We have consolidated for decision a number of appeals from orders by the district court approving a settlement of consumer-finance class action litigation, denying petitions to intervene, and awarding attorneys' fees. "Federal Rule of Civil Procedure 23(e) requires court approval of any settlement that effects the dismissal of a class-action. Before such a settlement may be approved, the district court must determine that a class-action settlement is fair, adequate, and reasonable, and not a product of collusion." Joel A. v. Giuliani, 218 F.3d 132, 138 (2d Cir.2000). The principal issue presented by these appeals is whether the district judge discharged the judicial duty to protect the members of a class in class-action litigation from lawyers for the class who may, in derogation of their professional and fiduciary obligations, place their pecuniary self-interest ahead of that of the class. This problem, repeatedly remarked by judges and scholars, see, e.g., Culver v. City of Milwaukee, 277 F.3d 908, 910 (7th Cir.2002); Greisz v. Household Bank (Illinois), N.A., 176 F.3d 1012, 1013 (7th Cir.1999); Rand v. Monsanto Co., 926 F.2d 596, 599 (7th Cir.1991); Duhaime v. John Hancock Mutual Life Ins. Co., 183 F.3d 1, 7 (1st Cir.1999); John C. Coffee, Jr., "Class Action Accountability: Reconciling Exit, Voice, and Loyalty in Representative Litigation," 100 Colum. L.Rev. 370,-385-93 (2000); David L. Shapiro, "Class Actions: The Class as Party and Client," 73 Notre Dame L.Rev. 913, 958-60 and n. 132 (1998), requires district judges to exercise the highest degree of vigilance in scrutinizing proposed settlements of class actions. We and other courts have gone so far as to term the district judge in the settlement phase of a class action suit a fiduciary of the class, who is subject therefore to the high duty of care that the law requires of fiduciaries. Culver v. City of Milwaukee, supra, 277 F.3d at 915; Stewart v. General Motors Corp., 756 F.2d 1285, 1293 (7th Cir.1985); In re Cendant Corp. Litigation, 264 F.3d 201, 231 (3d Cir.2001); Grant v. Bethlehem Steel Corp., 823 F.2d 20, 22 (2d Cir.1987).

We do not know whether the $25 million settlement that the district judge approved is a reasonable amount given the risk and likely return to the class of continued litigation; we do not have sufficient information to make a judgment on that question. What we do know is that, as in such cases as In re General Motors Corp. Engine Interchange Litigation, 594 F.2d 1106, 1124 (7th Cir.1979); Ficalora v. Lockheed California Co., 751 F.2d 995, 997 (9th Cir.1985) (per curiam); Holmes v. Continental Can Co., 706 F.2d 1144, 1150-51 (11th Cir.1983), and Pettway v. American Cast Iron Pipe Co., 576 F.2d 1157, 1214, 1218-19 (5th Cir.1978), the judge did not give the issue of the settlement's adequacy the care that it deserved.

This litigation arose out of refund anticipation loans made jointly by the two principal defendants, Beneficial National Bank and H & R Block, the tax preparer. When H & R Block files a refund claim with the Internal Revenue Service on behalf of one of its customers, the customer can expect to receive the refund within a few weeks unless the IRS decides to scrutinize the return for one reason or another. But even a few weeks is too long for the most necessitous taxpayers, and so Beneficial through Block offers to lend the customer the amount of the refund for the period between the filing of the claim and the receipt of the refund. The annual interest rate on such a loan will often exceed 100 percent — easily a quarter of the refund, even though the loan may be outstanding for only a few days. Block arranges the loan but Beneficial puts up the money for it. Not disclosed to the customer is the fact that Beneficial pays Block a fee for arranging the loan and also that Block owns part of the loan.

