Fierce competition produces better products at lower prices. Progress that bestows great benefits on consumers is cruel to producers that lag behind their rivals, as Unisys Corporation has. While other manufacturers of computers have grown, Unisys has been shrinking almost from the day it was formed by the merger of Burroughs and Sperry in 1986. In 1989 Unisys gave a pink slip to Edsel Gusman, who for 33 years had serviced and repaired its machines. The jury found that Unisys held Gusman's age, 59, against him and awarded $54,000 in damages for lost wages. Because the jury found that the age discrimination was wilful, the judge doubled this sum; he added $75,000 in front pay, for total damages of $183,000. Gusman's attorneys received a little more than $73,000 for their efforts. Unisys contests the decision on the merits, and Gusman contends that the sum allocated to fees should be increased.

Unisys reduced the size of Gusman's unit from 14 to 12. He does not deny that this was a genuine reduction, caused by the erosion in the firm's business, but he persuaded the jury that age rather than ability or some other proper reason led Unisys to exclude him from the 12. Unisys does not deny that the jury could reach this conclusion, although it does contend that the record does not support a decision that the discrimination was wilful. Because Unisys did not object to the jury instructions defining wilfulness, we need not await the Supreme Court's decision in Hazen Paper Co. v. Biggins, --- U.S. ----, 113 S. Ct. 960, 122 L. Ed. 2d 117 (1993). See also EEOC v. Century Broadcasting Corp., 957 F.2d 1446, 1457-60 (7th Cir.1992). Ample evidence warranted a conclusion that Gusman's immediate superior, Donald Troudt, believed that older workers are inferior--a thought he conveyed not only to his subordinates but also to the lawyers conducting his deposition in this litigation. Troudt made the recommendation that led to Gusman's discharge. Reasonable jurors could conclude that Troudt lied to his superiors about Gusman's skills, ensuring that they would approve his recommendation. (Troudt asserted, for example, that Gusman could repair only the computers that Unisys was phasing out rather than those it was currently selling; actually Gusman's skills lay in the newer batch of machines, and another person had to be trained to do that work after Gusman departed.) An employer cannot escape responsibility for wilful discrimination by multiple layers of paper review, when the facts on which the reviewers rely have been filtered by a manager determined to purge the labor force of older workers. Shager v. Upjohn Co., 913 F.2d 398, 405 (7th Cir.1990).

Unisys does not want Gusman back and does not say that the judge should have awarded reinstatement rather than front pay. Instead it says that the judge should have denied both reinstatement and front pay, insisting that Gusman would have been sacked during the next reduction in staff, or the one after that, even if age accounted for the decision in 1989. Maybe so, but the judge was not obliged to agree. (Front pay, as an alternative to the equitable remedy of reinstatement, is the judge's decision rather than the jury's, Fortino v. Quasar Co., 950 F.2d 389, 398 (7th Cir.1991), but a judge may accept advice from the jury on the subject. Here, as in Price v. Marshall Erdman & Associates, Inc., 966 F.2d 320, 324 (7th Cir.1992), the judge properly endorsed the jury's recommendation after giving the subject independent consideration.) As Gusman was making approximately $45,000 per year in salary and benefits, an award of $75,000 is appropriate if Gusman could have worked 1 1/2 years after (a hypothetical) reinstatement. A judge might conclude that Gusman's skills and seniority would have given him that much breathing space. And front pay was otherwise entirely appropriate. Gusman's human capital was not portable: he could repair Unisys's computers, not IBM's or Apple's. A skilled technician, however, could learn to work on other devices more quickly than a novice; an award of front pay tides him over until Gusman could regain his former productivity.

