Introduction: The Pasts & Futures of the Administrative State

Abstract To understand contemporary arguments about deconstructing and reconstructing the modern administrative state, we have to understand where that state came from, and what its futures might be. This introductory essay describes the traditional account of the modern administrative state's origins in the Progressive era and more recent revisionist accounts that give it a longer history. The competing accounts have different implications for our thinking about the administrative state's constitutional status, the former raising some questions about constitutionality, the latter alleviating such concerns. This introduction then draws upon the essays in this issue to describe three options for the future. Deconstructing the administrative state without adopting a program of across-the-board deregulation would entail more regulation by the legislature itself and would insist that Congress give clear instructions to administrative agencies. Tweaking would modify existing doctrine around the edges without making large changes. Reconstruction might involve adopting ever more flexible modes of regulation, including direct citizen participation in making and enforcing regulation.

P residential adviser Stephen Bannon might have simply been coining a phrase rather than outlining a program when he said that the Trump administration was interested in "deconstructing the administrative state." 1 Yet by replacing the familiar term deregulation with the unfamiliar deconstruction, Bannon captured a wider discomfort with how the modern administrative state was operating. That discomfort manifested itself in many forms: concern about the "ossification" of the process of adopting important regulatory rules across many domains, for example, and recognition that new regulatory tools could be more effective than traditional methods of prescriptive ("command and control") regulation. 2 This issue of Daedalus explores what deconstruction and its obverse reconstruction of the administrative state might be-and whether either is called for.
In their contributions to this volume, Susan Dudley and Peter Strauss lay out in some detail their accounts of the administrative state's emergence in the United States, situating the ensuing discussion of reconstruction and deconstruction. 3 Here I offer my highly simplified version-quite a bare sketch of those develop- The first story was told by the scholars who created the academic field of administrative law: Felix Frankfurter, James Landis, and, to a lesser degree, John Dickinson. Drawing on arguments made by an earlier group of Progressives, they found the origins of the administrative state in the late nineteenth century and argued that accommodating contemporary reality to the classical vision of U.S. constitutionalism required altering the latter so that the "new fourth branch" could fit within the Constitution. The second is a revisionist story offered by contemporary historians and legal scholars who have retrieved a longer history of the administrative state, dating to the early nineteenth century. For these scholars, the modern administrative state has always fit within the Constitution.
The Progressive story takes the 1887 creation of the Interstate Commerce Commission as the symbolic dawn of the modern era. The story identified a number of social and economic developments that, according to Progressives, had weakened the ability of the traditional institutions of government to provide effective governance. Technological change, again symbolized by the railroad but encompassing what we now refer to as information technology (the telegraph and the telephone), generated new problems: exploitation of workers and farmers, for example, and new political possibilities enabled by "yellow journalism." So did rapid urbanization and immigration; the modern city was overcrowded, rife with environmental dangers and crime.
The Progressive story asked: what institutions were best suited to dealing with these problems? Their answer had a negative side-not the existing system of legislatures, courts, and political parties-and a positive one-the new administrative agencies guided by professionals deploying the findings of contemporary science. Courts failed because they could intervene only episodically, when someone happened to bring a case before them. The cases the judges saw gave them a view of randomly selected parts of more general problems, and sometimes solving the problem at hand would perversely make things worse elsewhere. And to the extent that problems like workplace safety came to the courts' attention, the judges lacked both the expertise and the capacity to impose appropriate solutions: they might find railroads liable when they failed to provide workers with the equipment to allow them to disconnect and reconnect railroad cars safely, but they could not prescribe that the railroads use any particular system for doing so, even when engineers knew what the best system was.
Legislatures were inadequate in part because, dominated by politicians whose primary interest lay in holding on to power rather than advancing the public good, they failed to address new problems as they appeared-or at least failed to do so rapidly enough. By the time a problem became politically salient, the Progressive story had it, too much social damage had been done. A more nimble and self-starting 7 Mark Tushnet body that could identify problems rapidly-and without concern for whether voters cared about it enough to pressure their representatives-was needed. Further, even when legislatures did address real problems, they lacked the expertise to come up with the right solutions. Again, agencies staffed by professional experts in specialized fields would do better.
