ILANA DIAMOND ROVNER, Circuit Judge, with whom EVANS, DIANE P. WOOD, and WILLIAMS, Circuit Judges, join, dissenting.

In concluding that this PLRA provision satisfies rational relationship scrutiny, the majority as well as other courts have held that Congress could rationally believe the PLRA provisions furthered its interest in deterring frivolous or trivial litigation. The journey from the PLRA fee restrictions to the reduction in trivial or frivolous filings, however, requires so many leaps, many ridiculous, as to destroy any semblance of rationality. To believe that the fee restriction impacts the filing decision of pro se prisoners, Congress would have to believe that (1) prisoners would be inclined to bring frivolous or trivial lawsuits and (2) those prisoners would not be deterred from filing such suits by the likelihood that their own damages would be minimal or non-existent but (3) those prisoners would be deterred from filing those suits by the prospect that the fees available to their as-then non-existent attorneys were restricted under the PLRA even though (4) those attorneys would not have been entitled to fees for such suits even absent the PLRA and even though (5) the chances of those prisoners ever obtaining counsel was practically nil, given that only 1% of prisoner civil rights cases involve private attorneys, and another 4% involve appointed counsel (who presumably would not be appointed for our hypothetical trivial/flivolous cases). See Roger A. Hanson Henry W.K. Daley, Challenging the Conditions of Prison and Jails: A Report on Section 1983 Litigation 21-22 (1995); see, e.g., McKeever v. Israel, 689 F.2d 1315, 1320 (7th Cir. 1982) (in determining whether to appoint counsel, "the threshold question is whether the claim is of sufficient merit"). Such a belief is fanciful, not rational. In the alternative, to believe that the fee restriction impacts the decision of private attorneys bringing suits on behalf of such prisoners, Congress would have to believe that (1) there are private attorneys who are willing to bring suits on behalf of prisoners despite the difficult standards of proof and the absence of traditional damages such as lost wages and future earnings and (2) those attorneys would file a trivial or frivolous lawsuit even though fees for such lawsuits are not available under § 1988 but (3) those same attorneys would decide not to file the law suit once fees were merely restricted under the PLRA. This sequence is made even more ridiculous considering that only 1% of all prisoner civil rights litigation involves private attorneys, including groups such as the ACLU, but yet we are asked to believe that Congress rationally could think those attorneys are choosing to bring the frivolous or trivial lawsuits and would be deterred from that course by the fee restriction under the PLRA but not the fee prohibition under § 1988. If this connection is a rational one, then this test truly has lost all meaning. The only impact this provision will have is on the meritorious prisoner actions, rendering it even more difficult for prisoners to adequately present these meritorious claims to the courts, and more difficult for the court to find counsel willing to accept appointments to represent prisoners in such cases. The government does not purport to have a legitimate interest in deterring prisoners from filing meritorious suits, nor could it, and therefore the provision is unconstitutional.

Rather than apply traditional equal protection analysis to the classification at issue here, as did all of the other circuits to consider this claim, the plurality opinion engages in a free-ranging discussion of myriad unrelated statutes under the mantra of answering the question "equal compared to what?" But the classification at issue here is not a mystery — the singling out of prisoners from the § 1988 fee structure available to all other civil rights litigants — and the sole issue in this case is whether that classification is rationally related to a legitimate governmental goal. Rather than addressing that question, the plurality examines a host of unrelated policy issues including whether the fees available in civil rights actions are too generous as compared with those available in other causes of actions and whether prisoners are less worthy of benefits than other groups. Those are interesting issues for legislators or political candidates, but they are mere detours here, irrelevant to the equal protection claim before us. The plurality approach stands in stark contrast to that of the other circuits to consider the issue, which at least have engaged in a straightforward analysis of the classification and applied traditional equal protection analysis to the issue. The plurality opinion travels so far afield from equal protection jurisprudence in its opinion that before we can examine the rationality of the classification at issue here, we are compelled to clarify what is not at issue in this case.

I.

First, no one claims that the right of access to the courts is a right of subsidized access. In demonstrating that there is no constitutional right to attorney's fees, the plurality opinion lists a number of constitutional rights for which there is no right of funding, but it is a point that is undisputed and therefore the discussion is gratuitous. This is not a case concerning whether prisoners, or any other persons, have a constitutional right to attorney's fees in civil litigation. The only issue here is whether, in granting fees to prevailing plaintiff's in civil rights actions, Congress may constitutionally single out one group of persons for differing treatment, and all parties (as well as the other circuits to consider the issue) agree that the rational relationship test is the appropriate one for analyzing that question.

Second, this case is not a forum to analyze whether attorney's fees should be available in civil rights litigation as a whole, whether such litigation is more important than other types of litigation, or whether prisoners as a whole are a group less worthy of fees than "free persons." Although such a discussion in the current climate may well foster a visceral reaction of outrage (why should veterans have a lesser right to fees than prisoners'), it is an illusory comparison unrelated to the equal protection issue, and it foments the type of resentment towards a disfavored group that the equal protection clause is designed to address. It is difficult to understand why so much of the plurality opinion is devoted to a comparison of the fees available to prisoners in civil rights suits with the fees available to persons in miscellaneous other causes of actions. There is no constitutional requirement that fees be equal across causes of actions, nor is there any constitutional mandate that fees must be lower if a court deems the cause of action "less important" or the litigants "less worthy." Such a wide-ranging approach to the equal protection clause would transform courts into a "super-legislature." See Heller v. Doe by Doe, 509 U.S. 312, 319, 113 S.Ct. 2637, 125 L.Ed.2d 257 (1993) (rational basis review in equal protection does not "authorize `the judiciary [to] sit as a superlegislature to judge the wisdom or desirability of legislative policy determinations'"). We are not the proper body to determine whether the spectrum of benefits afforded prisoners is more than those given veterans, or whether prisoners or any other group of persons are less deserving of benefits than veterans and therefore should get less. The end result of such an approach would be one that I suspect the majority would not endorse. Should we compare benefits afforded the elderly with those given to impoverished children? What about those given to senior citizens regardless of income as opposed to those provided to the working poor, or to the unemployed? Are we to judge who is more worthy? Of course not. The court ventures into that dangerous territory in comparing the fees available to prisoners with those provided under other statutes, a comparison that has no basis in equal protection jurisprudence.

Moreover, the comparison is an illusory one for another reason. The PLRA applies to all civil rights actions by prisoners. This particular case happens to address the denial of appropriate medical care, which leads the plurality into a comparison with other unrelated statutes relating to medical care — including medical malpractice tort law, the FTCA, and Medicaid. But the case could have just as easily involved an entirely different field of law. Under the plurality's approach, if the civil rights claim concerned the conditions under which the prisoner was forced to live, we would examine all habitability provisions, or landlord-tenant laws. The logical extension of that approach would be that the constitutionality of the PLRA provision might well vary depending on the nature of the underlying lawsuit. The equal protection clause requires no such free-ranging examination of the fairness or wisdom of a provision in light of analogous but unrelated statutes. In fact, after raising the false comparisons, the plurality explicitly recognizes that the numbers upon which it bases much of its argument are "not directly comparable." Op. at 590.

