EDMUND G. BROWN, J
., GOVERNOR OF CAL-
IFORNIA, et al ., APPELLANTS v. MARCIANO
PLATA et al.
on appeal from the united states district courts for the eastern district and the northern district of california
Justice Scalia , with whom Justice Thomas joins, dissenting.
Today the Court affirms what is perhaps the most radical injunction issued by a court in our Nation’s history: an order requiring California to release the staggering number of 46,000 convicted criminals.
There comes before us, now and then, a case whose proper outcome is so clearly indicated by tradition and common sense, that its decision ought to shape the law, rather than vice versa. One would think that, before allowing the decree of a federal district court to release 46,000 convicted felons, this Court would bend every effort to read the law in such a way as to avoid that outrageous result. Today, quite to the contrary, the Court disregards stringently drawn provisions of the governing statute, and traditional constitutional limitations upon the power of a federal judge, in order to uphold the absurd.
The proceedings that led to this result were a judicial travesty. I dissent because the institutional reform the District Court has undertaken violates the terms of the gov- erning statute, ignores bedrock limitations on the power of Article III judges, and takes federal courts wildly beyond their institutional capacity.
The Prison Litigation Reform Act (PLRA) states that “[p]rospective relief in any civil action with respect to prison conditions shall extend no further than necessary to correct the violation of the Federal right of a particular plaintiff or plaintiffs”; that such relief must be “narrowly drawn, [and] exten[d] no further than necessary to correct the violation of the Federal right”; and that it must be “the least intrusive means necessary to correct the violation of the Federal right.” 18 U. S. C. §3626(a)(1)(A). In deciding whether these multiple limitations have been complied with, it is necessary to identify with precision what is the “violation of the Federal right of a particular plaintiff or plaintiffs” that has been alleged. What has been alleged here, and what the injunction issued by the Court is tailored (narrowly or not) to remedy is the running of a prison system with inadequate medical facilities. That may result in the denial of needed medical treatment to “a particular [prisoner] or [prisoners],” thereby violating (ac-cording to our cases) his or their Eighth Amendment rights. But the mere existence of the inadequate system does not subject to cruel and unusual punishment the entire prison population in need of medical care, including those who receive it.
The Court acknowledges that the plaintiffs “do not base their case on deficiencies in care provided on any one occasion”; rather, “[p]laintiffs rely on systemwide deficiencies in the provision of medical and mental health care that, taken as a whole, subject sick and mentally ill prisoners in California to ‘substantial risk of serious harm’ and cause the delivery of care in the prisons to fall below the evolving standards of decency that mark the progress of a maturing society.” Ante , at 7, n. 3. But our judge-empowering “evolving standards of decency” jurisprudence (with which, by the way, I heartily disagree, see, e.g. , Roper v. Simmons , 543 U. S. 551, 615–616 (2005) ( Scalia, J. , dissenting)) does not prescribe (or at least has not until today prescribed) rules for the “decent” running of schools, prisons, and other government institutions. It forbids “indecent” treatment of individuals—in the context of this case, the denial of medical care to those who need it . And the persons who have a constitutional claim for denial of medical care are those who are denied medical care—not all who face a “substantial risk” (whatever that is) of being denied medical care.
The Coleman litigation involves “the class of seriously mentally ill persons in California prisons,” ante , at 8, and the Plata litigation involves “the class of state prisoners with serious medical conditions,” ante , at 9. The plaintiffs do not appear to claim—and it would absurd to suggest—that every single one of those prisoners has personally experienced “torture or a lingering death,” ante , at 13 (internal quotation marks omitted), as a consequence of that bad medical system. Indeed, it is inconceivable that anything more than a small proportion of prisoners in the plaintiff classes have personally received sufficiently atrocious treatment that their Eighth Amendment right was violated—which, as the Court recognizes, is why the plaintiffs do not premise their claim on “deficiencies in care provided on any one occasion.” Ante , at 7, n. 3. Rather, the plaintiffs’ claim is that they are all part of a medical system so defective that some number of prisoners will inevitably be injured by incompetent medical care, and that this number is sufficiently high so as to render the system, as a whole, unconstitutional.