Beginning in 1990, more than twenty class actions were brought against the defendants on behalf of the refund anticipation borrowers. The suits charged a variety of violations of state and federal consumer-finance laws and also breach of fiduciary duty under state law. Some of the alleged violations appear to be technical. The most damaging charge appears to be that Block's customers are led to believe that Block is acting as their agent or fiduciary, much as if they had hired a lawyer or accountant to prepare their income tax returns, as affluent people do, whereas Block is, without disclosure to them, engaged in self-dealing.

Most of the suits failed on one ground or another; none has resulted in a final judgment against Beneficial or Block. But in the late 1990s several withstood motions to dismiss or motions for summary judgment, and at least one, a Texas suit, was slated for trial.

On September 3, 1997, two lawyers who had prosecuted two of the unsuccessful class actions, Howard Prossnitz and Francine Schwartz, had lunch in Chicago with Burt Rublin, who was and remains Beneficial's lead lawyer in defending against the class-action avalanche. Prossnitz and Schwartz brought with them to the lunch another lawyer, Daniel Harris. Although neither Prossnitz nor Schwartz, nor their friend Harris, had a pending suit against Beneficial (or against Block, which was not represented at the lunch), they discussed "a global RAL settlement" with Rublin. It is doubtful whether Prossnitz or Schwartz even had a client at this time; and certainly Harris did not. Schwartz later "bought" a client from another lawyer, to whom she promised a $100,000 referral fee. The necessity for such a transaction, when the class contains 17 million members, eludes our understanding.

In the hearing before the district judge on the adequacy of the settlement (the "fairness hearing," as it is called), Harris testified that at the lunch Rublin "`threw out' a number, for purposes of illustration, of $24 or $25 million." The judge described this testimony (which he elsewhere describes as "Harris believes he heard Rublin say the case was worth $23 or $24 million"), though it is vociferously denied by Rublin, as "credible." There was, however, no actual settlement negotiation at the lunch.

Prossnitz, Schwartz, and Harris, all solo practitioners, brought a substantial law firm, Miller Faucher and Cafferty LLP, into the picture. In April of the following year the foursome filed two class action suits against Beneficial similar to the others that had been filed and that were (those that hadn't flopped) wending their way through the courts of various states. One of the two suits filed also named as defendants H & R Block and three affiliated Block entities, but three of those, including Block itself, were voluntarily dismissed from the suit by the plaintiffs in October 1998 and the fourth was dismissed in February 1999. Shortly after the suits were filed, Harris made a settlement offer to Beneficial that was rejected, but after a hiatus negotiations began. Block was included in the settlement negotiations, despite the fact that there were by then no claims pending against it. It was included because Beneficial was reluctant to settle without Block, having promised to indemnify it for any liability resulting from Block's role in Beneficial's refund anticipation loans.

In October of 1999, a class jointly represented by the three solo practitioners and the Miller firm (we'll call these the "settlement class lawyers"), plus Beneficial and Block, entered into a settlement agreement which they submitted to the district court for its approval. The agreement contemplated the filing of an amended complaint naming H & R Block as a defendant, and by its terms covered claims against five Block entities, of which four were the entities originally named but subsequently dismissed as defendants in one of the two original class action complaints. The agreement defined the class as all persons who had obtained refund anticipation loans from Beneficial between January 1, 1987, and October 26, 1999, and provided for the release of all claims "arising out of or in any way relating to the tax refund anticipation loans (`RALs,' sometimes erroneously referred to as `Rapid Refunds') obtained by the Class at any time up to and through" that date. The defendants agreed to create a fund of $25 million against which members of the class could file a claim not to exceed $15. Any money left in the fund after the expiration of the period for filing claims was to revert to the defendants, who also agreed to injunctive relief in the form of certain required disclosures to future customers, primarily of the financial arrangements between Beneficial and Block, and to bear the cost of notice to class members and of the class counsel's legal fees out of their own pockets rather than out of the settlement fund. One RAL class action, the Basile suit pending in the Pennsylvania courts, was excluded from the agreement, apparently because Block thought it could get the supreme court of that state to reverse a lower court decision that had gone against the company. Beneficial and Block agreed to split the expense of the settlement 50-50.