On this account, legislatures would do best by identifying some general field of regulatory concern (such as prices for shipping goods by railroad, workplace safety, environmental degradation), creating an agency to deal with that field, and instructing the agency to develop regulations that best promoted public welfare. That latter instruction received the doctrinal label delegation. The Progressives argued that delegations probably had to be stated in quite general terms, such as "public convenience and necessity" or, in a modern statute, "requisite to protect the public health with an adequate margin of safety." 4 Congress could legislate in more detail, and sometimes did so, but, according to the Progressive account, broad delegations of regulatory authority were both inevitable and constitutionally permissible.
The breadth of the Progressive idea of science as the guide to public policy deserves special note. To them, science provided answers not merely to technical engineering problems but to all sorts of social ills. Economists could determine a "fair" rate of return on investment, for example. Sociologists could devise programs that would address the "root causes" of urban crime. Labor relations specialists knew how to mediate disputes between employers and workers in ways that would avoid strikes. 5 As I suggest later, contemporary arguments about whether or how much the administrative state should distinctively "follow the science" flow in part from a similarly expansive understanding of what science can tell us.
For students of administrative law, the first important revisionist work was William Novack's The People's Welfare: Law and Regulation in Nineteenth-Century America (1996), which illuminated a history of robust regulation at the state level well before 1887. What, though, of the national level? Writing in 1982, political scientist Stephen Skowronek described the national government in the nineteenth century as a state of "courts and parties." 6 Several decades later, historian Brian Balogh wondered about the "mystery of national authority" during that same period. 7 More recently, though, legal scholar Jerry Mashaw found a "lost" history of the administrative state. 8 Mashaw described scattered but persistently recurring forms of national regulation starting in the early republic that looked almost exactly like the forms of administrative governance that the Progressives celebrated. Other legal scholars have identified broad delegations of authority from Congress to executive branch officials from that same early period. 9 These stories matter today because the Progressive story has come to generate a response in the register of deconstruction. 10 If, as that story holds, the modern administrative state does not fit within traditional U.S. constitutionalism, and if, as is surely true, the Constitution was not formally amended to address that state's novelty, it follows for deconstructionists that important aspects of the administrative state must be revised. For them, the Constitution demands that Congress make major policy choices that, in the Progressive story, it has (improperly) delegated to administrative agencies, and that courts rather than agencies determine the scope of regulation that Congress has authorized. The revisionist story, in contrast, suggests that the contemporary administrative state is one of many possibilities that the original Constitution enabled. A deconstructed administrative state would of course be within the wide bounds the original Constitution created, but so is the modern administrative state, and so would a reconstructed administrative state. The revisionist story, that is, shifts our attention from constitutional limits on the administrative state to the policy choices open to us today. 11 F rankfurter, Landis, and Dickinson wrote about the administrative state in the 1930s. By then, modern administrative agencies had become part of the landscape. The New Deal produced a new group of "alphabet" agencies, the SEC (Securities and Exchange Commission) and the NLRB (National Labor Relations Board) being the most politically prominent. Conservatives assailed these agencies as unconstitutional and then, after the Supreme Court rejected their constitutional arguments, shifted attention to what came to be known as administrative law, a theretofore marginal legal category.
The attack combined several themes. 12 The first sounded in good-government. As each new agency was added to the system, a body of law developed about that agency, without any attention to how that body of law was related to the law governing other agencies. Courts applied a plethora of "standards of review" that differed in verbal formation and sometimes in practical application. Sometimes the courts gave an agency's findings great weight; at other times they found facts anew. Reformers sought a unified body of administrative law, eventually codified as the Administrative Procedure Act (APA), that would be, as the term was, "transsubstantive": that is, the same no matter what subject matter the agency dealt with.
The other theme was straightforwardly political. The new administrative agencies were out of control, dedicated to a transformation of the national economy that voters never truly endorsed. Suggestively, the initial proposals were to impose a uniform set of standards across all agencies. Then, as they proceeded through the legislative process, the proposals were pared down: the "traditional" agencies like the Interstate Commerce Commission and the Federal Trade Commission were dropped from the proposals' coverage, and aggressive judicial oversight was to attend only the actions of New Deal agencies.
