*382 Substantial evidence in the aggregate and at the firm level suggests that this apparatus works. From 1990 to 1995‑‑the period for which TURA data exist‑‑the use of toxic chemicals fell by twenty percent in Massachusetts, and the generation of toxic byproducts by thirty percent, after adjustments to take account of changes in the levels of production. [FN362] Beyond this overall result, it is clear that the requirement to plan reduction of toxics use has enabled firms to discover significant net benefits to doing so and to value the public institutions that facilitate the planning. [FN363] The peer inspectorate and related programs were crucial to the positive outcomes. Of all the services provided to facilitate planning, the responding firms were most enthusiastic about toxics use planner training, followed by site visits from the OTA, while firms were least enthusiastic about the provision of toxics use reduction information by their own trade associations. [FN364] Nor were these benefits offset for the firms by the costs to them of meeting the new reporting and planning requirements. [FN365] A summary measure of the favorable balance of costs and benefits of toxics use reduction planning to the firms is that eighty‑six percent of all respondents said they would continue to plan even if there were no longer a requirement that they do so. [FN366]
c. Project XL and the Common Sense Initiative.‑‑In the wake of this and related experience, the EPA has launched, by its own, plainly incomplete count, thirty‑nine programs to fashion an experimentalist regime at the federal level. [FN367] Two of the most prominent and widely noted programs,*383 Project XL and the Common Sense Initiative (CSI), encourage firms, singly or in sectoral groups, to propose and eventually to test experimentalist reforms. Other programs encourage similar initiatives by consortia of states collaborating in various ways with regional and national offices of the EPA. [FN368] Still others focus on the reorganization of the EPA's own regional offices to respond to and foster state‑level experimentalism. [FN369] This flotilla of programs is evidence of the compelling attraction of the experimentalist alternative. But its wavering course and halting progress call attention to the difficulties of proceeding from piecemeal demonstration of the feasibility of this kind of reform to its generalization. We focus on XL and CSI because, as the most advanced and publicly visible of the programs, their record sheds the most direct light on these difficulties and hence the federal prospects for experimentalism.
Project XL is a pilot program for encouraging and supervising pilot programs. [FN370] It allows the federal government to authorize state regulatory bodies to permit the entities that they regulate to adopt environmental performance strategies which deviate from traditional requirements, on the condition that "superior" environmental benefits result and detailed records of environmental performance are made public. [FN371] Project XL substantially extends TRI and TURA in offering private actors, by means of waivers, a more comprehensive version of the core experimentalist bargain of greater autonomy in the determination of precise ends and means of environmental regulation, in return for increased monitorability. But it differs crucially from these programs in providing for neither benchmarking, nor a peer inspectorate, nor any other systematic form of deriving rolling rules of fair and effective behavior from emerging practice.
*384 In the absence of any standards of comparison, discussions between a private actor and the state, and then between both of these and federal authorities, become an extended negotiation to agree on a distinct, quasi‑private regulatory regime. Government waivers are piled atop waivers and private undertakings atop other undertakings. Such negotiations are by their nature arduous and costly. They require exchanges so intimate, particular, and extended between the state, the private actor, and other concerned parties, as almost inevitably to suggest to some of the participants that others are colluding against them, even when they are not. No wonder that Project XL has so far been of interest chiefly to very large, capable corporations whose constant product and process innovation make regulatory permitting under fixed rules a potentially paralyzing burden. No wonder, too, that relatively few such corporations have been able to negotiate successfully waivers and reporting regimes that meet their own requirements and those of both levels of government and the other parties. While the EPA set an initial goal of fifty projects, [FN372] as of the end of 1997, only seven XL projects had been finally approved; three were listed by the EPA as "facilitated" but not yet final; nine had reached the intermediate "development" stage; and thirty XL proposals had been rejected or withdrawn for a variety of reasons. [FN373]
A brief look at a typical failed negotiation involving the Minnesota Mining and Manufacturing Co. (3M) facility in Hutchinson, Minnesota, shows why, stripped of a pooled, background understanding of what results can be achieved and what risks are worth taking, bargaining in isolation is likely to fail. [FN374] At Hutchinson, the company proposed to keep combined or multimedia emissions to a level below existing regulatory limits in return for waivers from the standard permitting procedures. Negotiations failed, for one thing, because the EPA could not provide 3M with satisfactory assurances that compliance with XL would immunize the *385 company from civil liability for technical violations of existing statutes. [FN375] For another, there was disagreement as to the level of "superior" performance expected as a condition of the waiver. [FN376] 3M claimed to be outperforming current standards, but refused to guarantee that it would increase or even maintain this level of "superiority" under the new arrangements. This uncertainty was compounded by disputes among the regulatory entities over the allocation of authority among the Minnesota Pollution Control Agency and the Washington and regional offices of the EPA. [FN377] Both sources of confusion were in turn the fruit of the initial decision by the federal level of the EPA to declare itself open to experimental reform without providing any but ad hoc means for establishing mutual accountability and coordinating its own decisions with those of its field offices, let alone other government entities or private actors. [FN378]