The district judge approved the settlement except for the reversion and the $15 cap, which at his insistence the parties raised to $30 for those members of the class (apparently the vast majority) who had received two or more tax refund anticipation loans from Block. With these changes the settlement was approved and notices mailed to 17 million persons — most of whom ignored them; several million of the notices, moreover, were undeliverable, presumably because the addressees had moved and left no forwarding address. Only 1 million of the recipients filed claims, which would be enough, however, to exhaust the settlement fund. Only about 6,000 of the recipients opted out of the class action so that they could seek additional relief against the defendants.

Incidentally, there is an unremarked conflict of interest within the class, between those class members who took out one or two refund anticipation loans and those who took out more than two and thus will receive no compensation for the additional damages that they incurred. Conflicts of interest can create serious problems for class action settlements, see, e.g., Amchem Products, Inc. v. Windsor, 521 U.S. 591, 626-27, 117 S. Ct. 2231, 138 L. Ed. 2d 689 (1997); Retired Chicago Police Ass'n v. City of Chicago, 7 F.3d 584, 598 (7th Cir.1993), and require the creation of separately represented subclasses. But in light of the modesty of the stakes even of class members who had multiple refund anticipation loans and the expense of subdividing the class (and how many subdivisions would be necessary to reflect the full range of damages?), we are not disposed to regard this particular defect in the settlement as fatal.

In finding that $25 million was an adequate settlement, the judge relied in part on an unsworn report by James Adler, an accountant who purported to estimate the damages caused by the defendants' alleged violations of law. He was not deposed or subjected to cross-examination and the judge did not discuss the adequacy of his methodology. Adler came up with a figure of $60 million, but it is unclear whether this was intended to be an estimate of the entire damages that the class might hope to recover if the case was tried and went to judgment and what legal assumptions underpinned the estimates.

The various objectors to the settlement, primarily intervening or would-be intervening plaintiffs who have claims that the settlement will release, contend that the settlement agreement is the product of a "reverse auction," the practice whereby the defendant in a series of class actions picks the most ineffectual class lawyers to negotiate a settlement with in the hope that the district court will approve a weak settlement that will preclude other claims against the defendant. Blyden v. Mancusi, 186 F.3d 252, 270 n. 9 (2d Cir.1999); Coffee, supra, at 392; Samuel Issacharoff, "Governance and Legitimacy in the Law of Class Actions," 1999 Sup. Ct. Rev. 337, 388; Marcel Kahan & Linda Silberman, "The Inadequate Search for `Adequacy' in Class Actions: A Critique of Epstein v. MCA, Inc.," 73 N.Y.U. L.Rev. 765, 775 (1998); John C. Coffee, Jr., "Class Wars: The Dilemma of the Mass Tort Class Action," 95 Colum. L.Rev. 1343, 1370-73 (1995). The ineffectual lawyers are happy to sell out a class they anyway can't do much for in exchange for generous attorneys' fees, and the defendants are happy to pay generous attorneys' fees since all they care about is the bottom line — the sum of the settlement and the attorneys' fees — and not the allocation of money between the two categories of expense. The defendants agreed to pay attorneys' fees in this case, to the three solo practitioners and the law firm that negotiated the settlement, of up to $4.25 million.

Although there is no proof that the settlement was actually collusive in the reverse-auction sense, the circumstances demanded closer scrutiny than the district judge gave it. He painted with too broad a brush, substituting intuition for the evidence and careful analysis that a case of this magnitude, and a settlement proposal of such questionable antecedents and circumstances, required. The initial agreement submitted for the judge's approval, remember, had provided for a reversion and also capped each class member's recovery at $15. If the parties had an inkling that only 1 million class members would file claims, they were agreeing to a settlement worth only $15 million, and probably less; for if 1 million class members filed claims capped at $30, fewer would have filed claims capped at $15. Yet according to a credibility determination by the district judge that we are not in a position to second guess, two and half years earlier, before RAL plaintiffs began having some success in the courts, Beneficial's counsel had indicated that $23 to $25 million were ballpark figures for a settlement with Beneficial alone. Beneficial's share of a $15 million settlement in which Block was a codefendant would be only $7.5 million (remember that Beneficial and Block agreed to split the cost of the settlement 50-50) — yet that is the settlement the lawyers for the settlement class agreed to, plus injunctive relief the value of which no one has attempted to monetize and which is barely discussed in the briefs or by the judge. The injunctive relief signally does not include a requirement that H & R Block disclose its interest in Beneficial's refund anticipation loans.

