Full opinion text
CARNES, Circuit Judge: When asked how much money would be enough for him, John D. Rockefeller reportedly said: “Just a little bit more.” The attorneys for the plaintiff class in this case want more than just a little bit more. They want a lot more money than they would receive from multiplying the number of hours they worked on this case by the hourly rate they charge. And the district court gave them a lot more— $4,500,000 more — out of the pockets of the taxpayers of Georgia. Not caught up in the district court’s spirit of generosity with their money, the Georgia taxpayers (through their governor and other officials who are defendants) have appealed the attorneys’ fees award because of that multi-million dollar enhancement and for other reasons. The plaintiffs’ attorneys, convinced that the district court was not generous enough, have cross-appealed (through the class they represent), insisting that they deserve even more than the $10,500,000 total fees that the court awarded them. I. Kenny A. and eight other named plaintiffs in this lawsuit are foster children in the custody of the Georgia Department of Human Resources. Kenny A. ex rel. Winn v. Perdue, 454 F.Supp.2d 1260, 1266 (N.D.Ga.2006) (Kenny A. III). In June 2002, on behalf of a class of all foster children in Fulton and DeKalb Counties and a subclass of African American foster children, the plaintiffs sued the governor, DHR, its commissioner, Fulton and De-Kalb Counties, each county’s department of family and children services, and the director of each of those departments. Id. at 1266-67. The complaint alleged systemic deficiencies in the counties’ foster care systems. Id. at 1267. These deficiencies, according to the complaint, included: (1) assigning excessive numbers of cases to inadequately trained and poorly supervised caseworkers; (2) not developing a sufficient number of foster homes properly screened to ensure the plaintiff children’s safety; (3) not identifying adult relatives who could care for the plaintiff children as an alternative to strangers or impersonal institutions; (4) failing to provide relevant information and support services to foster parents in order to prevent foster placements from being disrupted; (6) failing to develop administrative controls such as an information management system that ensures plaintiff children are expeditiously placed in a foster home matched to meet the children’s specific needs; (6) failing to provide timely and appropriate permanency planning, including failing to provide services that would enable plaintiffs to achieve their permanency planning goals; (7) placing, plaintiffs in dangerous, unsanitary, inappropriate shelters and other placements; (8) failing to provide appropriate and necessary mental health, medical, and education services to children in their custody; and (9) separating teenage mothers in foster care from their own children and separating siblings in foster care from each other without providing visitation. Id. The complaint asserted fifteen causes of action under federal and state law. Kenny A. ex rel. Winn v. Perdue, 218 F.R.D. 277, 283 (N.D.Ga.2003) (Kenny A. I). The federal law claims included alleged violations, brought under 42 U.S.C. § 1983, of the class members’ Fourteenth Amendment rights to substantive and procedural due process and their First, Ninth, and Fourteenth Amendment rights to liberty, privacy, and association. Id. The complaint also alleged violations of the class members’ statutory rights under the Adoption Assistance and Child Welfare Act of 1980, the Multiethnic Placement Act of 1994, as well as violations of the early and periodic screening, diagnosis, and treatment program of the Medicaid Act. Id. The state law claims asserted included violations of the class members’ substantive due process and equal protection rights under the Georgia Constitution, violations of various Georgia statutes, and claims of nuisance, breach of contract, and inadequate and ineffective legal representation. Id. The complaint sought declaratory and injunc-tive relief, as well as attorneys’ fees and expenses. Id. at 283 n. 1. The district court denied the defendants’ motions to dismiss and for summary judgment. Kenny A III, 454 F.Supp.2d at 1269; Kenny A. ex rel. Winn v. Perdue, 356 F.Supp.2d 1353, 1355 (N.D.Ga.2005) (Kenny A. IP); Kenny A. I, 218 F.R.D. at 305. It certified the class of all foster children in the Fulton and DeKalb County foster care systems, and the subgroup of African American foster children “who have had, or are subject to the risk of having, their adoption delayed or denied on the basis of their race or color,” Kenny A. I, 218 F.R.D. at 305, and set the case for trial, Kenny A III, 454 F.Supp.2d at 1268-69. At the same time, the district court referred the case to mediation. Id. at 1269. As that court would later describe it, “over the next four months the parties attended eighteen separate mediation sessions where they spent more than 110 hours trying to hammer out a settlement agreement.” Id. They were eventually successful. Id. In May 2006 the district court gave final approval to the settlement between the plaintiff class and Fulton and DeKalb Counties; the parties agreed to an attorneys’ fees award in that part of the case; the district court entered that award, and it has not been appealed. This appeal arises from the settlement involving the rest of the defendants, which was approved by the district court in October 2005. Id. The court summarized that settlement this way: Its centerpiece is a series of thirty-one outcome measures that State Defendants have agreed to meet and sustain for at least three consecutive six-month reporting periods. The outcome measures, many of them requiring phased-in results over a two-year period, seek to improve performance in the following areas: timely commencement and thorough completion of investigations of reported abuse or neglect; regular visits of foster children by case workers; approval and licensure of foster homes and other placements; the percentage of children who are the victims of substantiated maltreatment while in foster care; the percentage of children in foster homes that exceed their licensed capacity; the percentage of children who have experienced multiple moves while in foster care; and periodic judicial reviews of the safety and status of foster children. In addition, the Consent Decree requires comprehensive and periodic delivery of medical, dental, and mental health services to foster children; a detailed process for improved goal-setting, case planning and periodic reviews of children’s status while in foster care; limits on the placement of children in emergency shelters and group homes and institutions, and protections against overcrowding in foster homes; and the establishment of reimbursement rates to adequately compensate providers for caring for foster children. Moreover, State Defendants commit to reduced caseloads for all case managers and supervisors; a fully implemented single statewide automated child welfare information system; and maintaining or establishing placements and related services identified in a “needs assessment” to be conducted by a neutral expert. The settlement also includes processes for the supervision of private contract agencies that provide homes and services for foster children; improvements in foster parent screening, licensing and training, as well as foster parent support and communication; improvements