But what procedural principle justifies certifying a class of plaintiffs so they may assert a claim of systemic unconstitutionality? I can think of two possibilities, both of which are untenable. The first is that although some or most plaintiffs in the class do not individually have viable Eighth Amendment claims, the class as a whole has collectively suffered an Eighth Amendment violation. That theory is contrary to the bedrock rule that the sole purpose of classwide adjudication is to aggregate claims that are individually viable. “A class action, no less than traditional joinder (of which it is a species), merely enables a federal court to adjudicate claims of multiple parties at once, instead of in separate suits. And like traditional joinder, it leaves the parties’ legal rights and duties intact and the rules of decision unchanged.” Shady Grove Orthopedic Associates, P. A. v. Allstate Ins. Co. , 559 U. S. ___, ___ (2010) (plurality opinion) (slip op., at 14).
The second possibility is that every member of the plaintiff class has suffered an Eighth Amendment violation merely by virtue of being a patient in a poorly-run prison system, and the purpose of the class is merely to aggregate all those individually viable claims. This theory has the virtue of being consistent with procedural principles, but at the cost of a gross substantive departure from our case law. Under this theory, each and every prisoner who happens to be a patient in a system that has systemic weaknesses—such as “hir[ing] any doctor who had a license, a pulse and a pair of shoes,” ante , at 10 (internal quotation marks omitted)—has suffered cruel or unusual punishment, even if that person cannot make an individualized showing of mistreatment. Such a theory of the Eighth Amendment is preposterous. And we have said as much in the past: “If … a healthy inmate who had suffered no deprivation of needed medical treatment were able to claim violation of his constitutional right to medical care … simply on the ground that the prison medical facilities were inadequate, the essential distinction between judge and executive would have disappeared: it would have become the function of the courts to assure adequate medical care in prisons.” Lewis v. Casey , 518 U. S. 343, 350 (1996) .
Whether procedurally wrong or substantively wrong, the notion that the plaintiff class can allege an Eighth Amendment violation based on “systemwide deficiencies” is assuredly wrong. It follows that the remedy decreed here is also contrary to law, since the theory of systemic unconstitutionality is central to the plaintiffs’ case. The PLRA requires plaintiffs to establish that the systemwide injunction entered by the District Court was “narrowly drawn” and “extends no further than necessary” to correct “the violation of the Federal right of a particular plaintiff or plaintiffs.” If (as is the case) the only viable constitutional claims consist of individual instances of mistreatment, then a remedy reforming the system as a whole goes far beyond what the statute allows.
It is also worth noting the peculiarity that the vast majority of inmates most generously rewarded by the re-lease order—the 46,000 whose incarceration will be ended—do not form part of any aggrieved class even under the Court’s expansive notion of constitutional violation. Most of them will not be prisoners with medical conditions or severe mental illness; and many will undoubtedly be fine physical specimens who have developed intimidating muscles pumping iron in the prison gym.
Even if I accepted the implausible premise that the plaintiffs have established a systemwide violation of the Eighth Amendment , I would dissent from the Court’s endorsement of a decrowding order. That order is an example of what has become known as a “structural injunction.” As I have previously explained, structural injunctions are radically different from the injunctions traditionally issued by courts of equity, and presumably part of “the judicial Power” conferred on federal courts by Article III:
“The mandatory injunctions issued upon termination of litigation usually required ‘a single simple act.’ H. McClintock, Principles of Equity §15, pp. 32–33 (2d ed. 1948). Indeed, there was a ‘historical prejudice of the court of chancery against rendering decrees which called for more than a single affirmative act.’ Id., §61, at 160. And where specific performance of contracts was sought, it was the categorical rule that no decree would issue that required ongoing supervision. . . . Compliance with these ‘single act’ mandates could, in addition to being simple, be quick; and once it was achieved the contemnor’s relationship with the court came to an end, at least insofar as the subject of the order was concerned. Once the document was turned over or the land conveyed, the litigant’s obligation to the court, and the court’s coercive power over the litigant, ceased… . The court did not engage in any ongoing supervision of the litigant’s conduct, nor did its order continue to regulate its behavior.” Mine Workers v. Bagwell , 512 U. S. 821, 841–842 (1994) ( Scalia, J., concurring).