The CSI, established in 1994, has had, if anything, more difficulty than Project XL in moving from experimental idea to actual experiment. Much of the reason is manifest in a complex, ambiguous hierarchical structure, apparently intended to encourage a form of information pooling, that in fact discourages it. At the top is a thirty‑two‑person council composed of representatives of industry, small business, labor, federal, state, and local government, environmental justice groups, and community‑based and national environmental organizations. [FN379] Under the aegis of this council are six sectoral subcommittees grouping stakeholders from automobile manufacturing, iron and steel, metal finishing, computers and electronics, printing, and petroleum refining. Each sectoral subcommittee in turn organizes project groups and teams that elaborate detailed reform proposals and may undertake pilot programs to test them. Reform recommendations based on the studies and experience of the project groups, passed through the sectoral subcommittees, must be approved *386 by the council as a whole before being forwarded to the EPA as Common Sense Initiatives. [FN380] Hence, in addition to all of the difficulties of achieving consensus regarding concrete measures, there is the imponderable difficulty of anticipating which projects, and what forms of presentation, will be acceptable to the superintending council, not to mention the EPA itself. The predictable result is concentration of the activity, such as it is, in the project groups‑‑often, as in the case of metal finishing, in programs of self‑monitoring, reporting, and standard setting with clear affinities to TURA [FN381]‑‑and indecision at the top. [FN382] Worse still, it may be that project groups and especially pilot projects keep some or much of their results to themselves, guarding against the possibility of higher‑level interference, but also fragmenting and occluding information instead of pooling it.
The difficulties of XL and CSI point to what could prove to be a characteristic political dilemma in the diffusion and generalization of experimentalist methods. After the failures of standard programs have opened the way to alternatives, and these have progressed far enough to promise large‑ scale feasibility, public and private actors will often divide on what comes next. Those most inconvenienced by the current regime, or ideologically opposed to any form of government regulation, will urge abolition of the rump of traditional rules as a precondition for further and conclusive reform. One of their chief arguments will be that the experimentalist success would become self‑reinforcing if only the obstacle of current law (whose costly disutility it dramatizes) were abrogated. On the other side will be those who see themselves as beneficiaries of the current rules, or of regulatory protection against market forces in general. They will argue that the early successes could come undone if the old rules are discarded before new ones‑‑demonstrably ensuring that the *387new regime is, indeed, self‑reinforcing‑‑are in place. Caught in this to and fro, politicians and administrative officials will temporize, waiving the rules on all manner of conditions to placate those who want to try new things, but keeping the rules on the books to appease those who fear wholesale deregulation. Meanwhile, the authorities will hope that amidst the waiving there emerges just the experimentalist solution that conciliates both camps. The result will be the profusion of experimentalist activism in the small, and institutional immobilism in the large, that we observe in the EPA programs.
The experimentalist program we are advancing suggests a response to this dilemma. At almost any point in shifting from traditional to experimentalist regulation, it is possible to advance still further by extending the benchmarking apparatus, thus providing the instruments and the impetus for the actors themselves to demonstrate new possibilities for solving problems within current rules or devising alternatives to those rules. This is the lesson of TRI and TURA. Imagine, then, for example, that Congress amended the legislation governing the TRI to include the planning, peer‑inspectorate, and inter‑agency coordination features of TURA. In that case, actors nationwide would, as in Massachusetts, rapidly learn which kinds of public and private services and reporting regimes favor comprehensive environmental improvements, and the best of these could then be used to create a framework for an experimentalist framework for changing the traditional rump. Yet, whatever the advantages of such a system, the notion that it culminates in a framework for a framework suggests why, in the end, even augmented information pooling can only be a palliative, not a definitive solution. Eventually, the traditional rump stands in the way of further experiments. What then?
Here is where the basal ideas of experimentalism matter. Within representative democracy, "experiments" conducted within the confines of existing law, but with a view to changing existing law, are likely to turn into lobbying efforts rather than directly deliberative experiments. Those who entertain truly innovative plans that might put themselves and their ideas at risk will hold back for fear of overstepping legal boundaries. That is certainly part of the lesson of XL. [FN383] Many of those who participate in these confined "experiments" will do so in order to advance ideas they firmly hold, not to test them. That such participation leads to more maneuvering than forward motion is surely part of the lesson of CSI. Put another way, the effort to advance experimentalism by sparing it political conflict over the rump of rules leads to a paralyzing politicization of the experiments.