in case manager training; improvements in processes for addressing suspected abuse or neglect and suspected corporal punishment of children in foster care; and improvements in efforts to maximize available federal funding. Finally, the settlement provides that two child welfare specialists will serve as the Court’s independent accountability agents charged with the responsibility of measuring and reporting publicly on the State Defendants’ compliance with these and other undertakings as specified in the Consent Decree. Id. at 1289. The 2005 settlement agreement further acknowledged that “Plaintiff Class is entitled to recover its expenses of litigation, including reasonable attorneys’ fees and nontaxable costs, pursuant to 42 U.S.C. § 1988 and Fed.R.Civ.P. 28(h).” The agreement said that the parties would “attempt without court intervention to resolve the proper amount of Class Counsel’s fees and expenses of litigation.” If the parties could not reach an agreement, then “[t]he amount of any award shall be determined by the Court in accordance with the requirements of applicable law and procedures.” That is what happened. Because the parties could not agree on the amount of fees, the class filed a motion for the court to make a determination and award. Id. at 1269-70. It was accompanied by 2,500 pages of billing records, by affidavits from the two lead counsel for the class, and by affidavits from five Atlanta area attorneys who were not involved in the case. Id. at 1270, 1290. The motion sought a total of $14,342,860 in fees. Id. Half of that amount, $7,171,434.30, was to compensate the attorneys and their paralegals for the 29,908.73 hours they claimed to have worked at rates they sought; those rates ranged from $75 to $495. Id. The other $7,171,434.30 they sought was to be an enhancement of the fee award for a job well done. Id. The defendants objected to some of the hourly rates requested and to the number of hours claimed. Id. They broke down the 2,500 pages of submitted billing records into fifteen categories of hours representing different stages of the case, and argued that in fourteen of those categories the plaintiffs’ attorneys had expended more hours than was reasonably necessary to effectively represent the class. Id. at 1273-74 & n. 5. The defendants also argued that many of the entries on the billing records were too vague to support a claim for compensation. Id. at 1274. As for the request that the fee award be doubled by means of a $7 million enhancement, the defendants objected, arguing that this was not one of the rare cases in which an enhancement would be appropriate. They argued that enhancing the fee based on the results obtained would be improper because the skill of the plaintiffs’ attorneys in litigating the case would already have been taken into account in setting their hourly rates. Double counting the skill factor would be inappropriate, the defendants argued. The district court overruled all of the defendants’ objections to the hourly rates sought by the plaintiffs’ attorneys. Id. at 1285. It granted in full each hourly rate requested, ranging from $215 to $495 for plaintiffs’ attorneys, depending on their relative skill and experience. Id. at 1284-85. The court did not reduce the requested rates at all even though some of them were New York rates instead of Atlanta rates, and even though some of the attorneys (for example, Marcia Robinson Low-ry, one of the two lead attorneys) recently had been awarded a smaller hourly rate for work on another case of exactly the same type. Id. at 1284-86. Likewise, the court approved in full the requested hourly rates of $75 to $150 for paralegal work. Id. at 1285. The district court did partially sustain the defendants’ objection to the number of hours that the plaintiffs’ attorneys claimed to have worked on the case. Id. at 1274. The court agreed with the defendants that some of the entries on counsel’s billing records were vague and that the hours claimed for nine of the fifteen billing categories were excessive. Those nine excessively billed categories were: (1) preparing the complaint and mandatory disclosures; (2) litigating the preliminary injunction motion; (3) producing and analyzing documents; (4) litigating discovery motions; (5) conferencing with each other; (6) preparing expert witness reports; (7) responding to the summary judgment motion; (8) preparing for trial; and (9) travel. Id. at 1274-84,1286. The court dealt with the vague and excessive billing two ways, one for the non-travel hours and the other for travel hours. It made “an across-the-board 15% reduction in the number of non-travel related hours claimed by plaintiffs’ counsel.” Id. at 1286. This 15 percent reduction amounted to 4,371.22 hours being sliced off the 29,141.46 total number of non-travel hours for which the attorneys had requested compensation. In dollar terms, that amounted to a reduction of $1,040,176.92 in non-travel related attorneys’ fees. To remedy the vagueness and excessive billing problem as it related to travel hours, the court halved the hourly rate, thereby effectively halving the number of travel hours that were compensated from 767.27 to 383.64, which amounted to a reduction of $118,460.75. When travel and non-travel fees are considered together, the court reduced the overall $7,171,434.30 fee request by $1,158,631.40, which amounted to an overall reduction of 16 percent. After the reductions, the total award was $6,012,802.90 before enhancement. Id. On the applicability of an enhancement to the fee request, the court overruled the defendants’ objections. Id. at 1288. The court found that the plaintiffs’ attorneys were entitled to an enhancement because their hourly rate did “not take into account (1) the fact that class counsel were required to advance case expenses of $1.7 million over a three-year period with no ongoing reimbursement, (2) the fact that class counsel were not paid on an on-going basis as the work was being performed, and (3) the fact that class counsel’s ability to recover a fee and expense reimbursement were completely contingent on the outcome of the case.” Id. (footnote omitted). The district court also found that “the superb quality of [counsel’s] representation far exceeded what could reasonably be expected for the standard hourly rates used to calculate the [fee].” Id. at 1288-89. The court commented that “[q]uite simply, plaintiffs’ counsel brought a higher degree of skill, commitment, dedication, and professionalism to this litigation than the Court has seen displayed by the attorneys in any other case during its 27 years on the bench.” Id. at 1289. Finally, the court said that an enhancement was appropriate because “the evidence establishes that plaintiffs’ success in this case was truly exceptional.” Id. For these reasons, and based on the affidavits of some Atlanta area attorneys urging the court to enhance the fee award (even more than it did), the court multiplied the $6,012,802.90 fee award by 1.75, thereby enhancing it by 75 percent or $4,509,602.00. Id. at 1290. That enhancement boosted the total fee award to $10,522,405.08. Id. II. In reviewing the decisions of the district court raised in this appeal and cross-appeal, we look at questions of law anew but we review the court’s findings of fact only for clear error. Atlanta Journal & Constitution v. City of Atlanta Dep’t of Aviation, 442 F.3d 1283, 1287 (11th Cir.2006). The abuse of discretion standard