Structural injunctions depart from that historical practice, turning judges into long-term administrators of complex social institutions such as schools, prisons, and police departments. Indeed, they require judges to play a role essentially indistinguishable from the role ordinarily played by executive officials. Today’s decision not only affirms the structural injunction but vastly expands its use, by holding that an entire system is unconstitutional because it may produce constitutional violations.
The drawbacks of structural injunctions have been described at great length elsewhere. See, e.g., Lewis , supra , at 385–393 (1996) ( Thomas , J., concurring); Missouri v. Jenkins , 515 U. S. 70, 124–133 (1995) ( Thomas , J., concurring); Horowitz, Decreeing Organizational Change: Judicial Supervision of Public Institutions, 1983 Duke L. J. 1265. This case illustrates one of their most pernicious aspects: that they force judges to engage in a form of factfinding-as-policymaking that is outside the traditional judicial role. The factfinding judges traditionally engage in involves the determination of past or present facts based (except for a limited set of materials of which courts may take “judicial notice”) exclusively upon a closed trial record. That is one reason why a district judge’s factual findings are entitled to plain-error review: because having viewed the trial first hand he is in a better position to evaluate the evidence than a judge reviewing a cold record. In a very limited category of cases, judges have also traditionally been called upon to make some predictive judgments: which custody will best serve the interests of the child, for example, or whether a particular one-shot injunction will remedy the plaintiff’s grievance. When a judge manages a structural injunction, however, he will inevitably be required to make very broad empirical predictions necessarily based in large part upon policy views—the sort of predictions regularly made by legis-lators and executive officials, but inappropriate for the Third Branch.
This feature of structural injunctions is superbly illustrated by the District Court’s proceeding concerning the decrowding order’s effect on public safety. The PLRA requires that, before granting “[p]rospective relief in [a] civil action with respect to prison conditions,” a court must “give substantial weight to any adverse impact on public safety or the operation of a criminal justice system caused by the relief.” 18 U. S. C. §3626(a)(1)(A). Here, the District Court discharged that requirement by making the “factual finding” that “the state has available methods by which it could readily reduce the prison population to 137.5% design capacity or less without an adverse impact on public safety or the operation of the criminal justice system.” Juris. Statement App., O. T. 2009, No. 09-416, p. 253a. It found the evidence “clear” that prison overcrowding would “perpetuate a criminogenic prison system that itself threatens public safety,” id., at 186a, and volunteered its opinion that “[t]he population could be reduced even further with the reform of California’s antiquated sentencing policies and other related changes to the laws.” Id., at 253a. It “reject[ed] the testimony that inmates released early from prison would commit additional new crimes,” id., at 200a, finding that “shortening the length of stay through earned credits would give inmates incentives to participate in programming designed to lower recidivism,” id., at 204a, and that “slowing the flow of technical parole violators to prison, thereby substantially reducing the churning of parolees, would by itself improve both the prison and parole systems, and public safety.” Id., at 209a. It found that “the diversion of offenders to community correctional programs has significant beneficial effects on public safety,” id., at 214a, and that “additional rehabilitative programming would result in a significant population reduction while improving public safety,” id., at 216a.
The District Court cast these predictions (and the Court today accepts them) as “factual findings,” made in reliance on the procession of expert witnesses that testified at trial. Because these “findings” have support in the record, it is difficult to reverse them under a plain-error standard of review. Ante , at 38. And given that the District Court devoted nearly 10 days of trial and 70 pages of its opinion to this issue, it is difficult to dispute that the District Court has discharged its statutory obligation to give “substantial weight to any adverse impact on public safety.”
