Seeing Transparency More Clearly

In recent years, transparency has been proposed as the solution to, and the cause of, a remarkable range of public problems. The proliferation of seemingly contradictory claims about transparency becomes less puzzling, this essay argues, when one appreciates that transparency is not, in itself, a coherent normative ideal. Nor does it have a straightforward instrumental relationship to any primary goals of governance. To gain greater purchase on how transparency policies operate, scholars must therefore move beyond abstract assumptions and drill down into the specific legal, institutional, historical, political, and cultural contexts in which these policies are crafted and implemented. The field of transparency studies, in other words, is due for a “sociological turn.” David Pozen is a Professor of Law at Columbia Law School. This essay is a lightly edited and abridged version of a keynote address that he gave in June 2019 at the Sixth Global Conference on Transparency Research in Rio de Janeiro. E-mail: dpozen@law.columbia.edu Electronic copy available at: https://ssrn.com/abstract=3478005 1 Seeing Transparency More Clearly There is a famous scene in the television show The Simpsons in which Homer Simpson raises a beer mug for a drunken toast and proclaims: “To alcohol: the cause of, and solution to, all of life’s problems!” As the jumping-off point for this essay, I propose to adapt Homer Simpson’s maxim. For if the word “alcohol” is replaced in that toast by the word “transparency,” the revised version would encapsulate, to a disconcerting degree, the last decade’s worth of scholarship on open government policy. Transparency, that is, has been identified as the cause of, and solution to, a remarkable range of problems. On the one hand, commentators routinely assert or assume that transparency is indispensable to government accountability, democratic deliberation, citizen empowerment, public-spirited regulation, and public trust in the policy process. Transparency, in this discourse, has been endowed with “quasi-religious significance” (Hood 2006, 3). On the other hand, a growing number of commentators warn that open government regimes have contributed to the decline of deliberation, deal-making, and regulatory capacity in legislative and administrative bodies; to the empowerment not of ordinary citizens but of business lobbyists and commercial enterprises who exploit these regimes for private gain; and to the collapse of confidence in democratic institutions and the resurgence of interest in non-democratic alternatives (Pozen and Schudson 2018). A think tank called the Congressional Research Institute (2019) recently collected over 400 citations from political scientists, legal scholars, and journalists attesting to these “dark sides” of sunlight. In western academic circles, transparency thus finds itself more venerated and Electronic copy available at: https://ssrn.com/abstract=3478005 2 denigrated than ever: still believed by many to be “the sine qua non of good governance” (Michener 2015, 184), yet increasingly suspected of facilitating anti-regulatory or neoliberal agendas and of undermining the very values it is meant to promote. What are we to make of the proliferation of contradictory claims about transparency’s impacts? And how can transparency scholarship progress as a field in the face of such dissensus? In this essay, I will advance two main arguments as the beginning of an answer to these questions. First, as to the seemingly schizophrenic state of the literature, I will suggest that there is nothing incoherent about transparency policies yielding positive outcomes in certain settings and negative or even opposite outcomes in other settings because transparency is not, in itself, a coherent normative ideal. Nor does it have a straightforward instrumental relationship to any primary goals of governance. Second, it follows that to gain greater purchase on how transparency policies operate, scholars must move beyond abstractions and formalisms to drill down into the specific legal, institutional, historical, political, and cultural contexts in which these policies are crafted and implemented. As a shorthand, I will refer to this kind of scholarship as “sociological” and to the direction I am urging the field to take as a “sociological turn”—although as will become clear, I mean to use these terms in a broad sense that embraces the methodologies and insights of numerous disciplines. The Implausibility of Transparency as a Normative Ideal Public administration scholarship focuses on transparency’s role as a regulatory technique, generally in the form of legal directives or structural designs to make institutions more accessible to the outside world. It is uncontroversial to observe of most regulatory techniques—mandates, nudges, taxes, subsidies, incentives, voting rules, debate rules—that they may help certain groups and interests, and they may harm other groups and interests, potentially at the same time. Electronic copy available at: https://ssrn.com/abstract=3478005 3 Everything depends on the details. It is absurd to be pro-incentives or anti-subsidies, say, as a categorical matter. One has to know exactly what is being incentivized or subsidized, and how. It might seem obvious that the same would be true of transparency, except that a great deal of writing on the subject does not treat this as true at all. On the contrary, much of the writing takes its desirability as a given and is unabashedly pro-transparency in a way that, again, would seem odd for incentives, subsidies, or the like. For instance, in response to a recent critical study of transparency, a review quipped: “Is there ambivalence in the value of government transparency? Certainly, if you have something to hide” (Neacsu 2019, 53). Otherwise, the review implied, there is nothing to be ambivalent about; anyone who questions transparency consequently becomes suspicious. In the estimation of this reviewer and countless commentators, transparency is not just a regulatory technique. It is also a fundamental policy goal in its own right, a value to be prized and maximized. This widespread “reification” of transparency, as political theorist Darin Barney describes it, has converted the concept in the minds of many “from a social relationship or process into an object or thing; from a means into an end-in-itself” (2008, 91). On this view, researchers are apt to go astray when they ask about transparency’s impacts on bureaucratic performance or social improvement. The point of transparency policies is to generate transparency. And to the extent that a policy succeeds at generating transparency, that is a good thing, either because transparency is seen as inherently good or because it is assumed to have an inherently positive and proximate association with some other value that is seen as inherently good, most often democracy or accountability. This way of thinking about transparency, seeing it as an end in itself or the functional equivalent, seems to me deeply misguided. To see why, it may be helpful to draw on the philosopher Alasdair MacIntyre’s distinction between primary virtues and secondary virtues. The Electronic copy available at: https://ssrn.com/abstract=3478005 4 primary virtues, in MacIntyre’s formulation, are “directly related to the goals which [people] pursue as the ends of their life” (1967, 24). The secondary virtues “concern the way in which we ... go about our projects” (ibid.)—not what we are trying to achieve but how we try to achieve it. Translated to the public policy context, we might say that values such as social welfare, human and ecological flourishing, minimization of pain and suffering, and freedom from domination are primary virtues. These are the kinds of values that ought to guide the efforts taken by a decent government on behalf of its citizens. They may be articulated any number of ways and contested along any number of dimensions. But they reflect ideals that make people’s lives better in an immediate sense, to the extent they are realized; that are central to the construction of a good society on almost any account of the good; and that the success or failure of public policies can be meaningfully judged against. Transparency, for the most part, is not like that. A public policy that reduces the amount of pain and suffering in a country makes that country a better place, at least in one important respect. A public policy that reduces the amount of opacity in the decision-making of a government agency may or may not make the country a better place. One would have to learn how that increase in transparency is affecting the agency as well as the individuals and entities that interact with the agency. A case might be made for viewing transparency as a specially privileged procedural norm, or even “something suspended between primary objectives and secondary virtues” (Schudson 2015, 23). What is exceedingly difficult to imagine is a theory of political morality or justice that would demand the pursuit of transparency for transparency’s sake. Certain strains of deliberative democratic theory probably come closest here, but I am unaware of any version of deliberativism that makes a primary virtue out of transparency. The overriding objective, rather, is generally Electronic copy available at: https://ssrn.com/abstract=3478005 5 something like reasoned debate, equal communicative freedom, or an open and inclusive environment for collective will-formation (Bächtiger et al. 2018). Although transparency might contribute to good deliberation—might even, in some respects, be essential to it—the goal is not transparency. The goal is good deliberation, however defined. The Ambiguity of Transparency as a Practical Ideal Scholars and advocates, accordingly, risk making a category mistake when they treat transparency as a normative maximand or an end in itself. Instead, transparency must by and large be viewed in instr