The plurality in the end does not appear to be questioning the availability of attorney's fees to prisoners as much as the allowance of prisoner suits under § 1983 at all. Much of its analysis challenges why prisoners are allowed to litigate at all for deliberate indifference to their serious medical needs, comparing prisoners' litigation options to those of veterans and medicaid recipients among others. But that is not the subject of the PLRA provisions before us. We are concerned here only with the restriction on attorney's fees under the PLRA, not the wisdom of the varying litigation options among statutes, and therefore this discussion has no place in an equal protection analysis of the PLRA provision at issue here.

As we stated, the question "equal compared to what" is not answered by comparing prisoners in civil rights actions to litigants in unrelated causes of actions. For nearly all of those unrelated causes of action (i.e. malpractice tort law, the FTCA, etc.), all claimants filing suit under those actions are treated identically. Although there may be variations in the availability of fees among different causes of actions, there is no such discrepancy within a cause of action, with the exception of the IDEA which we will discuss shortly. The PLRA, however, creates just such a classification within a cause of action. Section 1988 grants attorney's fees to prevailing plaintiff's in civil rights litigation. The PLRA specifically targets those litigants, and creates a sub-group (prisoners) that will not have the same access to fees, stating that "[i]n any action brought by a prisoner . . . in which attorney's fees are authorized under section 1988 [footnote omitted] of this title, such fees shall not be awarded except that. . . ." It is that classification between classes of individuals under § 1988 that must pass equal protection scrutiny.

An example may illustrate the fallacy of the plurality's approach. Suppose instead of limiting fees for prisoners, the PLRA reduced the availability of fees under § 1988 for all blue-eyed litigants. Under the plurality's reasoning, there would be no equal protection problem with that statute. The same arguments used to justify the restriction for prisoners would apply to blue-eyed litigants: blue-eyed civil rights litigants would still be "decidedly better off' under those fee restrictions than veterans under the FTCA or persons claiming benefits under the Social Security program; litigants in those other causes of action manage to obtain attorney representation so there is no reason to believe that equal fees are necessary for the blue-eyed civil rights litigants to do the same; civil rights litigation is not nearly as important as collateral attacks on a conviction, and there are less fees available there; there is no constitutional right to attorney's fees; and finally, other statutes have widely divergent fee structures, and "what this means is that all of these different systems could be thought rational solutions." Yet most rational people would balk at the suggestion that Congress could, consistent with equal protection, single out persons with a certain eye color and deny them the same attorney's fees as all other litigants pursuing the same constitutional claims. That is because the detour into other causes of actions ignores the proper focus, which is on the classification itself viewed in the context of the statute authorizing fees for that particular cause of action. The classification must be rationally related to the legitimate governmental goal. That analysis is not even addressed until well into the plurality opinion, but it is the only relevant one for purposes of equal protection. The plurality's polemic may have (intuitive) appeal only because it exploits our own preconceptions about prisoners as a class and their worthiness — in other words, because it highlights why they are a disfavored group. But that tendency to act arbitrarily against a disfavored group is precisely the reason why the equal protection clause is necessary.

Only one of the examples offered by the plurality, that of the District of Columbia claimants under the IDEA, actually addresses a sub-group of litigants who were singled out for reduced fees as compared to others asserting the same cause of action. The D.C. Circuit in Calloway upheld that disparity only after determining that the classification was rationally related to the legitimate goal of addressing an acute funding need in the District. 216 F.3d 1. Notably, the court did not uphold the classification under the theory proposed by the majority that all fee structures vary and therefore all could be considered rational, nor did it uphold it because the fees that remained available were better than those afforded other groups such as veterans. Indeed, Calloway provides no support for the theories presented in this majority opinion. A glance at Calloway is revealing in its contrast to the present case.

Calloway considered the constitutionality of a provision that limited the fees that the District of Columbia could pay to attorneys for prevailing parties in IDEA lawsuits. The reason for singling out District of Columbia litigants from all other IDEA litigants was clear from the record, which demonstrated that in one year alone, "although the DCPS [District of Columbia Public Schools] served less than two-thousandths of one percent of the nation's disabled students, over forty-five percent of requests for due process hearings nationwide were made in D.C." Id. at 4. Because attorney's fees were available under the IDEA, legal representation in D.C. had grown from an obscure niche into a booming, lucrative industry, costing the DCPS over $10 million in attorney's fees in fiscal year 1998. Id. The Secretary of Education had determined that immediate compliance by the DCPS with the requirements of the IDEA was not feasible because the root causes of its failures were so extensive; the Secretary had thus entered into a compliance agreement with DCPS mandating full compliance within three years. Id. Against this unusual background, Congress acted to reduce the amount of attorney's fees available for IDEA actions in D.C. The D.C. Circuit considered whether the classification at issue there — the singling out of D.C. litigants from other IDEA litigants — was rationally related to a legitimate governmental goal. The government argued that "in view of DCPS's manifest inability to meet its obligations under IDEA, Congress could rationally have concluded that `it was more important for the District to spend its funds on remedying those systemic defects and providing primary services rather than upon litigation fees.'" Id. at 8. The D.C. Circuit held that Congress could rationally believe that limiting payments to attorneys would leave more funds available for direct services, and that assisting disabled children by allocating funds towards primary educational services rather than attorney's fees was a legitimate governmental goal. Id. at 8-9. Accordingly, the provision survived equal protection scrutiny.

The plurality opinion establishes no similar connection between reduced attorney's fees in prisoner civil rights litigation and any legitimate governmental goal. Unlike Calloway, there is no booming, lucrative business in the prisoner segment of civil rights litigation. With only 1% of all prisoner civil rights cases even involving attorneys not appointed by the court, that can not possibly be identified as a fee drain out of proportion to the other litigants under the civil rights statutes. It is that connection between attorney's fees and a legitimate governmental goal that the plurality never makes, choosing instead to devote much of the opinion to a condemnation of prisoner civil rights litigation generally (without regard to attorney representation or fees) or decrying the extent of fees available in civil rights actions generally as compared with other causes of actions that the majority appears to deem more significant. The Equal Protection Clause, however, addresses the fit between the particular classification and the legitimate governmental goal, and that is lacking here.

In a late effort to tie its discussion to traditional equal protection analysis, the plurality declares that the other statutes with lesser attorney's fees are relevant comparisons because litigants under those statutes do not have difficulties obtaining counsel, and therefore Congress could assume that prisoners under a more favorable statutory structure would not either. That comparison is so superficial as to be meaningless. First, as has been mentioned, the focus on medical cases is misplaced, because the PLRA limits fees in all types of prisoner civil rights actions. Moreover, damages cannot be compared with any meaning in prisoner versus nonprisoner litigation, because the absence of earning capacity in the former necessarily skews the damages. This case is a good example of the discrepancy. Diagnosed with end stage liver disease, Johnson lapsed into a coma multiple times. In February 1996, he was examined by Dr. Alexandru Musat, medical director of the Liver Transplantation Program at the University of Wisconsin Hospital, who recommended that Johnson should be evaluated to determine whether he was an appropriate candidate for a liver transplant, as "the only viable option for resolution of his symptoms and liver failure." Without a liver transplant, Dr. Musat estimated that Johnson's life expectancy would be in the range of three to four years.