But the idea that the three District Judges in this case relied solely on the credibility of the testifying expert witnesses is fanciful. Of course they were relying largely on their own beliefs about penology and recidivism. And of course different district judges, of different policy views, would have “found” that rehabilitation would not work and that releasing prisoners would increase the crime rate. I am not saying that the District Judges rendered their factual findings in bad faith. I am saying that it is impossible for judges to make “factual findings” without inserting their own policy judgments, when the factual findings are policy judgments. What occurred here is no more judicial factfinding in the ordinary sense than would be the factual findings that deficit spending will not lower the unemployment rate, or that the continued occupation of Iraq will decrease the risk of terrorism. Yet, because they have been branded “factual findings” entitled to deferential review, the policy preferences of three District Judges now govern the operation of California’s penal system.
It is important to recognize that the dressing-up of pol-icy judgments as factual findings is not an error pecu- liar to this case. It is an unavoidable concomitant of institutional-reform litigation. When a district court issues an injunction, it must make a factual assessment of the anticipated consequences of the injunction. And when the injunction undertakes to restructure a social institution, assessing the factual consequences of the injunction is necessarily the sort of predictive judgment that our system of government allocates to other government officials.
But structural injunctions do not simply invite judges to indulge policy preferences. They invite judges to indulge incompetent policy preferences. Three years of law school and familiarity with pertinent Supreme Court precedents give no insight whatsoever into the management of social institutions. Thus, in the proceeding below the District Court determined that constitutionally adequate medical services could be provided if the prison population was 137.5% of design capacity. This was an empirical finding it was utterly unqualified to make. Admittedly, the court did not generate that number entirely on its own; it heard the numbers 130% and 145% bandied about by various witnesses and decided to split the difference. But the ability of judges to spit back or even average-out numbers spoon-fed to them by expert witnesses does not render them competent decisionmakers in areas in which they are otherwise unqualified.
The District Court also relied heavily on the views of the Receiver and Special Master, and those reports play a starring role in the Court’s opinion today. The Court notes that “the Receiver and the Special Master filed reports stating that overcrowding posed a significant barrier to their efforts” and deems those reports “persuasive evidence that, absent a reduction in overcrowding, any remedy might prove unattainable and would at the very least require vast expenditures of resources by the State.” Ante , at 31–32. The use of these reports is even less consonant with the traditional judicial role than the District Court’s reliance on the expert testimony at trial. The latter, even when, as here, it is largely the expression of policy judgments, is at least subject to cross-examination. Relying on the un-cross-examined findings of an investigator, sent into the field to prepare a factual report and give suggestions on how to improve the prison system, bears no resemblance to ordinary judicial decisionmaking. It is true that the PLRA contemplates the appointment of Special Masters (although not Receivers), but Special Masters are authorized only to “conduct hearings and prepare proposed findings of fact” and “assist in the development of remedial plans,” 18 U. S. C. §3626(f)(6). This does not authorize them to make factual findings (unconnected to hearings) that are given seemingly wholesale deference. Neither the Receiver nor the Special Master was selected by California to run its prisons, and the fact that they may be experts in the field of prison reform does not justify the judicial imposition of their perspectives on the state.
My general concerns associated with judges’ running social institutions are magnified when they run prison systems, and doubly magnified when they force prison officials to release convicted criminals. As we have previously recognized:
“[C]ourts are ill equipped to deal with the increasingly urgent problems of prison administration and re- form… . [T]he problems of prisons in America are complex and intractable, and, more to the point, they are not readily susceptible of resolution by decree… . Running a prison is an inordinately difficult undertaking that requires expertise, planning, and the com-mitment of resources, all of which are peculiarly within the province of the legislative and executive branches of government. Prison is, moreover, a task that has been committed to the responsibility of those branches, and separation of powers concerns counsel a policy of judicial restraint. Where a state penal system is involved, federal courts have … additional reason to accord deference to the appropriate prison authorities.” Turner v. Safley , 482 U. S. 78, 84–85 (1987) (internal quotation marks omitted).