T here is a famous scene in the television show The Simpsons in which Homer Simpson raises a beer mug for a drunken toast and proclaims, "To alcohol: the cause of, and solution to, all of life's problems!" As the jumping-off point for this essay, I propose to adapt Homer Simpson's maxim. For if the word "alcohol" is replaced in that toast by the word "transparency," the revised version would encapsulate, to a disconcerting degree, the last decade's worth of scholarship on open government policy.
Transparency, that is, has been identified as the cause of, and solution to, a remarkable range of problems. On the one hand, commentators routinely assert or assume that transparency is indispensable to government accountability, democratic deliberation, citizen empowerment, public-spirited regulation, and public trust in the policy process. Transparency, in this discourse, has been endowed with "quasireligious significance" (Hood 2006, 3). On the other hand, a growing number of commentators warn that open government regimes have contributed to the decline of deliberation, dealmaking, and regulatory capacity in legislative and administrative bodies; to the empowerment not of ordinary citizens but of business lobbyists and commercial enterprises that exploit these regimes for private gain; and to the collapse of confidence in democratic institutions and the resurgence of interest in nondemocratic alternatives (Pozen and Schudson 2018). A think tank called the Congressional Research Institute (2019) has collected more than 500 citations from political scientists, legal scholars, and journalists attesting to these "dark sides" of sunlight. In Western academic circles, transparency thus finds itself more venerated and denigrated than ever: still believed by many to be "the sine qua non of good governance" (Michener 2015, 184), yet increasingly suspected of facilitating antiregulatory or neoliberal agendas and of undermining the very values it is meant to promote.
What are we to make of the proliferation of contradictory claims about transparency's impacts? And how can transparency scholarship progress as a field in the face of such dissensus?
In this essay, I will advance two main arguments as the beginning of an answer to these questions. First, as to the seemingly schizophrenic state of the literature, I will suggest that there is nothing incoherent about transparency policies yielding positive outcomes in certain settings and negative or even opposite outcomes in other settings because transparency is not, in itself, a coherent normative ideal. Nor does it have a straightforward instrumental relationship to any primary goals of governance. Second, it follows that to gain greater purchase on how transparency policies operate, scholars must move beyond abstractions and formalisms to drill down into the specific legal, institutional, historical, political, and cultural contexts in which these policies are crafted and implemented. As a shorthand, I will refer to this kind of scholarship as "sociological" and to the direction I am urging the field to take as a "sociological turn"-although as will become clear, I mean to use these terms in a broad sense that embraces the methodologies and insights of numerous disciplines.