The defendant Dr. George Daley, the medical director of the Bureau of Correctional Health Services for the Wisconsin Department of Corrections, refused to authorize that evaluation. On a number of subsequent occasions, Dr. Daley denied written requests from two prison doctors seeking authorization for the treatment recommended by Dr. Musat. Johnson then filed suit pursuant to 42 U.S.C. § 1983 alleging that Dr. Daley denied him adequate medical care for his liver disease in violation of the Eighth Amendment. In July 1998, the district court dismissed that action without prejudice for failure to exhaust administrative remedies, but the complaint was reinstated in December 1998, after Johnson satisfied the exhaustion requirement. In April 1999, Dr. Daley finally approved the request to have Johnson evaluated by the University of Wisconsin for a liver transplant, and in June 1999, the University of Wisconsin Hospital, with the approval of Dr. Daley, placed Johnson's name on the list as a, potential recipient for a donated liver. The case proceeded to trial on the issues of whether the initial denials of the evaluation constituted deliberate indifference to Johnson's serious medical needs and whether the three-year delay in evaluation and placement on the transplant list caused him any harm.

Despite the length of the delay and the clear risk to Johnson's life, recognized even by the prison doctors, Johnson was awarded only $10,000 in compensatory damages and $30,000 in punitive damages, amounts that undoubtedly would have been much higher were he not a pisoner. Even the plurality opinion reflects the devaluation of a prisoner's life, stating dismissively that "he appears to have suffered no long-term injury from the delay," but that "[n]onetheless, a jury agreed with Johnson . . . and awarded him . . . damages." I doubt that the plurality would be so cavalier about a three-year delay in being placed on a transplant list if it were they or a loved one awaiting a life-saving procedure, and the plurality opinion fails to mention that but for the lawsuit, Johnson would probably still not be on the transplant list.

The reality is that both the deep-seated societal antipathy towards prisoners as a class as well as the absence of lost wages or future earnings damages ensures that damages in prisoner cases will nearly always be minimal. See Roger A. Hanson Henry W.K. Daley, Challenging the Conditions of Prison and Jails: A Repoil on Section 1983 Litigation 37 (1995) (recognizing that prisoner settlements and verdicts are smaller because wages are nonexistent and a substantial component of compensatory damages is either lost wages or foregone earnings). More fundamentally, the comparison with the other statutes is flawed in its extrapolation of the experience under other statutes to this context. The plurality concludes that if attorneys are widely available under statutes with less liberal fee provisions, Congress could conclude they would be widely available under the PLRA. But even pre-PLRA, attorneys were rarely available to prisoners. Prisoners are represented by attorneys in only 4% of all prisoner litigation, and the vast majority of those attorneys are court appointed. Because attorneys were scarce in prisoner civil rights litigation even before the PLRA reduced the fees, the comparison to other statutes is meaningless.

The plurality opinion raises a host of other questionable arguments, not the least of which is that all prisoners are liars and all of their cases lost are lost because they lied. It is difficult to imagine any other group of persons for whom such sweeping (and unsupported) generalizations would be tolerated, let alone given credence, in a judicial opinion. Nevertheless, the generalizations do not further the plurality's argument. In fact, many of those alleged prisoner traits posited by the plurality would demonstrate that their filing decisions are entirely unaffected by attorney's fees. There is, for instance, no reason to believe that prisoners motivated to file lawsuits to get a day away from prison, to pass the time, or to harass prison staff, will be moved to abstain from such conduct by the prospect that their hypothetical future attorney will receive less in fees if the suit is successful. But that is precisely the basis of the rationality analysis offered by the plurality. The plurality further uses these preconceptions about character to engage in an even more fanciful calculation of the hypothetical effects of the PLRA on prisoner behavior. For instance, assuming all prisoners are liars, the plurality postulates that a frivolous case often will not be recognized as such until the trial, but that ignores the reality that the vast majority of prisoner cases are resolved without an evidentiary hearing. See Roger A. Hanson Henry W.K. Daley, Challenging the Conditions of Prison and Jails: A Report on Section 1983 Litigation 22 (1995) (noting that evidentiary hearings are not held in 97% of cases). The calculations are in large part based on unsupported assumptions and unrecognized factors. For instance, from the statistic that prisoners lose 90% of all fully tried cases, the majority then assumes that 9 out of 10 prisoners making claims are lying, and only 1 out of 10 guards is lying. That is unsupportable. Only 3% of cases reach evidentiary hearings or trials. Id. In the 97% that do not, courts assume the prisoners are truthful in their allegations but find that prisoners nonetheless lack the support necessary to satisfy the relatively demanding prerequisites to relief (i.e., in medical cases, prisoners need to demonstrate not mere malpractice, but deliberate indifference to a serious medical need). For the overwhelming majority of cases, then, there is no basis to assume that the case is lost because the prisoner is lying. Likewise, of the cases that go to trial, there is no reason to believe that the lack of success establishes that the prisoner was lying as opposed to the possibility that the prisoner lacked the corroboration to support a claim or simply could not meet the high burden, but in any case 90% of 3% is only 2.7%. All of the expected recovery calculations flow from that assumption.

The potential reduction in litigation that the majority portends as a result of these PLRA provisions is similarly baseless. For instance, the majority concludes that the PLRA fee restrictions will reduce the percentage of prisoners filing cases, but does not consider whether that decrease, by making it less risky for guards to engage in unconstitutional behavior, will increase such behavior and therefore offset some of that alleged "gain." In any case, the calculations are far removed from the issue here and the reality of filing decisions in the prison context. Furthermore, the majority declares that "even if 80% of prisoners are insensible to changes in the litigation process, different behavior by the other 20% would be welcome." Given that only 4% of prisoners will ever be represented by counsel, the 20% prediction is baseless. When we further consider that the vast majority of those 4% involve appointed counsel and therefore are cases that the court does not consider trivial, it becomes clear that the cases potentially deterred are not the trivial ones in which the government claims an interest, but the meritorious ones in which it does not. McKeever, 689 F.2d at 1320 (in determining whether to appoint counsel, "the threshold question is whether the claim is of sufficient merit"). No amount of statistical shell games can rationally establish otherwise.

We turn, then, to the only issue presented here — the connection between the classification and the legitimate governmental interest. A proper application of the rational relationship test renders the PLRA fee restrictions unconstitutional.

II.

We begin our analysis with a brief explanation of the genesis of the PLRA provisions. The PLRA was enacted in response to a veritable flood of prisoner litigation, most of it without merit, that clogged up the dockets of federal courts and drained resources away from meritorious cases. The magnitude of the problem was apparent. See, e.g., Roller v. Gunn, 107 F.3d 227, 230 (4th Cir. 1997) ("In 1995, prisoners brought over 25% of the civil cases filed in the federal courts. In this circuit alone, in forma pauperis (IFP) filings accounted for almost half of the court's 1995 caseload and prisoners were responsible for 75% of those filings.") (citations omitted); Boivin v. Black, 225 F.3d 36, 41 (1st Cir. 2000). In addressing this problem, Congress recognized that the unique circumstances of prison life provided a breeding ground for frivolous litigation. Inmates have a great deal of time on their hands, and no disincentive to spend it filing suits. Although a court appearance is a nuisance and often a real hardship on a non-incarcerated person, who must arrange for time off from work or child care, the inmate has no such concerns. To the contrary, a trip to the courtroom is a benefit for an inmate who is otherwise confined to a cell. In addition, the availability of IFP status, as well as the provision of all the inmate's daily needs in prison, removed any financial impediment to litigation. As a result, the type of cost-benefit assessment in which a non-prisoner might engage, to determine whether the constitutional issue was worth the effort and cost of pursuing federal court relief, was artificially skewed in the context of the prison suits. See Hadix v. Johnson, 230 F.3d 840, 844 (6th Cir. 2000); Boivin, 225 F.3d at 44; Tucker v. G. Branker, 142 F.3d 1294, 1300 (D.C. Cir. 1998); Mitchell v. Farcass, 112 F.3d 1483, 1488-89 (11th Cir. 1997); Roller, 107 F.3d at 234. Congress responded to that problem by enacting the PLRA in an attempt to redress the epidemic of frivolous prisoner litigation.