The alternative is to apply experimentalist principles of accountability as much to projects that challenge the rump of rules as to those that *388 do not. Imagine that the EPA asked Congress for explicit authority for itself and associated state entities to give temporary waivers from certain statutory permitting requirements for firms that meet generally applicable, rolling rules regarding environmental performance and reporting. This would create an open competition to devise a rolling alternative to current practice, acceptable to a wide range of public and private actors. Congress would, in the end, decide if the winner was an improvement sufficient to warrant a definitive switch. Experimentalism would be a new form of lawmaking, not a new method for influencing the lawmakers. If experimentalism, as we are projecting it, has a future, then environmental regulation is one of the first areas where, beyond the morass of waivers and pilot programs for pilot programs, we should see this kind of change taking place.
D. The Role of Courts in the National Experimentalist System
Even more than the other branches, the courts [FN384] are the institutions in which existing conceptions of constitutional democracy appear to flow seamlessly into experimentalism. "Experimentalist courts" are thus nothing more than traditional Article III courts transformed by the new methods of organization in the political branches of government and society around them. Experimentalist courts, like the traditional courts of constitutional democracy, function by a form of direct deliberation: [FN385] Citizens, as individuals or groups, speaking with the authority of their own experience, can demand that the government give reasons for its actions. In constitutional and administrative law litigation, such claims will ultimately be grounded in both traditional and experimentalist courts in norms of due process, freedom of expression, and equal protection, broadly understood as necessary elements of democracy, as well as the more concrete doctrines applying these and other textual guarantees. [FN386] But whatever the nominal source of the citizens' right to reasoned decisionmaking, the courts, experimentalist and conventional, are the place where individuals can insist that the polity, and the government that works in its name, justify again, by reference to its deepest values and its best understanding of relevant experience, the justifications given so *389 far for particular actions. It is in the courts, therefore, that experimentalism manifests its continuity with constitutional democracy, and constitutional democracy manifests its most directly deliberative and experimentalist aspect. [FN387]
The novel features of experimentalist courts result from their relations to the other experimentalist organs of government: Congress, state and local governments, and administrative agencies. As these entities adopt experimental methods, they come to elaborate new understandings of fundamental principles in the course of seeking solutions to concrete problems. The ebb and flow of reason giving is the political process of directly deliberative polyarchy. In effect, it obliges the actors to elaborate fundamental principles while assessing the practical consequences of different rules of order. This process substantiates the sovereign intent, the means associated with it, and the authorities' fidelity to the constraints imposed by both. Accordingly, a two‑ fold transformation in judicial decisionmaking serves as both a precondition and consequence of this enlarged role for social self‑explication.
First, the courts must develop an explicit understanding of fundamental legal norms deeply entrenched yet always provisional in the sense that the means by which core values are both protected and ultimately defined are deliberately exposed to experimentalist understanding. By insisting that actors respect the central experimentalist condition of declaring goals and measuring results, the courts can declare and defend inchoate rights without pretending to anticipate the manifold consequences of the finding.
Second, experimentalist courts defer to the political actors' exploration of means and ends only on the condition that the actors have in fact created the kind of record that makes possible an assessment of their linking of principle and practice. The system that experimentalist judicial review enables thus introduces constitutional values into the political decisions of everyday life while bringing the lessons of everyday life into the discourse of constitutional value. Put another way, experimentalist courts can serve democracy better not only because they presume to provide fewer definitive answers to legal, social, and ultimately political questions, but also because they can inquire into more of the political actors' own deliberative capacities.
Judicial review by experimentalist courts accordingly becomes a review of the admissibility of the reasons private and political actors themselves give for their decisions, and the respect they actually accord those reasons: a review, that is, of whether the protagonists have themselves *390 been sufficiently attentive to the legal factors that constrain the framing of alternatives and the process of choosing among them. Constitutional review in particular becomes a jurisprudence of impermissible arguments and obligatory considerations‑‑the former forbidding the actors to pursue ends found to be unconstitutional; the latter enjoining them to give particular attention to their choice of means when constitutional values appear to be at risk.