The Implausibility of Transparency as a Normative Ideal
Public administration scholarship focuses on transparency's role as a regulatory technique, generally in the form of legal directives or structural designs to make institutions more accessible to the outside world. It is uncontroversial to observe of most regulatory techniques-mandates, nudges, taxes, subsidies, incentives, voting rules, debate rules-that they may help certain groups and interests, and they may harm other groups and interests, potentially at the same time. Everything depends on the details. It is absurd to be proincentives or anti-subsidies, say, as a categorical matter. One has to know exactly what is being incentivized or subsidized, and how.
It might seem obvious that the same would be true of transparency, except that a great deal of writing on the subject does not treat this as true at all. On the contrary, much of the writing takes its desirability as a given and is unabashedly pro-transparency in a way that, again, would seem odd for incentives, subsidies, or the like. For instance, in response to a recent critical study of transparency, a review quipped, "Is there ambivalence in the value of government transparency? Certainly, if you have something to hide" (Neacsu 2019, 53). Otherwise, the review implied, there is nothing to be ambivalent about; anyone who questions transparency consequently becomes suspicious. In the estimation of this reviewer and countless commentators, transparency is not just a regulatory technique. It is also a fundamental policy goal in its own right, a value to be prized and maximized. This widespread "reification" of transparency, as political theorist Darin Barney describes it, has converted the concept in the minds of many "from a social relationship or process into an object or thing; from a means into an end-in-itself" (Barney 2008, 91).
On this view, researchers are apt to go astray when they ask about transparency's impacts on bureaucratic performance or social improvement. The point of transparency policies is to generate transparency. And to the extent that a policy succeeds at generating transparency, that is a good thing, either because transparency is seen as inherently good or because it is assumed to have an inherently positive and proximate association with some other value that is seen as inherently good, most often democracy or accountability.
This way of thinking about transparency, seeing it as an end in itself or the functional equivalent, seems to me deeply misguided. To see why, it may be helpful to draw on the philosopher Alasdair MacIntyre's distinction between primary virtues and secondary virtues. The primary virtues, in MacIntyre's formulation, are "directly related to the goals which [people] pursue as the ends of their life" (MacIntyre 1967, 24). The secondary virtues "concern the way in which we … go about our projects" (24)-not what we are trying to achieve but how we try to achieve it.
Translated to the public policy context, we might say that values such as social welfare, human and ecological flourishing, minimization of pain and suffering, and freedom from domination are primary virtues. These are the kinds of values that ought to guide the efforts taken by a decent government on behalf of its citizens. They may be articulated any number of ways and contested along any number of dimensions. But they reflect ideals that make people's lives better in an immediate sense, to the extent they are realized; that are central to the construction of a good society on almost any account of the good; and that the success or failure of public policies can be meaningfully judged against.
Transparency, for the most part, is not like that. A public policy that reduces the amount of pain and suffering in a country makes that country a better place, at least in one important respect. A public policy that reduces the amount of opacity in the decision making of a government agency may or may not make the country a better place. One would have to learn how that increase in transparency is affecting the agency as well as the individuals and entities that interact with the agency.
A case might be made for viewing transparency as a specially privileged procedural norm, or even "something suspended between primary objectives and secondary virtues" (Schudson 2015, 23). What is exceedingly difficult to imagine is a theory of political morality or justice that would demand the pursuit of transparency for transparency's sake. Certain strains of deliberative democratic theory probably come closest here, but I am unaware of any version of deliberativism that makes a primary virtue out of transparency. The overriding objective, rather, is generally something like reasoned debate, equal communicative freedom, or an open and inclusive environment for collective will formation (Bächtiger et al. 2018). Although transparency might contribute to good deliberation-might even, in some respects, be essential to it-the goal is not transparency. The goal is good deliberation, however defined.