We have already upheld a number of PLRA provisions as constitutional. See, e.g., Lewis v. Sullivan, 279 F.3d 526 (7th Cir. 2002) (holding constitutional PLRA requirement that prisoners prepay filing fees if they have a history of frivolous litigation except where under imminent danger of serious physical injury); Zehner v. Trigg, 133 F.3d 459 (7th Cir. 1997) (holding constitutional PLRA limitation that forbids recovery for mental or emotional damages without a prior showing of physical injury). Although we have upheld those provisions against equal protection challenges, each PLRA provision must be examined on its own merits.

We are concerned in this case only with §§ 1997e(d)(2) (3), which provide that in cases brought by an inmate in which fees are authorized under § 1988, the award of attorney's fees is limited to 150% of the amount of judgment, § 1997e(d)(2), and the hourly rate payable to a prisoner's attorney is restricted to 150% of the hourly rate allowed for court-appointed counsel in criminal cases under the Criminal Justice et ( 18 U.S.C. § 3006A), § 1997e(d)(3) (collectively referred to as "the PLRA fee restrictions").

Because those PLRA fee restrictions apply only to cases in which fees are authorized under § 1988, in order to assess the equal protection challenge we first must place those provisions in the context of the fees otherwise available under § 1988. See § 1997e(d); Martin v. Hadix, 527 U.S. 343, 363-64, 119 S.Ct. 1998, 144 L.Ed.2d 347 (1999) (Scalia, J., dissenting) ("[i]n reality . . . the PLRA simply revises the fees provided for by § 1988, and it seems that the underlying purpose of that provision must govern its amendment as well — which purpose was to provide an appropriate incentive for lawyers to work on (among other civil rights cases) prisoner suits."); Heller v. Doe by Doe, 509 U.S. 312, 321, 113 S.Ct. 2637, 125 L.Ed.2d 257 (1993) ("even the standard of rationality as we so often have defined it must find some footing in the realities of the subject addressed by the legislation"). See also Romer v. Evans, 517 U.S. 620, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996) (analyzing amendment in light of pre-existing law to determine that rational relationship test was violated); Lindsey v. Normet, 405 U.S. 56, 78, 92 S.Ct. 862, 31 L.Ed.2d 36 (1972) (analyzing the statutory double-bond requirement in the context of the pre-existing state statutes); Reitman v. Mulkey, 387 U.S. 369, 373, 87 S.Ct. 1627, 18 L.Ed.2d 830 (1967) (in equal protection challenge to state constitutional provision, court should consider its "`immediate objective,' its `ultimate effect' and its historical context and the conditions existing prior to its enactment."); In enacting § 1988, Congress recognized that civil rights actions for damages are distinct from private tort suits that benefit only the individual plaintiff's. Unlike ordinary tort litigation, civil rights suits "vindicate important civil and constitutional rights that cannot be valued solely in monetary terms." City of Riverside v. Rivera, 477 U.S. 561, 574, 106 S.Ct. 2686, 91 L.Ed.2d 466 (1986). Congress determined that many victims of civil rights violation were unable to obtain representation in the private market because of high attorney fees, and therefore § 1988 provides for payment of a reasonable fee-that is, "a fee large enough to induce competent counsel to handle the plaintiff's case, but no larger." Simpson v. Sheahan, 104 F.3d 998, 1002 (7th Cir. 1997).

Consistent with that purpose, § 1988 allows for the award of fees only to "prevailing parties." In order to be considered a prevailing party, the plaintiff must have obtained some judicially sanctioned relief on the merits of the claim. Specifically, a plaintiff prevails when actual relief on the merits of the claim, provided by a court judgment or consent decree, materially alters the legal relationship between the parties by modifying the defendant's behavior in a way that directly benefits the plaintiffc. Buckhannon Bd. and Care Home, Inc. v. West Virginia Dept. of Health and Human Resources, 532 U.S. 598, 604, 121 S.Ct. 1835, 149 L.Ed.2d 855 (2001); Farrar v. Hobby, 506 U.S. 103, 111-112, 113 S.Ct. 566, 121 L.Ed.2d 494 (1992); Simpson, 104 F.3d at 1001. Once that hurdle is surmounted, a court may award fees but only in an amount that is reasonable in light of the degree of overall success. Farrar, 506 U.S. at 114, 113 S.Ct. 566. Where a plaintiff is a prevailing party but the success is merely technical or de minimis, the presumption is that no fees should be awarded. See id. at 115, 113 S.Ct. 566 (where the plaintiff recovers only nominal damages "the only reasonable fee is usually no fee at all"); Cole v. Wodziak, 169 F.3d 486, 488 (7th Cir. 1999) ("a paltry jury award . . . implies that the only reasonable fee is zero"); Monticello School Dist. No. 25 v. George L. on Behalf of Brock L., 102 F.3d 895, 907 (7th Cir. 1996) ("when a plaintiff's success is simply technical or de minimis, no fees may be awarded, even if the plaintiff has succeeded on an issue in the litigation and may thus be technically a `prevailing party'"); Maul v. Constan, 23 F.3d 143, 145 147 (7th Cir. 1994) (award of attorney's fees is inappropriate where victory is de minimis).

The ultimate question in entitlement to fees under § 1988, therefore, is whether the plaintiffs victory is significant or merely de minimis. Maul 23 F.3d at 145. In recognition that "[n]ominal relief does not necessarily a nominal victory make," Farrar, 506 U.S. at 121, 113 S.Ct. 566 (O'Connor, J. concurring), courts may award fees in cases of minimal damages only if other factors establish that the victory, although monetarily small, is actually a significant one. In maldng that assessment, we look to the three factors set forth in Justice O'Connor's concurrence in Farrar. "(1) the difference between the judgment recovered and the recovery sought; (2) the significance of the legal issue on which the plaintiff prevailed; and (3) the public purpose of the litigation." Farrar, 506 U.S. at 122, 113 S.Ct. 566; Simpson, 104 F.3d at 1001; Briggs v. Marshall, 93 F.3d 355, 361 (7th Cir. 1996). Where those factors establish that the victory is not de minimis but is instead significant, fees may be available under § 1988. If those factors are not met, then the plaintiff achieved only a technical or de minimis victory and fees are unavailable under § 1988. Of course, if a case is fiivolous, no fees can be awarded under § 1988, and in fact plaintiff and plaintiff's counsel can be sanctioned for the pursuit of such a suit. Farrar, 506 U.S. at 111-12, 113 S.Ct. 566.

III.