Moreover, H & R Block appears to have faced substantial exposure in a Texas class action in which it was accused of breach of fiduciary obligations to its customers. The class in that suit was seeking disgorgement of all the fees paid to Block by the banks that made refund anticipation loans through it. The class argued that such a forfeiture was mandatory if Block was found to have violated its fiduciary duties. Disgorgement was also sought of all other fees that Block had received "in connection with each RAL transaction" — that is, the tax-preparation and electronic-filing fees that Block had charged its RAL customers to file their taxes for them — a form of relief that the class claimed was within the trial court's equitable discretion. The total amount sought could have reached $2 billion. The class had been certified, the case was proceeding in the Texas courts, and the theory of liability and damages could not be dismissed as frivolous; indeed, the case had been set for trial. Even if the class had only a 1 percent chance of prevailing, the expected value of its suit might reach $20 million. (This is on the unrealistic assumption that the only possible outcomes were a $2 billion judgment and a zero judgment. Realistic intermediate possibilities would make the $20 million estimate expand.)

Remarkably in view of the progress and promise of the Texas suit relative to the half-hearted efforts of the settlement class counsel, the district judge enjoined the Texas suit on the authority of the All Writs Act, 28 U.S.C. § 1651(a), reasoning that the suit might upend the settlement. In re VMS Securities Litigation, 103 F.3d 1317, 1323-24 (7th Cir.1996); In re Agent Orange Product Liability Litigation, 996 F.2d 1425, 1431-32 (2d Cir.1993); In re Baldwin-United Corp. (Single Premium Deferred Annuities Ins. Litigation), 770 F.2d 328, 335-38 (2d Cir.1985). The effect of the injunction is that the settlement release, if upheld, would release the claims in the Texas suit. For this release of potentially substantial claims against H & R Block the settlement class received no consideration. In fact the settlement class received no consideration for the release of any claims against Block. The only effect of bringing Block into the settlement was to allow Beneficial to cut its own expense of the settlement in half. The lawyers for the settlement class were richly rewarded for negotiations that greatly diminished the cost of settlement to Beneficial from the level that it had considered to be in the ballpark years earlier when the cases were running more in its favor than when the settlement agreement was negotiated. In effect, the settlement values the Texas and all other claims against Block at zero.

The district judge enjoined the lawyers for the Texas class from notifying the members of that class of the status of the Texas litigation to assist them in deciding whether to opt out of the settlement that the settlement class counsel had negotiated with Beneficial and Block and continue to litigate in the Texas courts. The judge should not have done this, especially since opting out was likely to be the sensible course of action given the ungenerosity of the settlement to the Texas class. A pattern of withholding information likely to undermine the settlement emerged when, after approving the settlement, the district judge encouraged the solo practitioners to submit their fee applications in camera, lest the paucity of the time they had devoted to the case (for which the judge awarded them more than $2 million in attorneys' fees) be used as ammunition by objectors to the adequacy of the representation of the class. There was no sound basis for sealing the fee applications, let alone for sealing the number of hours each of the settlement class counsel had devoted to the case. The applications are not in the appellate record and we do not know what the total number of hours devoted by the class counsel to this litigation was, but apparently it was a small number. This is not surprising, since the lawyers' efforts between the filing of the complaint and the settlement negotiations were singularly feeble, illustrated by their responding to the Block defendants' motion to dismiss for lack of personal jurisdiction with a voluntary dismissal of the claims against those defendants. Their representation of the class was almost certainly inadequate, an independent reason for disapproving a settlement. Ortiz v. Fibreboard Corp., 527 U.S. 815, 856 and n. 31, 119 S. Ct. 2295, 144 L. Ed. 2d 715 (1999); Culver v. City of Milwaukee, supra, 277 F.3d at 913; Linney v. Cellular Alaska Partnership, 151 F.3d 1234, 1238-39 (9th Cir.1998). But in addition it reinforces our concern with the adequacy of the district judge's consideration of the settlement.