The Ambiguity of Transparency as a Practical Ideal
Scholars and advocates, accordingly, risk making a category mistake when they treat transparency as a normative maximand or an end in itself. Instead, transparency must by and large be viewed in instrumental terms, as a means to other ends. Nevertheless, an objector might retort at this point, transparency's instrumental relationship to various public values could be so tight that little is lost by conflating the two. Even if transparency is not a coherent normative ideal, perhaps it serves reasonably well as a practical ideal in this sense-a standard of governance worth stressing and striving for, given that it has such a strong affinity with widely agreed-upon goals that may be harder to measure or specify.
The problem is that, in area after area, this is simply false. Transparency's instrumental relationship to most goals of governance turns out to be quite complicated. We know this by now from theory, from empirical observation, and from law.

Theoretical Ambiguity
We know this theoretically, because for every marvelous benefit that might flow from opening up a government initiative or institution, it does not take much creativity to posit a potential cost that undermines or nullifies that benefit. Consider the relationship between transparency and corruption. Legal scholar Michael Gilbert published a paper in 2018 pointing out that just as the state cannot deter corruption without information, bad actors cannot engage in corruption without information-who is buying, who is selling, who can be trusted. By making it easier and cheaper for all parties to obtain this sort of information, transparency has crosscutting effects. It deters some corrupt acts while facilitating others. Regulators, in Gilbert's model, face nothing less than a "tragic tradeoff " (2018, 138).
Or consider transparency and trust. It was a truism for 1960s and 1970s reformers that greater transparency from government would lead to greater trust in government (Pozen 2018). Over the past two decades, however, scholars have advanced a range of theoretical arguments that imply the exact inverse: for instance, because transparency mandates hold complex processes to unrealistic standards; because they foster a "culture of suspicion" (O'Neill 2002, 77) and a spirit "of conflict in state-citizen relations" (Erkkilä 2012, 25); or because they leave institutions unable to "respond to the demands placed on" them (Rosanvallon 2008, 259).
We cannot say with any confidence, a priori, whether these arguments are correct or whether the old 1960s-1970s arguments are correct. Both sets of claims are plausible. About all we can say with confidence, as a theoretical matter, is that transparency's relationship to trust is far from self-evident. The same goes for its relationship to the incidence of corruption, the quality of debate, the influence of powerful lobbies, and the content of policy outputs.