With that background, I turn to the constitutionality of the PLRA fee restrictions. The parties agree that the appropriate level of review is the rational relationship test because the challenged provisions do not impact fundamental rights and the prisoners do not constitute a suspect class which would have triggered a higher degree of scrutiny. The issue, therefore, is whether those PLRA fee restrictions are rationally related to a legitimate governmental interest. Although the district courts have diverged on the issue, the federal appellate courts that have considered the rationality of the PLRA fee restrictions have upheld them or split on the question. The Third Circuit sitting en banc split evenly on the issue, which had the effect of upholding the provisions as constitutional because the district court had so held. Collins v. Montgomery County Bd. Of Prison Inspectors, 176 F.3d 679, 686 (3d Cir. 1999). The First, Sixth, Eighth, Ninth, and Eleventh Circuits held the provisions constitutional. Boivin v. Black, 225 F.3d 36 (1st Cir. 2000); Hadix v. Johnson, 230 F.3d 840 (6th Cir. 2000) and Walker v. Bain, 257 F.3d 660 (6th Cir. 2001); Foulk v. Charrier, 262 F.3d 687 (8th Cir. 2001); Madrid v. Gomez, 190 F.3d 990 (9th Cir. 1999); Jackson v. State Board of Pardons and Paroles, 331 F.3d 790 (11th Cir. 2003). Those opinions are deserving of careful attention. Nevertheless, the split in the Third Circuit en banc and the dissents in the Sixth Circuit cases caution against a rush to judgment, and Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000), and McNally v. United States, 483 U.S. 350, 107 S.Ct. 2875, 97 L.Ed.2d 292 (1987), both of which overruled nearly uniform lower court precedent, certainly illustrate that strength of numbers does not always signify an analysis convincing to the Supreme Court. See Buckhannon, 532 U.S. at 621, 121 S.Ct. 1835 (noting that "our disagreeing with a `clear majority' of the Circuits is not at all a rare phenomenon. Indeed, our opinions sometimes contradict the unanimous and long-standing interpretation of lower federal courts.").

We begin with the interests that animate the PLRA fee restrictions. The government and Dr. Daley propose five possible governmental interests furthered by the fee provisions, which mirror the interests asserted in the other circuit cases. It does not matter whether Congress had those interests in mind when enacting the PLRA fee restrictions, or whether they are mere conjecture, because Johnson must negate every conceivable basis which might support the legislation whether or not the basis has any foundation in the record. Heller, 509 U.S. at 320, 113 S.Ct. 2637. The PLRA fee restrictions purportedly are related to the government's interest in: (1) deterring the filing of frivolous claims; (2) regulating the filing of suits predicated on trivial harms; (3) protecting the public fise; (4) regulating federal court intervention into States' management of their prison systems; and (5) preventing windfall fee awards. We will consider these in turn.

Two of the circuits as well as the concurrence here uphold the fee restrictions based solely on the governmental interest in deterring frivolous litigation. In the Ninth Circuit case of Madrid, the inmates acknowledged that the government had a legitimate interest in deterring frivolous filings, but maintained that it was not rational to distinguish between inmates and non-inmates in furtherance of that goal. 190 F.3d at 996. The court rejected that argument, holding that the government could rationally believe that prisoners are responsible for the majority of frivolous suits, and therefore could classify on that basis. Id. That argument is in fact consistent with our decision in Lewis, 279 F.3d at 528-29, and with the decisions of many other circuits upholding the constitutionality of the filing fee provisions of the PLRA. Congress could well have determined that inmates are responsible for a disproportionate percentage of frivolous litigation in the federal courts, and could properly create a disincentive to such filings, as by holding them responsible for the filing fees. In Madrid, the court held that Congress could rationally distinguish between inmates and laypersons in its effort to deter frivolous litigation.

The parties here concede Madrid's point that the government has a legitimate interest in deterring frivolous litigation, and that Congress could properly focus that effort on inmates who are disproportionately responsible for the problem. Johnson maintains, however, that the PLRA fee restrictions at issue here are not rationally related to that goal — in other words, that the fit between the end and the means is so attenuated as to render the restrictions arbitrary. The Madrid court never addressed that more critical issue of whether this particular classification — the attorney fee restitution — is related to that goal.

The First Circuit in Boivin faced a challenge more analogous to the one before us. In that case, Boivin presented "in skeletal form" an argument that there was a "complete lack of fit between the means that Congress chose (capping attorney's fees) and the end that it sought to achieve (reducing frivolous prisoner litigation)." 225 F.3d at 44-45. The First Circuit understood the argument to be that the fee cap would not deter the filing of frivolous actions because fees are awarded only to prevailing parties. The Boivin court rejected that argument:

[c]ommon sense suggests that this ex post view is untenable. Congress presumably feared the motivating effect of the prospect of attorneys' fees, ex ante, and the fee cap quells that effect by capping the potential payoff. This changes the odds, and forces both lawyer and client, out of self-interest, to assess likely outcomes with greater care before filing a suit that, even if nominally successful, might leave them holding a nearly empty bag.

Id. at 45. The court's " ex post" and " ex ante" distinction, however, fails to recognize that the PLRA fee restrictions do not impose any added disincentive ex ante to the filing of frivolous litigation, because § 1988 already prohibits fees for such litigation. Boivin theorizes that Congress feared that potential attorney's fees could provide motivation for litigation. It then holds that the prospect of a limited fee may alter that analysis at the beginning, thus diminishing frivolous litigation. That assumes an initial assessment by the inmate of the strength of the case, because the fee recovery under the PLRA is tied to the damages attained. If § 1988 authorized the payment of fees for even frivolous litigation, then the PLRA fee restrictions would indeed reduce the motivation for filing that such fee availability provides. Just the opposite is true, however.

Under § 1988, no fees are possible for a frivolous case, and in fact sanctions may be imposed for such litigation. Therefore, the PLRA fee restrictions, by their very terms, are inapplicable to such a case, because those restrictions apply only to inmate cases "in which attorney's fees are authorized under section 1988." See § 1997e(d)(1) (2). In addition, the § 1988 provisions deny all fees for frivolous litigation, thus imposing a greater disincentive ex ante to filing than the PLRA fee restrictions themselves would impose. The First Circuit failed to answer how an inmate filing frivolous litigation would be motivated to withdraw it by the prospect that he could obtain only 150% of the judgment as fees, when the law prior to the PLRA already established that he could obtain no fees at all for such litigation. If the goal is to reduce frivolous litigation, the PLRA fee restriction adds nothing to the equation. The law under § 1988 already removed the prospect of any fees in such circumstances, and therefore any contemplative effect that potential fees might have on the filing decision was already fully leveraged by § 1988. Given the prohibition on fees under § 1988 for frivolous litigation, the PLRA fee restrictions have no impact whatsoever on the filing of frivolous litigation that is not already provided by § 1988. The language of the statute acknowledges that by limiting the PLRA fee restrictions to cases in which fees are authorized under § 1988. Instead, the impact of the PLRA fee restrictions ex ante is only on the inmate contemplating the filing of non-frivolous litigation in which fees are potentially available under § 1988. See Hadix, 230 F.3d at 844 n. 3 ("with respect to truly frivolous claims, it could be plausibly argued that the provision would have at best a very attenuated effect on the litigation decision of a pro se inmate, whose claims must be certified as possessing at least some merit at an early stage in the litigation. Further, the fee cap provisions only apply to cases in which the prisoner has actually prevailed, thereby assuring that at least one claim was meritorious."). The government, however, does not and cannot argue that Congress has a legitimate interest in discouraging meritorious litigation by inmates.

In recognition of the weakness of the frivolousness argument, the Sixth, Eighth, and to some extent the Eleventh Circuits rested their decisions on the second asserted governmental interest — that of decreasing marginal or trivial lawsuits. I will assume for purposes of this dissent that the government may have a legitimate interest in decreasing the number of meritorious civil rights suits that involve trivial harms. The government links its interest in decreasing trivial litigation with the fee restrictions by opining that: (1) the fee restrictions will provide a disincentive that did not previously exist for an attorney to take a case involving trivial violations; (2) this will increase the likelihood that attorneys will decline to take such cases and (3) that prospect will in turn cause inmates to refrain from filing such cases because they will have to proceed pro se, or it will cause them to abandon their suits when unable to obtain representation.