applies to the district court’s determination of the number of compensable billable hours, the hourly rate at which plaintiffs’ counsel is compensated, the award of costs and expenses, and the enhancement deci sion. See Am. Civil Liberties Union of Ga. v. Barnes, 168 F.3d 423, 439 (11th Cir.1999); Richardson v. Ala. State Bd. of Educ., 935 F.2d 1240, 1248-49 (11th Cir.1991). In applying the abuse of discretion standard, we keep in mind that “[w]hen a district court has discretion, there are usually a range of choices it may make and still be affirmed; there is not only one right choice for the court to make.” Blasland, Bouck & Lee, Inc. v. City of N. Miami, 283 F.3d 1286, 1298 (11th Cir.2002); McMahan v. Toto, 256 F.3d 1120, 1128 (11th Cir.2001) (same). This is true “even though we would have gone the other way had it been our call.” Rasbury v. Internal Revenue Serv., 24 F.3d 159, 168 (11th Cir.1994). Of course, “[a] district court by definition abuses its discretion when it makes an error of law.” Koon v. United States, 518 U.S. 81, 100, 116 S.Ct. 2035, 2047, 135 L.Ed.2d 392 (1996); accord Schlup v. Delo, 513 U.S. 298, 333, 115 S.Ct. 851, 870, 130 L.Ed.2d 808 (1995) (O’Connor, J., concurring) (“It is a paradigmatic abuse of discretion for a court to base its judgment on an erroneous view of the law.”). We may also find an abuse of discretion if the district court failed to follow proper procedures in making its determination, based an award upon findings of fact that are clearly erroneous, or committed a clear error of judgment. Johnson v. Breeden, 280 F.3d 1308, 1326 (11th Cir.2002); Sun-America Corp. v. Sun Life Assurance Co. of Can., 77 F.3d 1325, 1333 (11th Cir.1996); BankAtlantic v. Blythe Eastman Paine Webber, Inc., 12 F.3d 1045, 1048 (11th Cir.1994). III. In their cross-appeal the plaintiffs contend that the district court abused its discretion in two respects: (1) by not applying the common fund and common benefit doctrines, which would have resulted in a fee award substantially higher than the one they received; and (2) by not compensating them for the $801,864.40 they spent on expert witness expenses. The district court resolved each of those two issues correctly, and we affirm its decision of them on the basis of the reasoning contained in its opinion. Kenny A. III, 454 F.Supp.2d at 1270-72, 1291-92. IY. The defendants conceded in the settlement agreement that the class had prevailed in the litigation and was entitled to recover reasonable attorneys’ fees under 42 U.S.C. § 1988. That section provides in relevant part that in any action, like this one, where the plaintiffs seek to enforce their constitutional rights under 42 U.S.C. § 1983, “the court, in its discretion, may allow the prevailing party, other than the United States, a reasonable attorney’s fee as part of the costs.” 42 U.S.C. § 1988(b). In their appeal the defendants contend that the district court abused its discretion in three ways. Their first contention is that the district court should not have awarded the plaintiffs’ attorneys all of the photocopying expenses they claimed. See Kenny A. III, 454 F.Supp.2d at 1294-95. Although we probably would not award the full amount of the claimed expenses if we were deciding the matter in the first instance, we cannot say that the district court abused its discretion in doing so. Y. We turn now to the defendants’ other contentions, starting with their argument that the district court erred in calculating the lodestar amount. “The most useful starting point for determining the amount of a reasonable fee is the number of hours reasonably expended on the litigation multiplied by a reasonable hourly rate.” Hensley v. Eckerhart, 461 U.S. 424, 433, 103 S.Ct. 1933, 1939, 76 L.Ed.2d 40 (1983). The product of this formula is the “lodestar,” Loranger v. Stierheim, 10 F.3d 776, 781 (11th Cir.1994) (per curiam), which is “the guiding light of our fee-shifting jurisprudence,” City of Burlington v. Dague, 505 U.S. 557, 562, 112 S.Ct. 2638, 2641, 120 L.Ed.2d 449 (1992). Although the defendants no longer contest the hourly rate component of the district court’s lodestar calculation, they do contend that the district court compensated the plaintiffs’ attorneys for an unreasonable number of hours. As we have already explained, the court essentially cut the total number of hours submitted by 16 percent because of vague entries and excessive billing. See supra at 1217-18. We have held that where the billing records are voluminous, as they are here, a district court may make a reasonable across the board reduction in hours instead of engaging in the pick and shovel work necessary to make a more precise determination. See Loranger, 10 F.3d at 783. Even so, the defendants argue, the district court’s cut was unreasonably shallow — it should have cut much deeper. The district court’s reduction of only 16 percent in the submitted hours does appear charitable, maybe even excessively so, in favor of the plaintiffs’ attorneys, but we are not quite convinced that it was a clear error of judgment, which is what we would have to conclude in order to find an abuse of discretion. See Johnson, 280 F.3d at 1326; SunAmerica Corp., 77 F.3d at 1333; BankAtlantic, 12 F.3d at 1048. Even though we would have cut the billable hours more if we were deciding the matter in the first instance, we cannot say that the result the district court reached was outside the range of permissible choices on this record. See Cook ex rel. Estate of Tessier v. Sheriff of Monroe County, Fla., 402 F.3d 1092, 1103-04 (11th Cir.2005); Cooper v. Southern Co., 390 F.3d 695, 711-12 (11th Cir.2004); Rasbury, 24 F.3d at 168. VI. The defendants’ final contention is that the district court should not have granted plaintiffs’ attorneys a $4.5 million enhancement to the $6 million lodestar amount. That contention can claim the favor of a presumption, and not a weak one either. The Supreme Court has instructed us that there is a “strong presumption” that the lodestar figure, without any adjustment, is the reasonable fee award. Dague, 505 U.S. at 562, 112 S.Ct. at 2641; Pennsylvania v. Del. Valley Citizens’ Council for Clean Air, 478 U.S. 546, 565, 106 S.Ct. 3088, 3098, 92 L.Ed.2d 439 (1986) (Delaware Valley I). That strong presumption can be rebutted only in “rare” and “exceptional” cases, Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098, where the fee applicant has shown that without an upward adjustment of the lodestar amount the fee would be unreasonable, Dague, 505 U.S. at 562, 112 S.Ct. at 2641. Even in the rare and exceptional case where an enhancement is permissible, it must be “supported by both ‘specific evidence’ on the record and detailed findings by the lower court.” Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098. A. In order to appreciate just how rare and exceptional a case must be for an enhancement of the lodestar amount to be permissible these days, a review of the evolution of Supreme Court thinking in this area is necessary. The history of the issue is one of early, tentative statements indicating more receptiveness to enhancements than the statements and holdings of later decisions actually permit. Suggestion has been followed by retrenchment, resulting in a decisional arc that bends decidedly against enhancements. Hensley, a 1983 decision, was the Court’s first stab at interpreting what Congress meant by a “reasonable fee” under § 1988 and similar fee-shifting statutes. In that case the Court adopted the lodestar method for calculating fees. Hensley, 461 U.S. at 433-34, 103 S.Ct. at 1939-40. In doing so, the Court said that “[t]he most useful starting point for determining the amount of a reasonable fee is the number of hours reasonably expended on the litigation multiplied by a reasonable hourly rate.” Id. at 433, 103 S.Ct. at 1939. It added, however, that this “does not end the inquiry,” because “[t]here remain other considerations that may lead the district court to adjust the fee upward or downward, including the important factor of the ‘results obtained.’ ” Id. at 434, 103 S.Ct. at 1940. The context in which that statement was made, however, was not one of upward enhancement. Instead, the actual adjustment issue decided in Hensley was whether the lodestar figure should be reduced for plaintiffs who have achieved only partial or limited success. See id. at 426, 103 S.Ct. at 1935-36 (“The issue in this case is whether a partially prevailing plaintiff may recover an attorney’s fee for legal services on unsuccessful claims.”); see also id. at 434-37, 103 S.Ct. at 1940-41. The Court decided that the fees should be reduced by the amount spent on claims that do not succeed. Id. at 440, 103 S.Ct. at 1943 (“Where the plaintiff has failed to prevail on a claim that is distinct in all respects from his successful claims, the hours spent on the unsuccessful claim should be excluded in considering the amount of a reasonable fee.”); see also id. (“[WJhere the plaintiff achieved only limited success, the district court should award only that amount of fees that is reasonable in relation to the results obtained.”). Even though the issue of enhancements was not presented in Hensley, the opinion does contain this dicta: “Where a plaintiff has obtained excellent results, his attorney should recover a fully compensatory fee. Normally this will encompass all hours reasonably expended on the litigation, and indeed in some cases of exceptional success an enhanced award may be justified.” Id. at 435, 103 S.Ct. at 1940. The use of “may be” instead of “will be” was a harbinger of decisions to come. The next term the Supreme Court decided Blum v. Stenson, 465 U.S. 886, 104 S.Ct. 1541, 79 L.Ed.2d 891 (1984), which, unlike Hensley, was a case in which the district court had actually awarded an enhancement. Id. at 891, 104 S.Ct. at 1545. The court had enhanced the lodestar figure by 50 percent “because of the quality of representation, the complexity of the issues, the riskiness of success, and the great benefit to the large class that was achieved.” Id. (quotation omitted). Unimpressed with that reasoning, the Supreme Court held that awarding the enhancement was an abuse of discretion. Id. at 896-902, 104 S.Ct. at 1547-50. While repeating the Hensley opinion’s statement that “in some cases of exceptional success an enhanced award may be justified,” id. at 897, 104 S.Ct. at 1548 (quotation omitted), the Blum Court qualified that possibility with the statement that the lodestar amount “is presumed to be the reasonable fee contemplated by § 1988,” id. The Court then considered and rejected each of the district court’s four justifications for the enhancement. It began: The novelty and complexity of the issues presumably were fully reflected in the number of billable hours recorded by counsel and thus do not warrant an upward adjustment in a fee based on the number of billable hours times reasonable hourly rates. There may be cases, of course, where the experience and special skill of the attorney will require the expenditure of fewer hours than counsel normally would be expected to spend on a particularly novel or complex issue. In those cases, the special skill and experience of counsel should be reflected in the reasonableness of the hourly rates. Neither complexity nor novelty of the issues, therefore, is an appropriate factor in determining whether to increase the basic fee award. Id. at 898-99, 104 S.Ct. at 1549. The Blum Court then moved on to the next factor: The “quality of representation” ... generally is reflected in the reasonable hourly rate. It, therefore, may [note the use, again, of the word “may” instead of “will”] justify an upward adjustment only in the rare case where the fee applicant offers specific evidence to show that the quality of service rendered was superior to that one reasonably should expect in light of the hourly rates charged and that the success was “exceptional.” Id. at 899, 104 S.Ct. at 1549 (words in brackets added). Otherwise, the Court said, “an upward adjustment for quality of representation is a clear example of double counting.” Id. The Court found the district court’s opinion that “[t]he quality of work performed by counsel throughout this case was high,” without any specific evidence to support it, insufficient to justify the enhancement. Id. at 899-90, 104 S.Ct. at 1549 (quotation omitted). Regarding the results obtained factor, the Court reasoned that successful results are to be considered when calculating counsel’s hourly rate. Id. at 900, 104 S.Ct. at 1549-50. “Because acknowledgment of the ‘results obtained’ generally will be subsumed within other factors used to calculate a reasonable fee, it normally should not provide an independent basis for increasing the fee award.” Id. (footnote omitted). The Blum Court left for another day the question of whether the risk borne by the plaintiffs’ counsel in taking the case on a contingency basis could ever justify an enhancement of the lodestar. Id. at 901 n. 17, 104 S.Ct. at 1550 n. 17. Assuming that it could, however, the Court said that the record did not identify any specific risks to counsel in that case. Id. at 901, 104 S.Ct. at 1550. Accordingly, the Court vacated the portion of the district court’s order enhancing the plaintiffs’ fees award. Id. at 901-02, 104 S.Ct. at 1550. The Court’s next “reasonable fee” case was Delaware Valley J. The district court had enhanced the plaintiffs’ lodestar calculation in that case because of the contingent nature of the fee and the quality of counsel’s work “which culminated in an outstanding result.” Delaware Valley I, 478 U.S. at 555, 106 S.Ct. at 3093 (quotation omitted). As it had done in Blum, the Delaware Valley I Court reviewed the district court’s reasons for enhancing the fee award and found all of them wanting. This time the Court did not just say that there’s a presumption, but emphasized that there is a “strong presumption” that the lodestar figure represents the reasonable fee. Id. at 565, 106 S.Ct. at 3098. The Court explained the basis for that strong presumption: A strong presumption that the lodestar figure — the product of reasonable hours times a reasonable rate — represents a “reasonable” fee is wholly consistent with the rationale behind the usual fee-shifting statute, including the one in the present case. These statutes were not designed as a form of economic relief to improve the financial lot of attorneys, nor were they intended to replicate exactly the fee an attorney could earn through a private fee arrangement with his client. Instead, the aim of such statutes was to enable private parties to obtain legal help in seeking redress for injuries resulting from the actual or threatened violation of specific federal laws. Hence, if plaintiffs, such as Delaware Valley, find it possible to engage a lawyer based on the statutory assurance that he will be paid a “reasonable fee,” the purpose behind the fee-shifting