Empirical Ambiguity
A wealth of descriptive research bears out this complexity. The U.S. Congress has attracted particular interest. In 2012, a task force was established by the American Political Science Association to study the breakdown of negotiation in that body. Its final report, published in 2013, looked closely at legislative developments since the 1970s and found that "gridlock in the American Congress has been exacerbated by the 'sunshine laws' that opened up committee deliberation to the public but also to lobbyists and other special interests" (Martin 2013, 127) and that "transparency often imposes direct costs on successful deal making" by tying politicians to partisan scripts and interfering with the good-faith search for compromise solutions (Binder and Lee 2013, 63). At this point, task force members concluded, "the empirical evidence on the deliberative benefits of closed-door interactions seems incontrovertible" (Warren and Mansbridge 2013, 108).
I agree with this assessment, but I would add that the potential deliberative downsides of closed-door interactions seem equally incontrovertible, from a greater risk of groupthink and epistemic closure to a greater risk of reliance on self-serving or illicit rationales (e.g., Lee 2008). The dysfunctional dynamics in the U.S. Congress, moreover, may not generalize to other less polarized settings. My point in highlighting this example is not to suggest that transparency is inimical to political bargaining, nor to deny that certain transparency reforms have had salutary effects; the Environmental Protection Agency's Toxics Release Inventory, to take another American example, has been credited "with playing a central role in driving improvements in pollution performance" since the late 1980s (Karkkainen 2001, 288). My point is that transparency's association with constructive deliberation, like its association with virtually every important governance goal, has proved in practice to be thoroughly "contextual and contingent" (Schauer 2011(Schauer , 1356cf. Meijer 2013).

Legal Ambiguity
Finally, we know that transparency's instrumental relationship to many goals of governance is complicated from the design of freedom of information laws the world over. These laws ubiquitously carve out exemptions for the protection of national security, public safety, personal privacy, commercial secrets, and internal deliberations (Pozen 2017(Pozen , 1106. While the precise contours of these exemptions can be controversial, their existence generally is not. The immanent normative structure of open government law thus reflects that transparency can do damage, can go too far, and must be negotiated carefully if an attractive balance of consequences is to be achieved. When we talk about transparency in governance, we are always talking about something short of full exposure. The landscape of open government law is covered in shadows, as it were, aimed at "translucency" more than true transparency (Samaha 2006, 923).

Implications for Transparency Scholarship
Where does this leave transparency scholarship? The implications are both positive and negative. Certain types of inquiries and methodologies look especially promising, in light of the points just reviewed, while other types of arguments and approaches seem especially unlikely to advance understanding.

Less Transparency Fetishism
Appreciating transparency's implausibility as a normative ideal and its ambiguity as a practical ideal counsels, above all, less romanticism and more realism about the subject. Some amount of romanticism may be useful in political advocacy; constant attention to detail has costs as well as benefits. But scholars and practitioners should be on guard against claims that deny, implicitly or explicitly, the contextual and contingent character of transparency's value. Such denialism can take many forms, among them: • sweeping assertions about what transparency can accomplish on its own; • unstated assumptions that greater transparency is necessarily desirable; • the reflexive attribution of bad motives to officeholders who seek to withhold information or otherwise chafe against the constraints of openness requirements; and • overconfidence in the external validity of experimental studies conducted in contexts far removed from the realm of public governance All of these moves can be seductive. None is well founded. The ubiquity of the first three suggests that, for all of the sophisticated analyses that have been produced, transparency continues to function as a "magic concept" (Alloa 2018) in mainstream academic and political discourse.
Scholars and practitioners may wish to adopt some thinner working premises. For instance, it seems reasonable to presume that more publicly available information is preferable to less publicly available information, all else being equal. It also seems reasonable to presume that certain sorts of transparency about certain sorts of actions-say, information about patterns and practices of state violence-are especially crucial to democratic accountability and especially prone to be withheld without just cause. Some "forms of transparency concerning the basic contours of government action," moreover, "may well be prerequisites to individual and collective self-determination and can be justified without consequentialist assumptions" Electronic copy available at: https://ssrn.com/abstract=3478005 (Pozen 2018, 161). Acknowledging the contingent and contextual nature of transparency's value does not require abandoning moral commitment or accepting that every aspect of transparency's value is contingent and contextual to the same degree.