Again, however, the PLRA adds no restriction on fees that was not already present under § 1988, and therefore does not add a disincentive for the prisoner to Me that trivial litigation. For the same reason, the PLRA fee restrictions would not impact an attorney's decision to file a frivolous or trivial suit in the first place. Under either the PLRA or § 1988, that attorney would need to initially assess the potential for fees before filing such a case, and because § 1988 already prohibits fees in a frivolous or trivial suit, the PLRA does not create any additional disincentive to filing such suits. All the PLRA does is to notify attorneys filing frivolous prisoner civil rights actions that they will get a maximum of 150% of zero instead of zero. The Sixth Circuit in Hadix and Walker failed to analyze the PLRA restrictions in the context of § 1988 fees generally, and therefore did not address whether the PLRA fee restriction added any disincentive that was not already provided by § 1988 fee restrictions. Because the Eighth Circuit's analysis consisted entirely of a quote from Walker, and the Eleventh Circuit merely quoted Walker and Hadix without separate analysis, those decisions manifest the same shortcoming. The § 1988 context is critical, however, in determining whether the means are rationally related to the ends; the search for a link between the classification and objective is what gives substance to the requirement of equal protection. Romer, 517 U.S. at 632-33, 116 S.Ct. 1620 (to survive rational relationship scrutiny, the law must be "narrow enough in scope and grounded in a sufficient factual context for us to ascertain some relation between the classification and the purpose it served").

The PLRA fee restriction, placed in the context of the § 1988 remedy that it modifies, fails to cross the first step in the three-step sequence that would supposedly reduce the volume of trivial litigation, because it does not provide a disincentive that did not previously exist for an attorney to take a case involving trivial violations. That is because attorney's fees are already presumptively unavailable in cases involving trivial or de minimis violations. See discussion supra at 590-592. Only where other factors establish that a seemingly minimal victory was actually significant, as where it involves a significant legal issue and serves an important public interest, will fees be available for a case that would otherwise be considered trivial or de minimis. Farrar, 506 U.S. at 122, 113 S.Ct. 566; Simpson, 104 F.3d at 1001. The government does not argue, nor could it, that it has a legitimate interest in discouraging the filing of constitutional claims involving a significant public interest. Therefore, that exception is inapplicable. We are left with the pre-existing § 1988 law that presumes that no fees are available for cases involving de minimis or trivial harins. Therefore, before the PLRA, an inmate deciding whether to file a lawsuit would be presented with the prospect of attorney's fees for meritorious cases involving significant harms, but with no fees available if the suit prevailed but involved only trivial harms, and with no fees and possibly sanctions if the suit was frivolous. The PLRA fee restrictions do nothing to alter the status quo for the inmate who brings the frivolous or trivial suit. It serves only to create a significant disadvantage for those presenting significant, meritorious challenges. Once again, however, the government does not allege that it has a legitimate interest in deterring the filing of such meritorious suits, nor does it argue that prisoners as opposed to others are responsible for more of those suits or are otherwise a proper class to single out for such a goal.

Even if fees were available under § 1988 to attorneys representing prisoners in trivial civil rights actions, the subset of cases in that category impacted by the PLRA fee restrictions is so small that it may actually be zero. We begin with the reality that only 1% of all prisoner cases even involve private attorneys. The other 4% involve appointed counsel, and by definition courts do not appoint counsel in trivial cases. Within that 1% of cases, we are asked to believe that some of those cases are trivial but that an attorney nonetheless agreed to represent the prisoner, that the attorney would not be deterred from filing by the prospect of little or no fees under the "qualitative" restitution of § 1988, but that the attorney will be deterred by the liquantitative" restriction of the PLRA. There is no reason to believe that any cases fall within that hypothetical reach.

The district court, faced with the reality of prisoner litigation on a daily basis, was skeptical of this line of reasoning. The court noted that "in nearly 100%" of prisoner cases involving representation, the attorneys were appointed by the court, and that of the 1,980 prisoner civil rights cases filed in the Wisconsin federal courts between January 1995 and September 1999, only 2.5% had counsel appointed. Johnson v. Daley, 117 F.Supp.2d 889, 898 (W.D.Wis. 2000); see also Roger A. Hanson Henry W.K. Daley, United States Dept. of Justice, Bureau of Justice Statistics, Challenging the Conditions of Prisons and Jails: A Report on Section 1983 Litigation, at 21-22 (Table 6) (Dec. 1994) (recounting that 96% of prisoner civil rights actions are filed pro se, with attorneys appointed in approximately 4% and private attorneys in approximately 1% of the cases). The court noted that in its thirty years of handling prison civil rights cases, it could think of none that were attorney initiated. 117 F.Supp.2d at 898. The court concluded that the asserted connection between the filing decision and attorney's fees was too tenuous:

At the heart of defendant's argument is the idea that prisoners perform a cost-benefit analysis in deciding whether to file a lawsuit, weighing a variety of factors. Although it is reasonable to assume that a pro se prisoner does do this, it is irrational to conclude that he bases his decision on the distant possibility that at some future time, his presently non-existent lawyer might recover a smaller rather than a larger amount of fees.

Id. at 896. The court further noted that although the purported goal was to discourage frivolous cases, there was "no chance" of counsel being appointed in frivolous cases, and accordingly the only prisoners affected by the fee provisions were those who filed meritorious complaints for whom the court could not secure counsel because of the limited fees available under the PLRA. Congress' goal of reducing the burden of prisoner suits on federal courts would be ill-served by a fee restriction that forced district courts to plod through pro se filings in meritorious cases because of the inability to obtain appointed counsel. The reality in which the fee restrictions operates negated the purported connection between the classification and the goals. Id. at 900; see Heller, 509 U.S. at 321, 113 S.Ct. 2637(noting that "even the standard of rationality as we so often have defined it must find some footing in the realities of the subject addressed by the legislation").

On the other hand, the PLRA fee restiictions have a very direct, foreseeable, and inevitable impact on the meritorious actions filed by attorneys, whether appointed or privately retained. Where the relation with the alleged governmental interest is that dubious, and the impact on an unrelated class of meritorious cases so obvious and severe, no rational connection exists between the classification and the purported interest.

I must emphasize that we are not concerned here with the possibility that the PLRA fee restrictions are merely cumulative of provisions in § 1988. Statutes do not violate equal protection solely because they overlap somewhat with other preexisting remedies. The problem in this case is that the government interest justifying the classification is in the deterrence of frivolous and trivial suits, yet the PLRA restrictions will not even apply to such suits. The language of the PLRA clarifies that the fee restrictions apply to actions in which § 1988 fees are authorized, thus making the contours of § 1988 the starting point in the analysis. Because no fees are available for frivolous or trivial cases, the PLRA fee restrictions will never apply to such cases, and therefore the governmental interest in deterring those cases is wholly inapposite.