President Franklin Roosevelt understood the Walter-Logan bill that reached his desk in 1941 as a challenge to the New Deal-and vetoed it. The good-govern-Mark Tushnet ment and political forces that had produced the bill remained in place, though, and Roosevelt promised a study to develop a statute that unified administrative law without threatening the advances, as he saw them, of the New Deal. The outcome was the APA, which for more than three-quarters of a century has provided the legal foundation for the administrative state. 13 S tability in the basic document, of course, does not mean that the administrative state has been static. As Dudley and Strauss show, a second proliferation of administrative agencies occurred in the 1960s and 1970s: the Equal Employment Opportunity Commission (1965), the Environmental Protection Agency (1970), and the Occupational Safety and Health Administration (1970), among others.
The politics associated with administrative law changed as well. As administrative law scholar Richard Stewart has argued, the Progressive account of the administrative state's rise, with its focus on expertise to advance the public good, ended in the 1930s just as that politics was about to change. 14 Rather than seeing politics as devoted to advancing the public good, scholars began to see it as the venue for interest group bargaining and administrative agencies as one of the forums for that bargaining. As such, they became targets for "capture" by the industries they regulated. Reformers developed several responses, the most important of which were expanded notions of standing to challenge agency action, which brought new players into the bargaining game, and creating agencies with economywide jurisdiction, which made them less susceptible to capture by any specific industry. Then, to recreate the synoptic view of problems that agencies were supposed to take, the president began to assert greater powers of supervision. 15 The upshot is clear. The politics associated with administrative law came to resemble that associated with legislation: administrative agencies became bogged down in exactly the same morass that legislatures were. They became inflexible, unable to respond nimbly to new problems, and committed to established routines that had "worked" before even when they might not work well today. Neomi Rao's contribution to this volume identifies several important pathologies she associates with the contemporary politics affecting agency operation. 16 Strikingly, the contemporary economic and social landscape appears to many observers quite similar to the regulatory domain that Frankfurter and Landis described as characterizing the late nineteenth century. Its most prominent feature, perhaps, is rapid technological development, for which "the Internet" and "the new social media" are shorthand. Demographic changes too are part of the landscape. Immigration changed the nation's largest cities in the late nineteenth century, and recent waves of immigration have changed smaller cities and even some rural areas. The urbanization of the late nineteenth century is paralleled by a widening rural-urban gap in the twenty-first.

Introduction: The Pasts & Futures of the Administrative State
And, important for our topic, contemporary forms of governance seem illsuited to deal with today's landscape. The critiques the Progressives leveled against courts and legislatures remain on the table. Courts proved to be innovative in developing what civil procedure scholar Abram Chayes called "public law litigation" over school desegregation and prison. 17 More recently, courts have entered "universal" injunctions, some of which aim at restructuring national policies rather than the state and local ones with which Chayes was concerned. Universal injunctions remain quite controversial, and they have proven more effective in blocking policy than in developing it. As of today and probably for the foreseeable future, these innovations are unlikely to spread broadly enough for courts to become general regulators in response to novel challenges.
Legislatures are gridlocked, unable to address new problems with anything like the alacrity that (some think) they demand. Legislators devote attention to problems that catch the public eye rather than those with deep roots that are largely invisible until they erupt into some policy disaster. And as I have already noted, the administrative agencies that Frankfurter and Landis saw as solutions to judicial and legislative failures are themselves caught up in what legal scholar Thomas McGarity has called "blood sport politics," unable to act quickly in response to new challenges and equally unable to produce stable policy responses when they manage to act. 18 And finally, the idea of disinterested scientific expertise has come under sustained assault from all sides. Some of the challenges are retrograde, as with climate change denialism, while some are purportedly sophisticated, as with postmodernist critiques of science. Some, though, have substantial merit, mostly because technocrats have in fact claimed that science provides more answers to public policy problems than is possible.
I offer what will surely be a controversial example: the public policy response to the COVID-19 pandemic. Epidemiologists and medical doctors gave us their best estimates of the risks associated with various policy options (border closings, mask mandates, temporary or prolonged curfews, and shutdowns) in light of the information available to them when they estimated the risks. Economists gave us their best estimates of the economic costs and, as Cass Sunstein emphasizes in his essay in this volume, ballpark estimates of the costs to human life and health associated with each option. 19 Neither epidemiologists nor economists, though, could tell us which policy we should adopt, in part because their estimates were inevitably fuzzy and, under the circumstances, should have changed as information accumulated and in part because, notwithstanding the economists' best efforts, only devoted technocrats believe that costs to the economy and costs to human life and health can be measured by a single metric. Technocratic-driven policy choices, which of course have to be implemented through politics, proved to be unstable in the face of public skepticism about how much the experts really could tell us. "Following the science" can bring policy-makers to the point where they Mark Tushnet could make reasonably well-informed choices, but "the science" could not and did not tell them what choice to make. P erhaps, then, today's administrative state is at a point structurally similar to the one the government had reached in the late nineteenth century: new problems posed by technological, economic, and social change, thrown at a governance system whose institutions are ill-adapted to deal with them. If so, what institutional responses might there be? After noting the possibility that the current situation is hardly as dire as deconstructionists and reconstructionists suggest, I describe the available institutional responses as deconstruction, tweaking, and reconstruction, acknowledging that the categories are not separated by sharp boundaries.