This Section focuses on the current problems courts face in construing ambiguous statutory and constitutional text, and the way the redistribution of reason giving in experimentalist courts provides a partial solution. Succeeding Parts rely on nascent constitutional doctrine to illustrate how experimentalist methods can reveal the boundaries of the legitimate zone of constitutional experimentation in federalism, separation of powers, and most important, a reinterpretation of the idea of individual rights that accompanies the new jurisprudence. [FN388]
1. The Dilemma of Judicial Review as the Muddle of Means and Ends.‑‑Under current conditions, the indeterminate relation of legislative means to ends bedevils judicial review of administrative action and the constitutionality of legislation. Absent well‑articulated connections between means and ends that the political actors have themselves elaborated, the courts must speculate about their relation. This speculation typically takes the form of balancing policy objectives against prima facie affronts to the legal order. Yet, because open balancing embroils the court in political decisions, it balances surreptitiously through the use of complex rules that determine the degree of judicial scrutiny to which an individual case will be subjected. But because the application of these tests in turn involves a suspiciously political calculus, the court must then defend the tests and categories in a form of second‑order balancing.
Consider the paradigmatic difficulty of determining the relation between the means and ends of a particular law, starting with ends. A straightforward way to know whether a law's ends are constitutional is to determine the legislature's intent in enacting it. But there is no simple way to make this determination: The majority that voted for a bill is composed of legislators who cast their vote for various reasons, [FN389] among which may be the desire to occlude the true reason for the vote. The alternative to looking to declared or subjective intent is to infer the so‑called "objective" intent from the statute itself. This is, of course, a judicial construction: Given the means chosen by the legislature, the courts *391 infer the statute's purposes or ends. But this maneuver of inferring the ends from the means yields a determinate result only if it is possible to assess the significance of the means without referring back, in circular fashion, to the indeterminate ends. [FN390]
Yet scrutiny of means does retrace the circle. A court bent on sustaining legislative power can simply define the statute's objective as the accomplishment of whatever it is the statute happens to accomplish. [FN391] Conversely, if a court first divines a legislative goal, then whatever means the legislature chooses, the court can devise other, less menacing ones, depending on its view of the legitimacy and urgency of the goal to be achieved. Suppose, for example, that the Supreme Court decides that preferential hiring of minority contractors is allowable as a means, provided that it is essential to achieving the allowable end of reduction of racial discrimination in the contracting industry in a particular city. [FN392] Or suppose, to return to the State Farm case, the Court finds that a particular, passive automobile safety restraint is an acceptable means to the end of increasing highway safety, provided that there is no better one. [FN393] Given a court's limited fact‑finding capability‑‑its limited ability to explore alternatives‑‑how can it know that means are not sufficiently closely tied to ends?
Under current practice in both the administrative and constitutional law contexts, the Court speculates about what are essentially empirical matters, asking questions about the advantages and disadvantages of alternatives that the protagonists might well have asked but did not: Was official discrimination in fact severe enough in the city in question to justify the inevitable social and individual costs of affirmative action? [FN394] Were passive restraints such as airbags feasible and therefore preferable to rescission of Standard 208? [FN395] Thus, in scrutinizing means by, for example, inquiring whether a statute is "narrowly tailored" to further a constitutional end, the Court in effect balances its estimates of the constitutional harm entailed by the legislative solution against its estimates of the greater costs (or benefits) to society that result from pursuing a constitutionally preferable (or equivalent) one. [FN396] The balance will often appear *392 capricious or partial because the standards for measuring costs and benefits of means can easily appear to have been derived from evaluations of the worthiness of ends. Thus, if the Court sets an extraordinarily high value on a particular vision of racial justice, almost any affirmative action program can be judged to produce benefits on balance greater than costs. In finding otherwise, by this logic, the Court subordinates racial justice to other values (including a competing vision of racial justice itself), even though it purports to answer the empirical question whether the challenged program could have achieved its worthy goals by other means. However, there is no satisfactory reason why the Court, rather than the legislature, is the appropriate institution to raise and answer such empirical questions and then make and compare the estimates of benefit and harm to which the answers lead. For these reasons, such scrutiny evokes the menace of the countermajoritarian difficulty. No wonder the Court is at pains to disguise the extent to which its formal constitutional tests rely on balancing techniques. [FN397]
The Court's efforts to reduce the need to engage in this suspect balancing by reducing its cause‑‑its own uncomfortable position as clarifier of ambiguous authoritative meaning‑‑lead to unpalatable results in representative democracy. One strategy is to require the legislature to articulate its purposes upon enacting legislation, and therefore disallow (re)interpretation of those purposes before a court responding to a subsequent constitutional challenge. This suggestion has been current for a quarter century; [FN398] the Supreme Court has followed it in some cases subjecting legislation to heightened scrutiny of constitutionality. [FN399] But a determined legislature could circumvent such a prohibition simply by declaring a purpose that immunized its legislation from later judicial challenge. Hence, the Court has not embraced a general exclusion of