Although the fit between means and ends need not be perfect, there must be some fit in order for the legislation to survive rational relationship scrutiny. If the interest is in deterring frivolous and trivial litigation, but the legislation impacts only non-trivial meritorious litigation, that fit is lacking. The present case bears witness to the lack of fit between the fee restrictions and the deterrence of frivolous litigation. Johnson's meritorious claim concerned actions which placed his life in danger. His claim cannot be characterized as trivial under any definition of that term. Nevertheless, he presented evidence to the district court that at least four private attorneys refused to take his case, and he obtained representation only after counsel was appointed for him by the district court. That attorney, having accepted the appointment from the court, is now faced with the prospect of recovering fees which are inadequate to compensate for the effort reasonably expended and the risk undertaken, and that meager compensation is attributable solely to the status of the client as incarcerated rather than to the relative merit of the claim. Of course, a statute does not fail the rational relationship test solely because its imprecise reach impacts persons beyond those within its legitimate interest. Neither, however, can we analyze the claim in some kind of Never-Neverland, where the context and the actual reach of the statute are ignored. See Romer, 517 U.S. at 632-33, 116 S.Ct. 1620 (to survive rational relation scrutiny, law must be grounded in sufficient factual context for Court to ascertain some relation between the classification and the purpose it serves); Heller, 509 U.S. at 321, 113 S.Ct. 2637("even the standard of rationality as we so often have defined it must find some footing in the realities of the subject addressed by the legislation"); Lindsey, 405 U.S. at 78, 92 S.Ct. 862 (analyzing the statutory doublebond requirement in the context of the pre-existing state statutes). The PLRA fee restrictions go well beyond being somewhat under-inclusive or over-inclusive; rather, essentially all of the interests that the restrictions allegedly target are excluded from its reach by § 1988, whereas the PLRA fee restrictions have a direct, foreseeable impact on the class of litigation in which Congress claims no interest. Burlington Northern R. Co. v. Ford, 504 U.S. 648, 653, 112 S.Ct. 2184, 119 L.Ed.2d 432 (1992) (recognizing that equal protection clause is violated if the under-inclusiveness and over-inclusiveness is so great that the rules can no longer be said rationally to implement the policy judgment); Romer v. Evans, 517 U.S. 620, 632, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996) (amendment fails rational relationship test where "its sheer breadth is so discontinues with the reasons offered for it that [it] seems inexplicable by anything but animus toward the class it affects"). Where the means are completely divorced from the end, as here, the litigation fails the rational relationship test. See Massachusetts Bd. of Retirement v. Murgia, 427 U.S. 307, 314, 96 S.Ct. 2562, 49 L.Ed.2d 520 (1976) (examining whether mandatory retirement statute "has the effect of excluding from service so few officers who are in fact unqualified as to render age 50 a criterion wholly unrelated to the objective of the statute.").

This conclusion is required if the rational relationship inquiry is to retain any meaning at all, and comports with an analogous Supreme Court case, Lindsey v. Normet, 405 U.S. 56, 92 S.Ct. 862, 31 L.Ed.2d 36 (1972). In Lindsey, the Court addressed the constitutionality of an Oregon Forcible Entry and Detainer statute that required tenants appealing an adverse determination to post bond in twice the amount of the rent expected to accrue pending the appellate decision. Id. at 64, 92 S.Ct. 862. The governmental interest advanced for that double-bond requirement was to prevent frivolous appeals for dilatory purposes, because absent an added cost the tenant would have no disincentive to forego the appeal. Additionally, the Oregon Supreme Court opined that the additional bond payment would compensate for waste or would be in lieu of damages for unlawful holding over. Id. at 76, 92 S.Ct. 862.

The U.S. Supreme Court analyzed the double-bond provision in the context of the pre-existing Oregon appeal provisions for civil cases generally. Id. at 74-78, 92 S.Ct. 862. In ordinary civil litigation, the litigant was required to file an undertaking with one or more sureties covering all damages, costs and disbursements which might be awarded against the appellant on appeal. Moreover, for an appellant to obtain a stay of execution in a case involving a monetary judgment, the undertaking had to also provide that the appellant would satisfy the judgment if she lost the appeal. Where the judgment was for real property, that undertaking had to provide that the appellant would commit no waste during the pendency of the appeal, and would pay for the use of property during that time should the appellant lose the appeal. Id. at 74-75, 92 S.Ct. 862. The doublebond provision imposed additional obligations for appeals from Forcible Entry and Detainer judgments. In addition to the undertaking required of all civil litigants, an appellant in such an action was compelled to provide an undertaking with two sureties for the payment of twice the rental value of the premises, and if the judgment was affirmed, the landlord was automatically entitled to twice the rents accruing during the appeal. Id. at 75, 92 S.Ct. 862.

The Court began its analysis of the double-bond provision by recognizing that even though the state is not required to provide appellate review, when an appeal is afforded, equal protection required that the right not be arbitrarily and capriciously granted to some litigants and not others. 405 U.S. at 77, 92 S.Ct. 862. We are presented with a similar situation here: Congress was not required to provide a mechanism for awarding attorney's fees, but once provided the fees cannot be arbitrarily and capriciously denied to one class of litigants without violating equal protection.

Although the Lindsey Court went on to acknowledge that Oregon had a legitimate interest in ensuring that appellants post adequate security for appeal, the Court held that the double-bond requirement did not effectuate that purpose. The Court emphasized that the undertaking already protected the landlord by assuring payment for accrued rent and protection against waste. Id. at 77-78, 92 S.Ct. 862. The Court rejected the claim that the double-bond provision would screen out frivolous appeals, because "it not only bars nonfrivolous appeals by those who are unable to post the bond but it also allows meritless appeals by others who can afford the bond." Id. at 78, 92 S.Ct. 862. Thus, the Court recognized that the statute was deficient because it was both over — and under-inclusive, and because pre-existing statutes already protected the state's asserted interest. Lindsey's criticisms parallel the problems presented by the PLRA. The asserted interests in deterring frivolous and trivial litigation are already covered by § 1988, which does not allow any fees at all in such cases. In that context, the limitation on the amount of fees is meaningless for those type of cases. Moreover, the PLRA inevitably impacts the non-trivial, meritorious civil rights cases which Congress claims no interest in deterring. The impact of the 150% limitations are magnified in the prison context, where damage awards are consistently lower than are provided in actions by non-prisoners and where the risk of no recovery is much greater. By linking fees to a percentage of those lower damages, the fee has a greater adverse impact in the prison context than might be experienced in nonprisoner litigation. Finally, the fee restrictions at issue in this case are just two of a number of other restrictions on fees in § 1997e(d). Section 1997e(d)(1) limits the fee amount in two ways: the fee must be directly and reasonably incurred in proving an actual violation, and the amount must be proportionately related to the court ordered relief or directly and reasonably incurred in enforcing the relief ordered for the violation. Therefore, the 150% cap on total fees and on the hourly rate will in all cases operate to reduce the fee award below that baseline amount of fees that were directly, reasonably incurred and proportionate to the relief ordered. This impact further undermines the arguments made in this case. The fee restrictions have no impact on the frivolous and trivial cases, and operate only to reduce the fee award in meritorious litigation below the amount that was both reasonably incurred and proportionate to the relief. The restrictions therefore lack the fit between classification and governmental interest that is necessary to ensure that the classification is not drawn for the purpose of disadvantaging the group burdened by the law. Romer, 517 U.S. at 633, 116 S.Ct. 1620. The fit between means and ends is even more attenuated in this case than in Lindsey, and therefore the PLRA fee restrictions violate equal protection. See also Rinaldi v. Yeager, 384 U.S. 305, 310, 86 S.Ct. 1497, 16 L.Ed.2d 577 (1966) (invalidating requirement that only incarcerated persons repay the costs of transcripts; holding that the statutory interest in deterring frivolous appeals cannot survive equal protection scrutiny because the statute inevitably burdens many whose unsuccessful appeals were nonfrivolous, while leaving untouched many whose appeals were frivolous.)