Perhaps, as Sunstein suggests, we should recommit ourselves to the Progressives' technocratic vision, in the contemporary form of a comprehensive costbenefit state. To Sunstein, skeptics about monetizing all sorts of costs and benefits are simply mistaken. On this view, students of the modern administrative state should do their best to show legislators and those they represent that cost-benefit analysis produces regulatory decisions that are better than any available alternative. Here David Lewis's observation that the modern administrative state has been battered by decades of criticism of its performance comes into the picture. 20 We should tout such major successes of regulation as the dramatic improvement in the nation's air quality to show that the modern administrative state works rather well. The point generalizes: regulating well is the best way to vindicate technocrats' claims about the contributions they make to public welfare.
If the administrative state is not working well, though, what to do? In early work, Stewart glimpsed the possibility of deconstruction but thought that it could not take the forms most prominently offered by the administrative state's conservative critics. 21 Those forms were deregulation and privatization. Both would remove the state entirely from the domain of regulation.
Today, I think, we should focus on the administrative part of "deconstructing the administrative state." Deconstructing that state would mean dramatically scaling back the activities of administrative agencies without becoming committed in principle to no regulation at all. In a deconstructed administrative state, governments could regulate but would have to do so through detailed legislation rather than through delegations to administrative agencies. Privatization would mean not the complete transfer of authority to private corporations, but rather the design and interpretation of contracts between governments and those companies. The law of privatization would focus on what contract terms would best accomplish the purposes lying behind privatization, which typically involve taking advantage of entrepreneurs' incentives to find cost-effective methods to achieve public-regarding goals such as safety or education.

Introduction: The Pasts & Futures of the Administrative State
Tweaking the administrative state today is a modestly conservative program to scale back judicially imposed additions (in the guise of interpretations) to the Administrative Procedure Act. In their contributions to this Daedalus issue, Aaron Nielson and Christopher Walker describe different routes to retrieving the APA's original goals of constructing an administrative law that gives proper scope to regulatory authority within bounds set by Congress, and Sean Farhang, in his essay, describes how tweaking is already occurring. 22 Placed in the argument developed here, tweaking is motivated by the adverse effects of importing the interest-group model of policy development into administrative law. Reducing the scope of "public interest" standing to initiate or challenge administrative action would enhance the agencies' independent expertise. Taking seriously the APA's requirement that rules be accompanied (only) by a short description of their purposes would allow agencies to replace efforts to "litigation-proof" their rules by providing extremely detailed explanation with efforts to develop better regulation. And, as Lewis emphasizes in his essay in this volume, tweaking the administrative state by funding it adequately would allow managers to get on with the work of regulating well. E fforts to deconstruct and tweak the administrative state have models to build on. Reconstruction, in contrast, requires more imagination, even speculation. Artificial intelligence and automated regulation, discussed by Bernard Bell and Cary Coglianese in this volume, provide some hints of possibilities and might soon give us something like a proof of concept about a reconstructed administrative state. 23 So do Beth Noveck's descriptions in her essay of recent uses of big data in regulatory design and enforcement. 24 An implicit exchange between Walker and Charles Sabel and Jeremy Kessler illustrates some possibilities. Both essays note that regulators have begun to use "guidance" documents-formally, statements about an agency's plans for implementing its interpretation of the statutes it is charged with enforcing-as forms of regulation. Though, in form, guidance documents merely state intentions, the targets of regulation, facing the possibility that they will have to mount an expensive defense of their practices, have strong incentives to conform to the agency's interpretations, which thereby have the same effect that a full-fledged regulation would. Except, as Walker observes in his essay, under current interpretations of the APA, targets who could readily challenge a statutory interpretation embodied in a regulation face substantial obstacles in obtaining judicial review of an interpretation offered in a guidance document. Echoing many others, Walker would tweak the current regime by expanding the opportunities to obtain judicial review of regulation by guidance.