The judge approved the settlement primarily because he thought the prospects for the class if the litigation continued were uncertain. They might lose in the end, or win little; and even if they won a lot, the delay in winning would make the relief eventually awarded the class worth much less in present-value terms. To most people, a dollar today is worth a great deal more than a dollar ten years from now. It is especially likely to be worth more to the members of the class in this litigation. Only a person with a very high discount rate (that is, a strong preference for present over future dollars—a preference that may reflect desperation rather than fecklessness or shortsightedness) would borrow at an astronomical interest rate in order to get a sum of money now rather than a few weeks from now.

All this is true, but in the suspicious circumstances that we have recited the judge should have made a greater effort (he made none) to quantify the net expected value of continued litigation to the class, since a settlement for less than that value would not be adequate. Determining that value would require estimating the range of possible outcomes and ascribing a probability to each point on the range, In re General Motors Corp. Engine Interchange Litigation, supra, 594 F.2d at 1132-33 n. 44, though as just noted those outcomes must be discounted to the present using a reasonable, and in this case perhaps a steep, interest rate. In re General Motors Corp. Pick-Up Truck Fuel Tank Products Liability Litigation, 55 F.3d 768, 806 (3d Cir.1995); cf. Transcraft, Inc. v. Galvin, Stalmack, Kirschner & Clark, 39 F.3d 812, 819 (7th Cir.1994). We say "perhaps" because even a person with a high discount rate may not care much whether he receives $15 to $30 now or in the future, since it is such a trivial amount of money even to a person who is usually strapped for funds. If, moreover, the court would award prejudgment interest in a case litigated to judgment, discounting might wash out of the picture altogether.

A high degree of precision cannot be expected in valuing a litigation, especially regarding the estimation of the probability of particular outcomes. Still, much more could have been done here without (what is obviously to be avoided) turning the fairness hearing into a trial of the merits. For example, the judge could have insisted that the parties present evidence that would enable four possible outcomes to be estimated: call them high, medium, low, and zero. High might be in the billions of dollars, medium in the hundreds of millions, low in the tens of millions. Some approximate range of percentages, reflecting the probability of obtaining each of these outcomes in a trial (more likely a series of trials), might be estimated, and so a ballpark valuation derived.

Some arbitrary figures will indicate the nature of the analysis that we are envisaging. Suppose a high recovery were estimated at $5 billion, medium at $200 million, low at $10 million. Suppose the midpoint of the percentage estimates for the probability of victory at trial was .5 percent for the high, 20 percent for the medium, and 30 percent for the low (and thus 49.5 percent for zero). Then the net expected value of the litigation, before discounting, would be $68 million; discounting, depending on an estimate of the likely duration of the litigation, would bring this figure down, though probably not to $25 million — and any discounting might be inappropriate, as we explained. These figures are arbitrary; our point is only that the judge made no effort to translate his intuitions about the strength of the plaintiffs' case, the range of possible damages, and the likely duration of the litigation if it was not settled now into numbers that would permit a responsible evaluation of the reasonableness of the settlement.