statute has been satisfied. Id. The Court in Delaware Valley I closed some doors for enhancements that it had left ajar in earlier cases. It held that the strong presumption that the lodestar figure is the amount to be awarded- — the presumption against enhancements — -cannot be overcome by the special skill and experience of counsel, the quality of representation, or the results obtained. Id. Those factors, the Court reasoned, “are presumably fully reflected in the lodestar amount, and thus cannot serve as independent bases for increasing the basic fee award.” Id. This is the Court’s reasoning: [Wjhen an attorney first accepts a case and agrees to represent the client, he obligates himself to perform to the best of his ability and to produce the best possible results commensurate with his skill and his client’s interests. Calculating the fee award in a manner that accounts for these factors, either in determining the reasonable number of hours expended on the litigation or in setting the reasonable hourly rate, thus adequately compensates the attorney, and leaves very little room for enhancing the award based on his post-engagement performance. In short, the lodestar figure includes most, if not all, of the relevant factors constituting a “reasonable” attorney’s fee, and it is unnecessary to enhance the fee for superior performance in order to serve the statutory purpose of enabling plaintiffs to secure legal assistance. Id. at 565-66, 106 S.Ct. at 3098. Not using the overall quality of performance to enhance the fee removes “any danger of ‘double counting.’ ” Id. at 566, 106 S.Ct. at 3099. The Delaware Valley I Court reversed the enhancement part of the fee award insofar as the enhancement was based on the quality of the attorneys’ work and the outstanding result obtained. Id. at 568, 106 S.Ct. at 3099-100. It left undecided the question of whether the enhancement was proper insofar as it was based on the contingency nature of the fee — the risk that the attorneys would lose the case and receive no fees. That issue was to be reargued in the same case the next term, id., 106 S.Ct. at 3100, but that attempt to decide it misfired when the Court fractured, with no opinion garnering a majority vote, see Pennsylvania v. Del. Valley Citizens’ Council for Clean Air, 483 U.S. 711, 107 S.Ct. 3078, 97 L.Ed.2d 585 (1987) (Delaware Valley II). It was not until five years later, after the membership of the Court had changed, that the contingency issue was definitively resolved. In Dague the Court squarely rejected the proposition that a court “may enhance the fee award above the ‘lodestar’ amount in order to reflect the fact that the party’s attorneys were retained on a contingent-fee basis and thus assumed the risk of receiving no payment at all for their services.” Dague, 505 U.S. at 559, 112 S.Ct. at 2689. The Court reasoned that an enhancement for contingency would duplicate in substantial part factors that had already been considered in arriving at the lodestar. It explained that “[t]he risk of loss in a particular case (and, therefore, the attorney’s contingent risk) is the product of two factors: (1) the legal and factual merits of the claim, and (2) the difficulty of establishing those merits.” Id. at 562, 112 S.Ct. at 2641. The second factor, the Court said, is “ordinarily reflected in the lodestar — either in the higher number of hours expended to overcome the difficulty, or in the higher hourly rate of the attorney skilled and experienced enough to do so.” Id. “Taking account of it again through lodestar enhancement amounts to double counting.” Id. at 563, 112 S.Ct. at 2641. The Court recognized that, unlike the second contingency factor, the first one — the relative merits of the claim — is not taken into consideration in determining the lodestar figure. That is no problem, the Court decided, because the relative merits of the claim “should play no part in the calculation of the award.” Id. For one thing, there is always a risk that a case will be lost — “no claim has a 100% chance of success.” Id. As a result, permitting adjustments based on risk would mean that the “computation of the lodestar would never end the court’s inquiry in contingent-fee cases.” Id. That would contradict the Court’s repeated instruction that there is a strong presumption that the lodestar is the fee to be awarded and adjustments to it are to be the rare exception, not the general rule. Not only that, the Court said, but awarding enhancements based on the financial risk in taking the case would also “provide attorneys with the same incentive to bring relatively meritless claims as relatively meritorious ones.” Id. The “social cost of indiscriminately encouraging non-meritorious claims” would be too high, and encouraging attorneys to take less meritorious cases is “an unlikely objective of the ‘reasonable fees’ provisions.” Id. at 563, 112 S.Ct. at 2642. Accordingly, the Court concluded that a contingency enhancement is “not consistent with our general rejection of the contingent-fee model for fee awards, nor is it necessary to the determination of a reasonable fee.” Id. at 566, 112 S.Ct. at 2643. The Court in Dague did not merely reverse the application of a contingency or risk enhancement in that case, but also ruled out one in any case, “concluding that no contingency enhancement whatever is compatible with the fee-shifting statutes at issue.” Id. at 565, 112 S.Ct. at 2643. The Dague decision is the last word we have from the Supreme Court on the issue of whether, and if so when, the lodestar amount may be enhanced in calculating the attorneys’ fees award under one of the federal fee-shifting statutes. B. The district court’s $4.5 million enhancement to the $6 million lodestar figure in the present case cannot be squared with the Supreme Court decisions we have discussed. The district court did not explicitly mention, much less give full effect to, the strong presumption that the lodestar amount is a reasonable fee and therefore the fee to be awarded. See Dague, 505 U.S. at 562, 112 S.Ct. at 2641 (“We have established a strong presumption that the lodestar represents the reasonable fee, and have placed upon the fee applicant who seeks more than that the burden of showing that such an adjustment is necessary to the determination of a reasonable fee.” (quotation and citation omitted)); Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098 (“A strong presumption that the lodestar figure — the product of reasonable hours times a reasonable rate — represents a ‘reasonable’ fee is wholly consistent with the rationale behind the usual fee-shifting statute, including the one in the present case.”). The district court relied on several factors in granting the multi-million dollar enhancement to the lodestar figure, most of which are cited in the following passage from its opinion: First, the evidence establishes that the quality of service rendered by class counsel, including their extraordinary commitment of capital resources, was far superior to what consumers of legal services in the legal marketplace in Atlanta could reasonably expect to receive for the rates used in the lodestar calculation. Specifically, the evidence shows that the hourly rates used in the lodestar calculation do not take into account (1) the fact that class counsel were required to advance case expenses of $1.7 million over a three-year period with no ongoing reimbursement, (2) the fact that class counsel were not