Less Transparency Formalism
Scholars and practitioners should also be on guard against drawing strong conclusions from the texts of transparency laws. Legal theorists have long known that it is risky to fixate on the "law on books" to the neglect of the "law in action," as the two constantly come apart (Pound 1910). Yet it is especially risky to do so in the open government field, given the heterogeneity of transparency's effects and the notoriously imperfect state of compliance with transparency mandates.
The Right-to-Information (RTI) Rating system, which holds itself out as "the leading global tool for assessing the strength of national legal frameworks for accessing information held by public authorities" (Centre for Law and Democracy 2019), illustrates the concern. RTI Ratings are based solely on the law on books. Hence, the United States is awarded points in the "Sanctions" category because its Freedom of Information Act (FOIA) provides for sanctions against agency personnel who improperly withhold records (Youm and Mendel 2018, 260). In practice, however, these sanctions are rare to nonexistent. One scholar investigating this issue found exactly one case over the course of two decades in which such a sanction was applied (Winters 1996, 618).
Equally stark, the U.S. Congress amended FOIA in 1974 to clarify that courts must review de novo all claims of exemption, including claims of Exemption 1 for classified materials (5 U.S.C. § 552(a)(4) (B)). On the books, this looks stunningly transparent. No judicial deference is to be afforded to the executive branch on national security secrecy. Zero! In practice, however, the executive almost always wins when it asserts Exemption 1. Typically, the court grants summary judgment to the executive without allowing discovery or inspecting any of the withheld records (Pozen 2017(Pozen , 1118. Meaningful victories in national security FOIA challenges, the New York Times' top newsroom lawyer has written, "remain legal unicorns" (McCraw 2016).
So, it is simplistic in the extreme to celebrate FOIA for the vigorous judicial review of national security secrecy that the statute appears to demand. At the same time, it is also too simple to draw conclusions exclusively from rates of wins and losses in court, for in the shadow of judicial review, agencies sometimes declassify requested records before a judge ever issues a ruling (Kreimer 2008(Kreimer , 1055. If the statute books overstate the degree to which transparency is realized in the national security domain, the judicial record understates it. Neither source, moreover, speaks to the pervasive practice of executive branch officials' "leaking" classified information to the media, notwithstanding an expansive set of criminal prohibitions on this behavior (Pozen 2013).
It is impossible, then, to understand what is really going on with government transparency in the United States by parsing the language of the relevant statutes or, for that matter, judicial opinions. One would be utterly lost. The same no doubt holds true for other jurisdictions. Analyses that confine themselves to the law on books may shed a little light on the actual operations and upshots of transparency regimes-but only a little. And they are liable to obscure as much as they reveal. Afghanistan, ranked 177 out of 180 countries in Transparency International's 2017 Corruption Perceptions Index (2018), currently enjoys the top RTI Rating in the world (Centre for Law and Democracy 2018).

More Conditional Thinking
If fetishism and formalism are pitfalls to be avoided in this area, which sorts of approaches are especially well suited to grappling with the "opacity of transparency" (Fenster 2006)?
On the normative side, it seems to me that a great deal of work is still needed regarding the conditions under which certain forms of nontransparency, or partial transparency, are more or less justified. For instance, when exactly is ex post (rather than ex ante or realtime) disclosure sufficient? When are high-level summaries (rather than complete transcripts) sufficient? When is divulgence to an oversight body (rather than the public at large) sufficient?
These sorts of institutionally grounded, middle-range questions have received only modest attention in the theoretical literature. Most "are hardly explored" (Cucciniello, Porumbescu, and Grimmelikhuijsen 2017, 41) in the empirical literature as well.
They are pivotal questions for the next wave of transparency reformers, however. And they are difficult. The answers may vary depending not only on the character of the underlying information but also on the attributes and incentives of the information-holding bodies and the degree to which they are held accountable through other mechanisms. It is not enough to ask whether certain sorts of processes or documents may legitimately be shielded from full and immediate public view. Transparency scholarship must press further to ask which parts of those processes and documents realistically can and should be shielded, to what extent, by whom, from whom, for how long, and pursuant to what safeguards.