Two classes were absorbed into the settlement even though their claims were sharply different from those of the classes represented by the settlement counsel. A class action brought by Belinda Peterson years before the suit that gave rise to the settlement complained of Block's promise, for which it levied a separate charge, that a customer would receive a "rapid refund" from the IRS. The Peterson class alleges that Block knew that customers such as Peterson who were seeking a refund on the basis of the Earned Income Tax Credit would not receive it rapidly because the IRS was subjecting this class of refund requests to special scrutiny. No loan was involved. The district judge nevertheless held that the Peterson class was embraced by the settlement and release, meaning that the members of the class would be entitled to damages of $15 apiece, period. The settlement class counsel had never complained about, or so far as appears knew anything about, the rapid refunds, and indeed Harris testified that the Peterson class was intended to be excluded from the settlement. "Rapid Refunds" are mentioned in the release only because the term is sometimes used to describe refund anticipation loans, an entirely different practice from Block's promise of a rapid refund. Whatever injury the promise caused Block's customers is not measured by the $15 award, which was an estimate of the injury inflicted by violations connected with refund anticipation loans. The Peterson class was included in the settlement as a result of a purely semantic coincidence.

Another class that got swept up in the settlement as it were accidentally, the Carbajal class, was complaining about the defendants' "intercepting" IRS refunds in order to offset debts owed by the customer on previous refund anticipation loans. Block would apply to the IRS for a refund for its customer but direct that the refund be paid to it, and it would then deduct whatever the customer owed it and its partner lending institutions and forward the balance (if any) to the customer. This practice was distinct from nondisclosure, misrepresentation, and other alleged wrongs concerning the making of refund anticipation loans. The relation was closer than in the case of the rapid refunds, because the interception practice, though not involving a loan as such, involved an effort to collect debts based on past refund anticipation loans. But the $15 damages award that the members of the Carbajal class received as part of the settlement that the district judge approved was just as unrelated to whatever injury the interception inflicted on them.

All things considered, we conclude that the district judge abused his discretion in approving the settlement. Because of this conclusion, the other issues raised by the appeals need not be decided, but for guidance on remand we will address the principal ones, which concern attorneys' fees. To begin with, we disapprove the practice (a practice we had never heard of and can find no case law concerning) of encouraging or permitting the submission of fee applications in camera. In the unlikely event that some confidential information is contained in the applications, that information can be whited out. To conceal the applications and in particular their bottom line paralyzes objectors, even though inflated attorneys' fees are an endemic problem in class action litigation and the fee applications of such attorneys must therefore be given beady-eyed scrutiny by the district judge. Gunter v. Ridgewood Energy Corp., 223 F.3d 190, 192 (3d Cir. 2000); 7B Charles Alan Wright, Arthur R. Miller & Mary Kay Kane, Federal Practice and Procedure § 1803, pp. 510-11 (2d ed.1986). Second, class counsel's compensation must be proportioned to the incremental benefits they confer on the class, not the total benefits. Supposing that with absolutely minimal effort the class counsel could, as appears to be the case, have obtained a settlement for $20 million back in 1997 when they met with Rublin, the question would be how much they should be rewarded for having pushed the settlement up to $25 million. Surely not the $4.25 million that the judge pursuant to the agreement between the parties to the settlement awarded, especially since the class counsel, but for prodding by the judge, would have settled for less than Rublin appears to have been prepared to offer years earlier.

Several lawyers, each representing a class member, appeared at the fairness hearing to object to various features of the proposed settlement, primarily the reversion which the judge later struck. They wanted a fee for having conferred a benefit on the class by arguing successfully against the reversion. The judge turned them down.