paid on an ongoing basis as the work was being performed, and (3) the fact that class counsel’s ability to recover a fee and expense reimbursement were completely contingent on the outcome of the case. (Chandler Deck ¶¶5-8; Fellows Decl. ¶¶ 5-8; Knowles Deck ¶¶ 7-12; Rawls Deck ¶¶ 5-8; Lowry Deck ¶ 25; Bram-lett Deck ¶¶ 7(a) & 10,13-14.) Kenny A. III, 454 F.Supp.2d at 1288 & n.8 (citations omitted). None of the three factors the district court relied on in that passage to justify boosting the award is a proper basis for doing so. 1. Starting from the top of the list, the district court’s reliance on the quality of service and superior performance, which are essentially the same thing, conflicts with the Supreme Court’s teachings that those considerations are adequately accounted for “either in determining the reasonable number of hours expended on the litigation or in setting the reasonable hourly rates,” and that it is “unnecessary to enhance the fee for superior performance in order to serve the statutory purpose of enabling plaintiffs to secure legal assistance.” Delaware Valley I, 478 U.S. at 565-66, 106 S.Ct. at 3098; see also id. at 565, 106 S.Ct. at 3098 (“Hence, if plaintiffs ... find it possible to engage a lawyer based on the statutory assurance that he will be paid a ‘reasonable fee,’ the purpose behind the fee-shifting statute has been satisfied.”). In this case, the district court specifically considered “the stellar performance of plaintiffs’ counsel throughout this long and difficult case” in approving their hourly rates, which reached $450 and $495 for lead counsel. Kenny A. III, 454 F.Supp.2d at 1284, 1286. Having already been used to increase the lodestar amount, the quality of the services rendered and superior performance could not also be used to enhance the award above the lodestar amount that had been calculated using those higher rates. See Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098. Double counting simply is not allowed. It was after the successful settlement had been achieved, after they knew how well they had performed, after they knew the degree of difficulty involved, and after they knew how much capital resources and legal effort had been required that the plaintiffs’ attorneys asked for hourly rates they thought that they deserved. Over the objections of the defendants, the district court gave them those rates in full. It did not reduce a single hourly rate for any attorney or paralegal by so much as a penny. In requesting their hourly rates, the lead counsel for the plaintiffs represented that those rates were adequate, subject to considerations they specified, to compensate them. Marcia Robinson Lowry, one of the two lead counsel, testified by affidavit that: The standard hourly rates reflected in Exhibit 2 are fair, reasonable, and consistent with hourly rates in the Atlanta market for the price of legal services of comparable quality rendered in cases demanding similar skill, judgment and performance. These standard hourly rates do not, however, take into account the fact that class counsel was required in this case to advance the entire $1.7 million expense of prosecuting this case for the benefit of the class, or the fact that class counsel’s compensation was totally contingent upon prevailing in this action. Jeffrey O. Bramlett, the other lead counsel, testified by affidavit that: “The hourly rates set forth in Exhibit 1 correctly reflect the hourly rates my law firm currently charges and collects from clients who hire us to perform legal services on a Standard Hourly Rate basis.” However, he further testified: Standard Hourly Rates are predicated on the assumptions that the client will pay in full on a 30-60 day cycle, that counsel is not required to shoulder any significant financial risk of unreim-bursed case expense, and that counsel will be paid currently regardless of the result ultimately achieved. Here, Class Counsel was forced to advance case expenses approaching $1.7 million to protect the Class’ interests, largely because State Defendants refused Plaintiffs’ suggestion of a neutral case record review that would have saved both sides (and ultimately the State) duplicative cost. Here, Class Counsel was forced to personally advance Plaintiffs’ portion of that cost in the face of State Defendants’ vigorous assertion that it would not and could not be recovered. Here, Class Counsel’s recovery of any cost, let alone fee, was utterly dependent upon the contingency of a successful result. Here, Class Counsel was forced to invest more than $8.85 million of professional time and out-of-pocket expense over a three- and-one-half year period with no assurance of any recovery. There is no indication that Mr. Bramlett’s firm’s regular hourly rates are anywhere near the low side of the Atlanta legal market or even the midpoint. The position that the two lead counsel took in their affidavits is that the hourly-rates used to calculate the lodestar figure were the same rates paying clients would have been charged. They each represented that those rates would fully compensate them except for three facts that the district court listed in its opinion: “(1) the fact that class counsel were required to advance case expenses of $1.7 million over a three-year period with no ongoing reimbursement, (2) the fact that class counsel were not paid on an on-going basis as the work was being performed, and (3) the fact that class counsel’s ability to recover a fee and expense reimbursement were completely contingent on the outcome of the case.” Kenny A. III, 454 F.Supp.2d at 1288. 2. The first two of those three considerations are delayed payment factors, and if they were permissible bases for enhancing a fee award, an enhancement would be required in virtually every class action covered by one of the traditional federal fee-shifting statutes. It is a rare § 1983 action where those whose rights have been violated finance the cost of the litigation. Attorneys for the plaintiffs in these types of cases almost always have to advance the cost of the litigation and do not receive any payment for their time until the case is completed. That is simply the nature of the beast. If the delays in reimbursement for costs and payment for services that these cases inevitably entail justified enhancement, there would be an enhancement in almost every case. We know that cannot be, because the Supreme Court has instructed us that we must employ a strong presumption against enhancements and confine them to the rare and exceptional case. Dague, 505 U.S. at 562, 112 S.Ct. at 2641; Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098. Particularly instructive on this point is some of the Supreme Court’s reasoning in concluding that there could be no enhancement for the contingency nature of a case — for the risk that the plaintiffs’ attorneys would not be paid at all for the costs they advanced and the time they expended. See Dague, 505 U.S. at 563, 112 S.Ct. at 2641. Because there is always a risk that a case will be lost, permitting enhancements based on the degree of risk would mean that calculation of the lodestar would never end the inquiry in a contingent-fee case, and that, the Court decided, would be unacceptable. Id. Likewise, permitting enhancements for the delay in compensation for expenses incurred and services rendered would mean that calculation of the lodestar would be merely the first step, not the presumptive last step in the process. That would be unacceptable. Finally, we have said that any delay in payment for professional services rendered is offset by the fact that the hourly rates used are those that prevail at the completion of the case instead of the usually lower rates that were in effect at the time the earlier work was done. See Norman, 836 F.2d at 1302 (“[W]here there is a delay the court should take into account the time value of money and the effects of inflation and generally award compensation at current rates rather than at historic rates.”). In this case, as in most, the hourly rates used were those in effect after the case had been completed. (See R32:495:Ex.C^ 6) (“The hourly rates set forth in Exhibit 1 correctly reflect the hourly rates my firm currently charges and collects from clients who hire us to perform legal services on a Standard Hourly Rate basis.”). That covers two of the three factors the district court relied on in determining that the hourly rates were underinclusive and an enhancement was warranted. Kenny A. III, 454 F.Supp.2d at 1288. 