristotle (De Vries, 2004), political theorists drawing from Arendt (Cavarero, Reference Cavarero, White and Moon 2004; Krause, Reference Krause 2015), but also political realists like Bernard Williams who define politics as the absence of tyranny (Williams, Reference Williams 2005). It, lastly, resonates with arguments made by critical tech/data scholars on how tech companies “preempt” our capacity to act through the prediction and optimization of our human conditions (Amoore, Reference Amoore 2020; Delacroix and Veale, Reference Delacroix, Veale, O’Hara and Hildebrandt 2021; Powell, Reference Powell 2021).
Emphasizing the open-ended, dynamic and continuous character of politics also points into the direction of situations where this is absent, and the “fields of possibilities” (Tully, Reference Tully 2008, p. 125) have disappeared. For Foucault, such situations have transformed into relations of domination (Foucault, Reference Foucault and Rabinow 1997, p. 283; Dean, Reference Dean 2014, p. 47). Others prefer to characterize these situations as hegemonic or tyrannical (Mouffe, Reference Mouffe 2013), and more Deleuze-inspired writers talk about “bad bondage”. Footnote 22 Such lacks of politics, for the more Foucauldian-inspired writers, equal a lack of power and capacity to act, and in the end, a diminishment of freedom (Tully, Reference Tully 2008, p. 125).
Understanding politics in the way described above should be understood as a proposition or working hypothesis to be tested empirically (Gobo, Reference Gobo 2008, p. 88). The distinction between political practices and practices of domination thus functions initially and merely as an analytic tool to help the researcher grasp the issues of government, power, freedom, and the lack thereof, at play (Dean, Reference Dean 2014, p. 49). Whether or not it can and should function more substantively as an evaluative benchmark is a question to be worked out empirically, in and with practices.
What does this then imply for open data research? Most importantly, it forces researchers to be attentive to how and in what ways practices of data production, processing, and dissemination are conducive to political action. Or, in other words, how and in what way open government data indeed increase the range of possibilities individuals and collectives have to be in charge of their futures. While communicating the results of a decision-making procedure to the inhabitants of a municipality might at first sight seem to be promoting a municipality’s transparency, Footnote 23 , it is to be questioned whether this open data practice really increases the quality of the municipality’s political arena if it is not accompanied with an opportunity to contest these decisions, or an invitation to come up with alternative proposals (or to act “politically”).
This last example shows that it matters how an open data practice and the values it tries to promote is described. The kind of transparency promoted here by the municipality clashes—arguably—with the one held dear by its inhabitants, illustrating the different practices of “transparency” at play. The analysis of such tensions is needed to describe the character of this open data practice, and this analysis combined with our rough grasp of the difference between politics and domination helps researchers to answer the question whether this type of open data policy promotes politics and democracy, and in what way. The challenge here, once more, is the see the values open data promote—e.g., transparency—as practical, i.e., as the results of individuals interacting with technologies in particular situations. The usual starting point of open data government research and monitoring—the extent to which various values are brought about—glances over the difference between a value and practices of valuation (Berthoin Antal et al., Reference Berthoin Antal, Hutter and Stark 2015). Rather than presuming the relevance of such values in practice, should the open data scholar show through her careful redescription of an open data practice, the existence of such valuations, their interactions, tensions, and their location on the continuum between politics and domination.
The key question to be answered when studying and implementing open data is how the solutions openness promise latch onto the problem they aim to solve. This is not something to be taken for granted. Moreover, the discussion of the three issues in this commentary—open data’s practical nature, its politics, its valuation—motivates researchers to explicate how and in what way technologies like open data present such solutions. Such explications should be based on a detailed analysis of how different actors do or do not do open data, which is important because open data and its implicated values should be seen as the results of interactions, rather than their input. To be taken into consideration when redescribing open data in this way is the distinction between politics and domination. How and in what way open data practices are to be situated on this continuum, is a question to be asked and answered continuously.
The author thanks Daan Kolkman and two anonymous reviewers for their comments on drafts of this commentary.
This commentary is based on PhD research funded by the Dutch Research Council (NWO 313-99-330) and forthcoming as From communicating to distributing: studying open government and open data in the Netherlands. The revised version of this commentary was written while being funded by the European Research Council (grant number 716971). Both funders had no role in the writing of this commentary.
The author declares no competing interests exist.
Conceptualization: G.V.M.; Investigation: G.V.M.; Methodology: G.V.M.; Writing—original draft: G.V.M.; Writing—review & editing: G.V.M.
Data availability is not applicable to this article as no new data were created or analyzed in this study.