These principles apply doubly to a prisoner-release order. As the author of today’s opinion explained earlier this Term, granting a writ of habeas corpus “ ‘disturbs the State’s significant interest in repose for concluded litigation, denies society the right to punish some admitted offenders, and intrudes on state sovereignty to a degree matched by few exercises of federal judicial authority.’ ” Harrington v. Richter , 562 U. S. ___, ___ (2011) (slip op., at 13) (quoting Harris v. Reed , 489 U. S. 255, 282 (1989) ( Kennedy, J., dissenting)). Recognizing that habeas relief must be granted sparingly, we have reversed the Ninth Circuit’s erroneous grant of habeas relief to individual California prisoners four times this Term alone. Cullen<linebreak> v. Pinholster , 563 U. S. ___ (2011); Felkner v. Jackson , 562 U. S. ___ (2011) (per curiam); Swarthout v. Cooke , 562 U. S. ___ (2011) (per curiam); Harrington , supra . And yet here, the Court affirms an order granting the functional equivalent of 46,000 writs of habeas corpus, based on its paean to courts’ “substantial flexibility when making these judgments.” Ante , at 41. It seems that the Court’s respect for state sovereignty has vanished in the case where it most matters.
The Court’s opinion includes a bizarre coda noting that “[t]he State may wish to move for modification of the three-judge court’s order to extend the deadline for the required reduction to five years.” Ante , at 46–47. The Dis-trict Court, it says, “may grant such a request provided that the State satisfies necessary and appropriate preconditions designed to ensure the measures are taken to implement the plan without undue delay”; and it gives vague suggestions of what these preconditions “may include,” such as “interim benchmarks.” Ante , at 47. It also invites the District Court to “consider whether it is appropriate to order the State to begin without delay to develop a system to identify prisoners who are unlikely to reoffend,” and informs the State that it “should devise systems to select those prisoners least likely to jeopardize public safety.” Ibid. (What a good idea!)
The legal effect of this passage is unclear—I suspect intentionally so. If it is nothing but a polite remainder to the State and to the District Court that the injunction is subject to modification, then it is entirely unnecessary. As both the State and the District Court are undoubtedly aware, a party is always entitled to move to modify an equitable decree, and the PLRA contains an express provision authorizing District Courts to modify or terminate prison injunctions. See 18 U. S. C. §3626(b).
I suspect, however, that this passage is a warning shot across the bow, telling the District Court that it had better modify the injunction if the State requests what we invite it to request. Such a warning, if successful, would achieve the benefit of a marginal reduction in the inevitable murders, robberies, and rapes to be committed by the released inmates. But it would achieve that at the expense of in-tellectual bankruptcy, as the Court’s “warning” is en- tirely alien to ordinary principles of appellate review of injunctions. When a party moves for modification of an injunction, the district court is entitled to rule on that motion first, subject to review for abuse of discretion if it declines to modify the order. Horne v. Flores , 557 U. S. ___, ___, ___ (2009) (slip op., at 10, 20). Moreover, when a district court enters a new decree with new benchmarks, the selection of those benchmarks is also reviewed under a deferential, abuse-of-discretion standard of review—a point the Court appears to recognize. Ante , at 45. Appellate courts are not supposed to “affirm” injunctions while preemptively noting that the State “may” request, and the District Court “may” grant, a request to extend the State’s deadline to release prisoners by three years based on some suggestions on what appropriate preconditions for such a modification “may” include.
Of course what is really happening here is that the Court, overcome by common sense, disapproves of the results reached by the District Court, but cannot remedy them (it thinks) by applying ordinary standards of appellate review. It has therefore selected a solution unknown in our legal system: A deliberately ambiguous set of suggestions on how to modify the injunction, just deferential enough so that it can say with a straight face that it is “affirming,” just stern enough to put the District Court on notice that it will likely get reversed if it does not follow them. In doing this, the Court has aggrandized itself, grasping authority that appellate courts are not supposed to have, and using it to enact a compromise solution with no legal basis other than the Court’s say-so. That we are driven to engage in these extralegal activities should be a sign that the entire project of permitting district courts to run prison systems is misbegotten.