More Sociological Study
On the descriptive side, it seems to me that we need to expect both less and more from transparency: less in the way of unqualified upsides or immediately apparent consequences, and more in the way of unintended second-and third-order effects.
Gregory Michener's recent Viewpoint essay explains persuasively that many transparency policies "generate gradual, indirect, and diffuse impacts" (2019, 136), which may be missed by studies that concentrate on short time frames or on the most readily tabulated outcome variables. Michener worries that a focus on direct, quantifiable metrics leads scholars to undervalue the benefits that transparency policies bring, even in situations of lackluster implementation, as when these policies help foster bureaucratic competency or coordination in the medium run. A focus on such metrics, however, can likewise lead scholars to overvalue transparency. Qualitative research on the U.S. administrative state, for instance, suggests that certain transparency policies can lead, over time, to the hollowing out of bureaucratic capacity when transparency is weaponized by political opponents and regulated parties or enlisted repeatedly as a substitute for stronger regulation (Pozen 2018).
It cannot be taken for granted that by investigating a transparency policy's gradual, indirect, and diffuse impacts, the policy will come off looking better than if it were examined in a narrower fashion. The scholarly payoff from investigating these impacts is not that we will thereby reaffirm our priors and restore faith in transparency. The payoff is that doing so will allow us to learn more about how transparency works, whether for good or for ill.
To operationalize his insights, Michener calls for "greater use of qualitative and mixed methods," including "careful tracing of impact processes and indicators, combined with sensible counterfactual reasoning" (2019, 136). I wish to echo and extend this call, and to suggest a label that can capture the sort of project Michener envisions and the many different forms of academic inquiry that can participate in it: a sociological turn. If this label sounds strange, recall that transparency, like secrecy, is social by nature. Just as secrets are kept from certain actors, transparency is provided to certain actors. Transparency is an interpersonal, intersubjective, and relational phenomenon. Sociology, as conventionally defined (e.g., American Sociological Association 2019), integrates the insights of multiple fields to explore such phenomena as they play out in and across particular institutional settings.
A sociological approach is hardly a straitjacket. It need not require the adoption of any discrete theoretical paradigm. It can have a more or less historical flavor, a more or less ethnographic flavor, a more or less comparative flavor, and so on. The one thing to which a sociological approach (at least in the loose conception I am invoking) commits the researcher is critical scrutiny of the social dimensions of transparency policies: the cultural, political, and organizational environments in which they arise and evolve; the unwritten norms that condition their usage; the power dynamics and distributional disparities they reflect and create, within and beyond the corridors of government; the adaptive behaviors and counter-behaviors and counter-counter-behaviors they inspire; and all the other ways in which these policies shape, and in turn are shaped by, human agency and collective action.
A sociological approach, on this understanding, is less a matter of specialized disciplinary training than of emphasis and "imagination" (Mills 1959), less a matter of technique than of the questions that are posed and the answers that are taken to be satisfactory. It involves moving beyond threshold indicators such as response rates and processing times-suggestive as these data may be-and attending to broader features of transparency policies, with an eye toward the iterated interactions between formal legal structures and informal developments in the communities that supply, demand, and interpret information. In short, it reflects a commitment to reckoning with transparency in its full complexity as a social phenomenon-not in every study, but in the overall balance of work that is produced and in the orientation and aspirations of the field.

Conclusion
A sociological turn in transparency studies would also be something of a return. As exemplified by Georg Simmel (1906) and Max Weber (1922), the leading prewar theorists of transparency's twinsecrecy-were intensely interested in the social life of secrets and revelations, their uses as tools of social control and their effects on the circulation of power within institutions. Needless to say, this is not the place to delve into Simmel's or Weber's thought. My submission is simple: the sociological study of secrecy generated powerful insights in the past. It can do so for transparency today. If this essay is right that transparency is (1) implausible as a primary virtue in governance, (2) ambiguous as a secondary virtue, and (3) inherently social in nature, then sociological inquiry, broadly conceived, gives us our best hope for developing a deep understanding of transparency policies and their many and varied impacts.