In their essay, Sabel and Kessler, in contrast, place guidance in a broader framework. 25 As I have already noted, the Progressive themes of rapid technolog-Mark Tushnet ical change have reemerged in thinking about reconstructing the administrative state. Sabel and Kessler see regulation by guidance as an innovative response to such change. Administrative law in its current form makes it time-consuming and costly for an agency to regulate by rule-one source of ossification. By the time an agency can work through rule-making, the problem it is trying to address will have changed shape. Regulation by guidance, Sabel and Kessler argue, restores the nimbleness and flexibility that modern governance requires.
But, in their view, it offers more. Retrieving ideas offered a century ago by Progressive philosopher and political theorist John Dewey, they argue that regulation by guidance is one of a family of alternatives to command-and-control or prescriptive regulation that can yield policies that can be adjusted to produce increasingly beneficial outcomes. They see regulation by guidance as similar to a more familiar form: regulation by output rather than input. Regulation by output sets goals that regulated companies or other regulated entities must reach-levels of pollution emitted, for example-and lets the regulated entity figure out how to reach them. The Deweyian insight is that the agency can then observe the choices the companies make and use that information to push forward more prescriptive regulations. Similarly with regulation by guidance: agencies can see how the entities they regulate respond to the incentives the guidance documents provide and deploy their expertise to evaluate the effectiveness of alternative responses, then move to more prescriptive regulation, or to less regulation, if experience shows that things are going reasonably well.
One feature of a reconstructed administrative state, then, might be building the process of learning-by-doing into the state's institutions. But, we might ask, who is to do the "doing"? One feature of the blood-sport politics of contemporary administrative law is deep contention over who gets to be a member of the leadership of today's alphabet agencies. Combine this with the "deliberative deficit" Avery White and Michael Neblo identify in their essay and the imaginative possibilities for substantial institutional innovation open up. 26 White and Neblo put on the table the possibility of regulation by an administrative law parallel to the citizens' assemblies and similar bodies that have been used to develop policies across a range of topics. Within this framework, a "modest" program would have a central body prescribe regulatory goals and have citizens' assemblies devise implementation techniques appropriate to their local circumstances. That program can be founded upon the Deweyian idea that ordinary people combine common sense with local knowledge in ways not readily accessible to regulators more removed in time and space from the point of implementation. 27 A more ambitious program would take participatory budgeting exercises as a model. The legislature would single out a relatively discrete problem, such as waste-water pollution at some industrial facilities. Groups of citizens would meet locally (both where the plants are located and where the plants' products are con-sumed), discuss and debate regulatory proposals, and send the one they adopt to the next level, where another citizens' assembly would debate the various proposals they received, adopt one, and again send it up to the next level, ultimately with a single policy adopted, perhaps by a regulatory agency or a legislature, but perhaps instead by a "grand" national citizens' assembly. 28 A variant would use a single, relatively small citizens' assembly as the sole decision-maker. In this model, a host of such assemblies is convened. Each deals with a single regulatory problem, defined narrowly (such as disposition of polluted waste water from fracking operations) or broadly (such as enhancing air quality in specified locations). 29 Members, who we can describe as "members of a regulatory agency," are chosen at random from the general population, compensated appropriately for their time, provided with general information about the problem, authorized to call upon whatever experts they think will be helpful, and-cruciallycharged with coming up with fully enforceable regulations. With a large number of these "participatory regulatory agencies," every citizen would have some opportunity to be a lawmaker in some regulatory domain, satisfying at least some definitions of democracy, through a deliberative process. O f course I have pushed the suggestions from the contributions to this Daedalus issue far beyond the limits of any individual essay. Yet in my view, serious consideration of deconstructing and reconstructing the administrative state requires highly speculative proposals coupled with small-scale efforts to provide proof of concept. 30 The modern administrative state emerged in the late nineteenth century and took its current form in the late twentieth century. It was shaped by economics, technology, politics, and, of course, the U.S. Constitution. If it is to be reconsidered-defended anew, deconstructed, or reconstructed-those same forces will come into play, or perhaps better, have already come into play. From my narrow perspective as a constitutional lawyer, the next step will be to see whether or how any new form of the administrative state can be accommodated to the existing Constitution, or an amended one.