But perhaps I am being too unkind. The Court, or at least a majority of the Court’s majority, must be aware that the judges of the District Court are likely to call its bluff, since they know full well it cannot possibly be an abuse of discretion to refuse to accept the State’s proposed modifications in an injunction that has just been approved (affirmed) in its present form. An injunction, after all, does not have to be perfect; only good enough for government work, which the Court today says this is. So perhaps the coda is nothing more than a ceremonial washing of the hands—making it clear for all to see, that if the terrible things sure to happen as a consequence of this outrageous order do happen, they will be none of this Court’s responsibility. After all, did we not want, and indeed even suggest, something better?
In view of the incoherence of the Eighth Amendment claim at the core of this case, the nonjudicial features of institutional reform litigation that this case exemplifies, and the unique concerns associated with mass prisoner releases, I do not believe this Court can affirm this injunction. I will state my approach briefly: In my view, a court may not order a prisoner’s release unless it determines that the prisoner is suffering from a violation of his constitutional rights, and that his release, and no other relief, will remedy that violation. Thus, if the court determines that a particular prisoner is being denied constitutionally required medical treatment, and the release of that prisoner (and no other remedy) would enable him to obtain medical treatment, then the court can order his release; but a court may not order the release of prisoners who have suffered no violations of their constitutional rights, merely to make it less likely that that will happen to them in the future.
This view follows from the PLRA’s text that I discussed at the outset, 18 U. S. C. §3626(a)(1)(A). “[N]arrowly drawn” means that the relief applies only to the “particular [prisoner] or [prisoners]” whose constitutional rights are violated; “extends no further than necessary” means that prisoners whose rights are not violated will not obtain relief; and “least intrusive means necessary to correct the violation of the Federal right” means that no other relief is available. * *
I acknowledge that this reading of the PLRA would se-verely limit the circumstances under which a court could issue structural injunctions to remedy allegedly unconstitutional prison conditions, although it would not eliminate them entirely. If, for instance, a class representing all prisoners in a particular institution alleged that the temperature in their cells was so cold as to violate the Eighth Amendment , or that they were deprived of all exercise time, a court could enter a prisonwide injunction ordering that the temperature be raised or exercise time be provided. Still, my approach may invite the objection that the PLRA appears to contemplate structural injunctions in general and mass prisoner-release orders in particular. The statute requires courts to “give substantial weight to any adverse impact on public safety or the operation of a criminal justice system caused by the relief” and authorizes them to appoint Special Masters, §3626 (a)(1)(A), (f), provisions that seem to presuppose the possibility of a structural remedy. It also sets forth criteria under which courts may issue orders that have “the purpose or effect of reducing or limiting the prisoner population,” §3626(g)(4).
I do not believe that objection carries the day. In addition to imposing numerous limitations on the ability of district courts to order injunctive relief with respect to prison conditions, the PLRA states that “[n]othing in this section shall be construed to … repeal or detract from otherwise applicable limitations on the remedial powers of the courts.” §3626(a)(1)(C). The PLRA is therefore best understood as an attempt to constrain the discretion of courts issuing structural injunctions—not as a mandate for their use. For the reasons I have outlined, structural injunctions, especially prisoner-release orders, raise grave separation-of-powers concerns and veer significantly from the historical role and institutional capability of courts. It is appropriate to construe the PLRA so as to constrain courts from entering injunctive relief that would exceed that role and capability.
* * *
The District Court’s order that California release 46,000 prisoners extends “further than necessary to correct the violation of the Federal right of a particular plaintiff or plaintiffs” who have been denied needed medical care. 18 U. S. C. §3626(a)(1)(A). It is accordingly forbidden by the PLRA—besides defying all sound conception of the proper role of judges.
* * Any doubt on this last score, at least as far as prisoner-release orders are concerned, is eliminated by §3626(a)(3)(E) of the statute, which provides that to enter a prisoner-release order the court must find “by clear and convincing evidence that— (i) crowding is the primary cause of the violation of a Federal right; and (ii) no other relief will remedy the violation of the Federal right.”