3. The third factor was the contingent nature of the case. Enhancing a lodestar based on contingency is flatly forbidden by the Dague decision. The district court was under the mistaken impression that the Dague rule forbids only an enhancement based on contingency alone. Id. at 1288 n. 8. The rule is not so limited. The Supreme Court’s description of the scope of its holding leaves no wiggle room. The Court stated that it was “concluding that no contingency enhancement whatever is compatible with the fee-shifting statutes at issue.” Dague, 505 U.S. at 565, 112 S.Ct. at 2643. In case anyone missed the point of that, the Court unequivocally stated in the penultimate sentence of its opinion: “[W]e hold that enhancement for contingency is not permitted under the fee-shifting statutes at issue.” Id. at 567, 112 S.Ct. at 2643-44. Period. Even if we were to go beyond that clear language of the Supreme Court’s command, the reasons it gave for concluding that contingency enhancements are incompatible with the fee-shifting statutes apply with equal force to using contingency as one of several reasons for an enhancement. For example, taking the risk of loss into account both in the lodestar and in an enhancement amounts to double counting the difficulty of the case, id. at 562-63, 112 S.Ct. at 2641, regardless of whether contingency is the sole enhancement factor. No double counting means no double counting. Allowing contingency to figure into an enhancement to any extent would also provide attorneys with an incentive to bring relatively meritless claims. See id. at 563, 112 S.Ct. at 2642; see also id. at 566, 112 S.Ct. at 2643 (“Contingency enhancement would make the setting of fees more complex and arbitrary, hence more unpredictable, and hence more litigable.”). That, the Court reasoned, would be bad. Dague, 505 U.S. at 566, 112 S.Ct. at 2643. In summary, none of the three factors that the district court enumerated as a basis for determining that the professional services rendered by plaintiffs’ counsel were “far superior to what consumers of legal services in the legal marketplace in Atlanta could reasonably expect to receive for the rates used in the lodestar calculation,” Kenny A. III, 454 F.Supp.2d at 1288, is a permissible basis for enhancing the lodestar amount. An enhancement based on them is inconsistent with controlling Supreme Court decisions. 4. Making a further attempt to justify the enhancement, the district court also stated that: In addition, based on its personal observation of plaintiffs’ counsel’s performance throughout this litigation, the Court finds that the superb quality of their representation far exceeded what could reasonably be expected for the standard hourly rates used to calculate the lodestar. Quite simply, plaintiffs’ counsel brought a higher degree of skill, commitment, dedication, and professionalism to this litigation than the Court has seen displayed by the attorneys in any other case during its 27 years on the bench. The foster children of Fulton and DeKalb Counties were indeed fortunate to have such unparalleled legal representation, and the Court would be remiss if it failed to compensate counsel for this extraordinary level of service to their clients. Id. at 1288-89. This rationale is little more than a restatement of the district court’s position that the lodestar amount should be enhanced because of the quality of the representation. Two points about that. First, the district court itself found that plaintiffs’ attorneys had submitted vague records and overbilled in nine of fifteen categories, forcing the court to cut more than 4,700 hours off those they claimed, amounting to a reduction of more than $1 million of billable time. See supra at 1217-18. And, as we have indicated, if anything, the district court was too kind to plaintiffs’ attorneys in that respect. See supra at 1219-20. Any kindness aside, in Delaware Valley I the Supreme Court held that the “elimination of a large number of hours on the grounds that they were unnecessary, unreasonable, or unproductive is not supportive of the court’s later conclusion that the remaining hours represented work of ‘superior quality.’ ” Delaware Valley I, 478 U.S. at 567, 106 S.Ct. at 3099. See generally id. at 554 n. 2, 106 S.Ct. at 3093 n. 2. The holding reflects the common sense idea that bad and excessive billing is inconsistent with superb law-yering. The district court failed to consider that inconsistency in its findings. The second, and more fundamental, point about the district court’s reliance on the quality of the representation as a ground for enhancing the lodestar amount is that it amounts to double counting and is contrary to the Supreme Court’s decision in Delaware Valley I. As we have already explained, in that decision the Court reversed an enhancement for superior performance, holding that “the special skill and experience of counsel” and “the quality of representation” are “presumably fully reflected in the lodestar amount, and thus cannot serve as independent bases for increasing the basic fee award.” Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098 (quotation and citation omitted); see also id. at 566, 106 S.Ct. at 3098 (“In short, the lodestar figure includes most, if not all, of the relevant factors constituting a ‘reasonable’ attorney’s fee, and it is unnecessary to enhance the fee for superior performance in order to serve the statutory purpose of enabling plaintiffs to secure legal assistance.” (emphasis added)). The final reason the district court gave for the $4.5 million enhancement was that “plaintiffs’ success in this case was truly exceptional.” Kenny A. III, 454 F.Supp.2d at 1289. The result was so good, the court said, that: “After 58 years as a practicing attorney and federal judge, the Court is unaware of any other case in which a plaintiff class has achieved such a favorable result on such a comprehensive scale.” Id. at 1290. Not only that, but the district court also said that “even if the plaintiffs had prevailed in a trial of this case, it is doubtful that they would have obtained relief as ‘intricately detailed and comprehensive’ as that contained in the Consent Decree.” Id. at 1289-90. To the extent that the district court rewarded plaintiffs’ counsel with an enhancement for obtaining better results than the class would have received had the case been resolved on the merits, that is plainly wrong. The purpose of the fee-shifting statutes, as the Supreme Court explained in Delaware Valley I, is “to enable private parties to obtain legal help in seeking redress for injuries resulting from the actual or threatened violation of specific federal laws.” Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098. The statutes are designed to ensure that civil rights plaintiffs are adequately represented by counsel. See Hensley, 461 U.S. at 429, 103 S.Ct. at 1937 (“The purpose of § 1988 is to ensure ‘effective access to the judicial process’ for persons with civil rights grievances.” (quoting H.R.Rep. No. 94-1558, at 1 (1976))). For that reason, a fee award should “result[] in fees which are adequate to attract competent counsel, but which do not produce windfalls to attorneys.” Blum, 465 U.S. at 893-94, 104 S.Ct. at 1546 (quoting S.Rep. No. 94-1011, at 6 (1976), as reprinted in 1976 U.S.C.C.A.N. 5908, 5913). Stated from the other direction, the fee-shifting statutes are not designed to provide representation that will win plaintiffs more than a correct application of substantive and remedial law entitles them to receive. Their purpose is not to provide representation that will secure settlement relief that is more “intricately detailed and comprehensive,” Kenny A. III, 454 F.Supp.2d at 1289-90, than the plaintiffs would have received if their claims had been litigated to final judgment. The purpose of the statutes, most assuredly, is not to provide plaintiffs with representation that dazzles or bedazzles the district court judge. Fee awards should be calculated in a way that furthers the purpose of the fee-shifting statutes; they should not be used to encourage or reward results that go beyond that purpose. See Blum, 465 U.S. at 893-95, 104 S.Ct. at 1546-47. Some of the discussion in the Dague opinion is useful in understanding why enhancements should not be given for merits-exceeding results. As we have recounted, in that case the Supreme Court rejected the idea of enhancements for contingency or risk of loss in part because permitting them would provide attorneys with some incentive to bring relatively meritless claims. Dague, 505 U.S. at 563, 112 S.Ct. at 2641-42. The social cost of encouraging attorneys to press claims of dubious merit would be too high, the Court reasoned, and providing attorneys with incentives to bring them is “an unlikely objective of the ‘reasonable fees’ provision.” Id. at 563, 112 S.Ct. at 2643. A result that obtains more or better relief than plaintiffs are entitled to receive under the law is, to the extent it exceeds their entitlement on the merits, analogous to relief on a meritless claim. Just as Dague instructs us that fee awards should not underwrite efforts to obtain relief where none is due under the law, neither should they underwrite efforts to receive more or better relief than that due under the law. Just as the societal costs for fee awards for non-meritorious claims are too high, so also are they too high for results that exceed what the law allows. Just as encouraging non-meritorious claims cannot have been an objective of the fee-shifting provisions, neither can encouraging results that go beyond what the law allows have been an objective. To put it in an either-or manner, superb results are either what a fair application of the law produces, which means that they are not truly “superb,” or they are results that exceed what the law allows and for that reason are beyond the purpose of the fee-shifting statutes. Those statutes are designed to provide a reasonable fee for a reasonable result, not an extraordinary fee for a result that goes beyond what the law would provide if the claims were litigated to their correct conclusion on the merits. Look at it this way. A merits-exceeding result for plaintiffs must be the product of one, or some combination, of the following factors: superior lawyering by plaintiffs’ counsel, bad lawyering by defendants’ counsel, poor decision making by the court, or dumb luck. The Supreme Court held in Delaware Valley I that superior lawyering by the plaintiffs’ counsel is not a proper basis for an enhancement. Delaware Valley I, 478 U.S. at 565-68, 106 S.Ct. at 3098-100. So, the first possible cause of results that go beyond the merits cannot be used to justify an enhancement. Nor can it plausibly be argued that plaintiffs’ attorneys ought to reap a windfall, and defendants ought to have to pay more in fees than they otherwise should, because of bad lawyering on the defense side. Surely a defendant suffers enough from the additional relief granted against it because of inferior representation without making the defendant pay a surcharge to the other side for the privilege of having been the victim of bad lawyering. Nor can it be argued, with tongue out of cheek, that fees should be increased to reward plaintiffs’ attorneys for being on the side that happens to benefit from bad judging or good luck. That exhausts the possible explanations for excessively favorable results, and none supports awarding an enhancement. 5. The district court relied to some extent on four of the five affidavits that the plaintiffs’ attorneys in this case obtained from their friends who do similar work. Kenny A. III, 454 F.Supp.2d at 1290. The district court found that the affiants were “disinterested Atlanta attorneys.” Id. That finding is clearly erroneous. At oral argument, one of the two lead counsel for the plaintiffs conceded that he not only had recruited these attorneys to provide the affidavits but that in the past he had also supplied some of them with similar affidavits to help boost their own fee awards. Aside from the need to support those who support them, the lawyers who signed the affidavits have a financial interest in keeping the fee award in this case and every case like it as high as possible. The higher this fee award is the more useful it will be as precedent for the lawyer signing the affidavit when he seeks a high fee award in his own cases. The affiants are anything but “disinterested.” The lodestar amount will never suffice for attorneys who practice in this area. They will always believe, in all sincerity, that they deserve more and that the justice system will function better if they are paid more. Lawyers who handle these kinds of cases cannot be disinterested witnesses because they are financially interested. To state this is not to slam lawyers in general or plaintiffs’ lawyers in particular. It simply recognizes that because self-interest is hard-wired into human circuitry, no group is disinterested when it comes to the question of what members of the group are to be paid. Cf. H.L. Mencken, A Little Book in C Major 22 (John Lane Co.1916) (“It is hard to believe that a man is telling the truth when you know that you would lie if you were in his place.”). Aside from the obvious self-interest of the affiants, the contents of the affidavits that were filed in this case are also flawed in other ways. For example, in each of them the affiant solemnly swore that the amount of compensable time claimed in this case was “reasonable,” was “reasonable, fair, and fully compensable,” or was “fair, reasonable, and fully deserving of adequate and full compensation.” Yet, we know those assertions are not true. As we have already noted, the district court found that plaintiffs had inflated the number of hours for which they were due compensation in most of the fifteen billing categories and by a substa