An initial question is whether we have jurisdiction of their appeals, since their clients, although members of the class, were denied intervention in the district court and are not appealing that denial. These plaintiffs say they have no reason to be parties at this stage because they are content with the revised settlement, which axed the reversion. Ordinarily only a party to a litigation, including a class action, can appeal. Marino v. Ortiz, 484 U.S. 301, 304, 108 S. Ct. 586, 98 L. Ed. 2d 629 (1988) (per curiam); Karcher v. May, 484 U.S. 72, 77, 108 S. Ct. 388, 98 L. Ed. 2d 327 (1987); In re Navigant Consulting, Inc., Securities Litigation, 275 F.3d 616, 617-18 (7th Cir.2001). (Not all courts agree that a class member can appeal only if he has intervened and thus become a named party; the line up is reviewed in Scardelletti v. Debarr, 265 F.3d 195, 205-06, 209 (4th Cir.), cert. granted, ___ U.S. ___, 122 S. Ct. 663, 151 L. Ed. 2d 578 (2001).) But to this as to most legal generalizations there are exceptions — see, e.g., United States Catholic Conference v. Abortion Rights Mobilization, Inc., 487 U.S. 72, 76, 108 S. Ct. 2268, 101 L. Ed. 2d 69 (1988), recognizing the right of third-person witnesses, not named parties to the case, to appeal from a contempt citation for noncompliance with subpoenas served on them by a named party — most relevantly for cases in which a lawyer appeals from a sanction imposed on him, Vollmer v. Publishers Clearing House, 248 F.3d 698, 701, 705, 711 (7th Cir.2001); Corroon v. Reeve, 258 F.3d 86, 88, 90, 92 (2d Cir.2001), or from the denial of a motion for fees. Gaskill v. Gordon, 160 F.3d 361, 362-63 (7th Cir.1998); Florin v. Nationsbank of Georgia, N.A., 34 F.3d 560, 562 and n. 1 (7th Cir.1994); In re Continental Illinois Securities Litigation, 962 F.2d 566, 568 (7th Cir.1992). The lawyer doesn't have to move to intervene in the district court in order to appeal to us; nor does the objector have to move to intervene in order to appeal the lawyer's fee. Powers v. Eichen, 229 F.3d 1249, 1251, 1256 (9th Cir.2000); Rosenbaum v. MacAllister, 64 F.3d 1439, 1441-43 (10th Cir.1995); see also Zucker v. Occidental Petroleum Corp., 192 F.3d 1323, 1326 (9th Cir.1999). We just the other day rejected the proposition that a lawyer must be made a party before he can be ordered to disgorge a fee wrongfully retained, and though not a party the lawyer was permitted to appeal the order. Dale M. v. Board of Education, 282 F.3d 984 (7th Cir.2002).

The situation here is similar. The clients being content with the settlement, the only issue is whether their lawyers here are entitled to a fee. They are in effect volunteer lawyers for the class asking that they receive a fee for their efforts. We think they can appeal without their clients' having intervened. Intervention would not only be a pointless formality. Powers v. Eichen, supra, 229 F.3d at 1256. It would be a futile one. For the clients, being content with the settlement, could not appeal even if they were parties, because they seek no relief from us and standing must continue throughout the entire litigation. United States Parole Comm'n v. Geraghty, 445 U.S. 388, 397, 100 S. Ct. 1202, 63 L. Ed. 2d 479 (1980); Levin v. Attorney Registration & Disciplinary Comm'n, 74 F.3d 763, 767 and n. 5 (7th Cir.1996); Chong v. District Director, INS, 264 F.3d 378, 383-84 (3d Cir.2001). Unless the lawyers can appeal, there will be no appellate review of the district judge's decision not to award them fees.

It would be a different case if the claim for attorneys' fees rested on a fee-shifting statute, so that the money would come from the defendants and not diminish the $25 million fund. Then the objectors would have to intervene, because it is the litigants rather than the lawyers who hold the entitlement to awards under fee-shifting statutes. Central States, Southeast & Southwest Areas Pension Fund v. Central Cartage Co., 76 F.3d 114, 116 (7th Cir. 1996). The objectors might be obliged by contract to pay these sums to their lawyers, but that would not make the lawyers parties. To be entitled to appeal, however, the objectors would have to become parties.