The remaining interests asserted by the government similarly are insufficient to survive even that minimal level of scrutiny. The government maintains that the PLRA fee restrictions are rationally related to the government interests in reducing federal court intervention into state management of prisons, preventing windfall fee awards, and protecting the public fisc. The first argument is that successful prison litigation may result in consent decrees, involving the federal court in state management of prisons, and that the fee restrictions may reduce the incentive for prisoners to file such claims, thus diminishing federal court involvement in state management of the prisons. The remaining arguments similarly focus on the impact of meritorious litigation on the state, asserting that the fee restrictions would cabin judicial discretion in awarding fees, thus reducing the potential for windfall fee awards, and would protect the public fise by decreasing the fees that must be paid out of the state treasuries. All of these justifications focus on the impact occasioned by meritorious prisoner litigation. Certainly, the states would save money if less meritorious litigation were filed against them, and if fees available for such litigation were restricted (although, again, the government's interest must be a legitimate one to survive scrutiny.) Similar savings could be obtained if Congress passed a similar fee restriction applicable to all blue-eyed litigants, all government employee litigants, or all litigation concerning the educational system. Equal protection, however, requires more than just a showing that some goal is furthered by the legislation; rather, equal protection requires "some rationality in the nature of the class singled out." Rinaldi, 384 U.S. at 308-09, 86 S.Ct. 1497; Schilb v. Kuebel, 404 U.S. 357, 368, 92 S.Ct. 479, 30 L.Ed.2d 502(1971).

The Court examined the need for that fit between classification and goal in Rinaldi. There, a New Jersey statute required indigent persons convicted of a crime, who were confined to prison, to repay the cost of the transcript if they were unsuccessful on appeal. It required no such repayment, however, from those who received a suspended sentence, probation, or only a fine. The state argued that its statute was designed to replenish the county treasury from those who had benefited from county expenditures, and that it would deter frivolous appeals. 384 U.S. at 309-10, 86 S.Ct. 1497. The Court rejected that argument, however, holding that the statute was unconstitutional under the Equal Protection Clause. Certainly, the statute would help replenish the county treasury, but that furtherance of the fiscal goal did not end the Court's inquiry. Instead, the Court held that the Equal Protection Clause requires that, "in defining a class subject to legislation, the distinctions that are drawn have `some relevance to the purpose for which the classification is made.'" Id. at 308-09, 86 S.Ct. 1497. In Rinald the classification distinguished institutionalized unsuccessful appellants from those not institutionalized. The Court held, however, that "[t]here is no defensible interest served by focusing on that distinction as a classifying feature in a reimbursement statute, since it bears no relationship whatsoever to the purpose of the repayment provision." Id. at 309, 86 S.Ct. 1497. Whether a person was institutionalized or not, or the nature of the sentence imposed, was a trait unrelated to the fiscal objective of the statute. Id. at 309-310, 86 S.Ct. 1497. The Court further rejected the state's claim that the classification was a matter of administrative convenience, because those punished by fines could be reached through ordinary garnishment, and repayment could easily be made a condition of probation. Id. at 310, 86 S.Ct. 1497. Thus, the Court again recognized that the constitutionality of a statute cannot be analyzed in a vacuum, but must include a recognition of the real context in which it operates. Because incarceration status had no rational connection to a transcript cost, the Court held that the statute lacked rationality in the nature of the class singled out and was unconstitutional under the Equal Protection Clause. Id.; Schilb, 404 U.S. at 368, 92 S.Ct. 479; see also James v. Strange, 407 U.S. 128, 92 S.Ct. 2027, 32 L.Ed.2d 600 (1972) (relying on Rinaldi and holding that statute giving state right to recover legal defense costs from indigent defendants and depriving them of protective exemptions violates equal protection).

We are presented with the same situation in addressing the remaining governmental interests asserted here. The government asserts an interest in diminishing the fiscal burden of fees on the states, and also in decreasing the costs associated with defending and monitoring compliance in cases of meritorious litigation. There is no rationality, however, in imposing such a restriction on attorneys representing institutionalized persons, because the locale of the litigant is unrelated to the fiscal objective. Although Congress can approach a problem piece-meal, there is no attempt in the briefs to explain why inmates should be singled out over non-inmates for the restriction in cases involving meritorious litigation, which is all that is addressed by these last three proffered governmental interests. For instance, the prisoner/nonprisoner classification makes no sense if the interest is in simply protecting the public fisc, because there is no reason to believe that prisoners are filing more nontrivial meritorious cases than non-poisoners, or that they are achieving windfall fee awards in greater numbers in such cases. In fact, the record indicates that at the time the PLRA fee restrictions were enacted, nearly 96% of all prison litigation was pro se, and there is no correlation between inmate status and windfall fee awards (at least no correlation that would indicate such awards are more likely to be provided to inmates; the reverse, however, might be true). There is no rational basis to connect inmate status with the goal of reducing the fiscal burden any more than Congress could single out blue-eyed litigants and restrict their fees. The classification itself has to be related to the interest asserted, and it is not. Congress in enacting the PLRA was concerned with the flood of frivolous litigation, not the flood of meritorious litigation. There is simply no basis in the record or in theory upon which we could hold that meritorious litigation by prisoners is creating an economic burden justifying the singling out of inmates for this type of restriction. There is, in short, no connection between the fiscal goals of the statute and the status of the meritorious litigant as either an inmate or a non-institutionalized individual.

As a means of discouraging frivolous or trivial suits by prisoners, the PLRA imposes no restriction on fees that § 1988 does not already impose. As a means of reducing the burden on states, there is no rational reason to single out prisoners. The PLRA fee restrictions will, however, have a significant, predictable impact on the ability of prisoners with meritorious cases to obtain representation. As the district court recognized, "the only prisoners affected are those who file meritorious complaints for whom the court cannot secure counsel because of the limited fees available to lawyers in such cases." Johnson, 117 F.Supp.2d at 896. Rather than decrease the burden on courts, these restrictions will have the opposite effect, making it more difficult for courts to persuade lawyers to accept appointment. Id. at 898. The inevitable result will be that constitutional violations against prisoners will go unremedied, and that is contrary to the purposes of the civil rights acts and § 1988. Congress has no legitimate governmental interest in deterring the filing of meritorious lawsuits. As such, the classification is not rationally related to the goals and fails Equal Protection scrutiny. I respectfully dissent.

FLAUM, Chief Judge, dissenting.

In my judgment, as Judge Rovner has indicated in her dissent, the plurality's proposed benchmarks for answering the "compared to what?" question miss the mark. The appropriate focus should be between prisoner and nonprisoner litigants seeking fees under § 1988 — indeed, I beheve that perhaps the most apt comparison would be between former prisoners bringing suit for violations that occurred during their imprisonment and current prisoners bringing suit for identical violations. Cf. Kerr v. Puckett, 138 F.3d 321, 323 (7th Cir. 1998) (holding that prisoner as defined in 42 U.S.C. § 1997e(h) does not comprehend a felon who has been released). Concluding that Judge Rovner has identified the correct approach for comparison purposes and agreeing with the well-reasoned analysis contained in Parts Il and III of her opinion, I respectfully dissent.