But in fact the lawyers are claiming fees in their own name, so that any fees awarded to them would come from the $25 million fund under the common-fund doctrine. When a lawyer lays claim to a portion of the kitty, he becomes a real party in interest; and should he therefore have to intervene — not only for purposes of taking an appeal himself but so that he will be an appellee if class members oppose the district court's award and want the money back? We do not see why a person who has received money by order of the district court at the expense of a party must be named as a party in order for the party harmed by the order to be able to appeal. After all, a defendant could appeal from a fee award to the plaintiff's lawyer that he thought excessive.

We need not pursue this interesting question, on which the Supreme Court may shed considerable light when it decides the Scardelletti case, any further, since the claim for attorneys' fees falls with the settlement. But assuming these lawyers can appeal, let us for the sake of guidance for the future consider the merits of their appeals.

The law generally does not allow good Samaritans to claim a legally enforceable reward for their deeds. Nadalin v. Automobile Recovery Bureau, Inc., 169 F.3d 1084, 1086 (7th Cir.1999); Saul Levmore, "Explaining Restitution," 71 Va. L.Rev. 65 (1985). But when professionals render valuable albeit not bargained-for services in circumstances in which high transaction costs prevent negotiation and voluntary agreement, the law does allow them to claim a reasonable professional fee from the recipient of their services. Gaskill v. Gordon, supra, 160 F.3d at 363; In re Continental Illinois Securities Litigation, supra, 962 F.2d at 568, 571; Levmore, supra, at 66. That is the situation of objectors to a class action settlement. It is desirable to have as broad a range of participants in the fairness hearing as possible because of the risk of collusion over attorneys' fees and the terms of settlement generally. This participation is encouraged by permitting lawyers who contribute materially to the proceeding to obtain a fee. Gottlieb v. Barry, 43 F.3d 474, 490-91 (10th Cir.1994); Fisher v. Procter & Gamble Mfg. Co., 613 F.2d 527, 547 (5th Cir. 1980); White v. Auerbach, 500 F.2d 822, 828 (2d Cir.1974).

The principles of restitution that authorize such a result also require, however, that the objectors produce an improvement in the settlement worth more than the fee they are seeking; otherwise they have rendered no benefit to the class. Class Plaintiffs v. Jaffe & Schlesinger, P.A., 19 F.3d 1306, 1308 (9th Cir.1994) (per curiam); see also Stewart v. General Motors Corp., supra, 756 F.2d at 1294; Petrovic v. Amoco Oil Co., 200 F.3d 1140, 1156 (8th Cir.1999).

The judge denied a fee to the objectors in part on the ground that he had already decided, without telling anybody, not to accept the reversion. But objectors must decide whether to object without knowing what objections may be moot because they have already occurred to the judge. A compelling ground for denying the objectors a fee in this case does exist, however. It is that other lawyers at the hearing, notably counsel for the Carnegie intervenors, had vigorously objected to the reversion; the objectors added nothing. In re Synthroid Marketing Litigation, 264 F.3d 712, 722-23 (7th Cir.2001). For similar reasons the judge was correct to deny intervention to class representatives (Mitchell, Westfall, Ramsey, Jones, Vaughn, and Longo) whose arguments had already been covered by existing plaintiffs. Fed.R.Civ.P. 24(a)(2).

The judgment approving the class action settlement and awarding attorneys' fees is reversed and the case is remanded to the district court for further proceedings consistent with this opinion. The injunction against the Texas class action must be vacated in light of our disapproval of the settlement. See Bradford Exchange v. Trein's Exchange, 600 F.2d 99, 101-02 (7th Cir.1979) (per curiam); cf. Rufo v. Inmates of Suffolk County Jail, 502 U.S. 367, 383, 112 S. Ct. 748, 116 L. Ed. 2d 867 (1992); In re Hendrix, 986 F.2d 195, 198 (7th Cir.1993). The action of a panel of this court in upholding the district judge's interlocutory injunction against the Texas suit, entered while settlement negotiations were in process, does not bear on the validity of the final injunction, which is an incident of the settlement agreement and falls with it.

Finally, we have decided that Circuit Rule 36 shall apply on remand.

REVERSED AND REMANDED.