Only one arrow remains in Unisys's quiver. Sarah Siskind, one of Gusman's lawyers, testified during the trial. The roles of attorney and witness usually are incompatible. A witness is supposed to present the facts without a slant, while an attorney's job is to advocate a partisan view of the significance of the facts. One person trying to do both things is apt to be a poor witness, a poor advocate, or both. See United States v. Trapnell, 638 F.2d 1016, 1025 (7th Cir.1980); United States v. Johnston, 690 F.2d 638, 642-44 (7th Cir.1982) (in banc); ABA, Model Rule of Professional Conduct 3.7(a). Yet the limits on being lawyer and witness in the same case are not absolute. The ABA's Model Rule, for example, makes exceptions for uncontested issues and cases in which disqualification of counsel "would work a substantial hardship on the client". So we must examine the circumstances.

Claims of discrimination are hard to prove one case at a time. An employer can offer some proper explanation for almost any decision. A pattern of treating older (or black, or female) employees worse than others speaks more loudly. The law of large numbers smoothes over the quirks and turns of fate that make finding "the" cause of a particular discharge so hard. Gusman's lawyers obtained discovery that they hoped would enable them to show that Unisys had it in for older employees. Attorney Siskind, who reviewed the materials Unisys furnished, used them to construct a chart that, she believed, demonstrated such a pattern. Unisys wanted to prevent use of the chart and objected on substantive grounds. Unexpectedly (so Gusman's lawyers say) Unisys also objected on foundation grounds. That is, Unisys required Gusman to establish that the chart actually summarized the information Unisys had provided in discovery. Now one would expect that any questions about the provenance of such a chart could be resolved behind the scenes. Gusman's lawyers could show Unisys's lawyers how the chart had been compiled. Whether such a meeting took place we do not know. Unisys demanded that the foundation be laid before the jury. And as Siskind had prepared the chart herself (rather than turning the mass of discovery materials over to a paralegal for duplicative review and compilation), the only way to lay the foundation was by her testimony. So she took the stand, and in two pages of transcript (followed by two pages of cross-examination) described how she had prepared the chart.

As the district court observed, Unisys did not need to force this appearance by counsel. Judge Shabaz suspected, as do we, that Unisys precipitated this episode in the hope of salting the record with reversible error. Well, we are not cooperating. Truculent litigation tactics should be discouraged, not rewarded. It may be that Gusman's team should have taken more precautions, preparing another way to lay the foundation in case Unisys decided to play hardball. Here the principle of "no harm, no foul" governs. Ultimately, the judge excluded the chart on substantive grounds. Siskind's testimony did not present the sort of conflict in roles that underlies the attorney-witness rule. Only errors affecting substantial rights permit reversal in civil litigation. See 28 U.S.C. § 2111; Fed.R.Civ.P. 61. If permitting Siskind to testify was a mistake, that error was harmless.

Having prevailed, Gusman is entitled to collect from Unisys a "reasonable attorney's fee" with which to compensate his lawyers. 29 U.S.C. § 626(b), incorporating 29 U.S.C. § 216(b). Gusman submitted a request for a little more than $103,000, accompanied by time sheets, affidavits, and sample billings to other clients showing rates of $200 per hour for Siskind, $225 per hour for senior partner Charles Barnhill, and $140 per hour for a third lawyer. The affidavits asserted that these are the lawyers' usual rates, which they receive when working for paying clients--not only in Chicago, where their firm has its principal office, but also in Madison, Wisconsin, where the firm has a branch (and where the trial occurred). Unisys did not contest the reasonableness of the time counsel devoted to the litigation but contended that $200 per hour is too steep for Madison. The district judge agreed, reducing the rate for Barnhill and Siskind to $150 per hour (and $125 per hour for the junior litigator). The court allowed that the rates solvent clients are willing to pay Barnhill and Siskind are interesting but added:

[F]ar more persuasive are the affidavits of Madison attorneys suggesting a relevant prevailing rate in the Madison area of $125 to $150 per hour.... This Court is familiar with the billing rates for legal services provided by attorneys [names omitted] and others who litigate discrimination cases in the Madison community. It is of the opinion that it does, indeed, know the value of lawyers' legal services as well as the market itself, which suggests a rate between $125 and $150 per hour.... A reasonable fee should approximate the prevailing market rate in the relevant community.... The prevailing market rate is not based solely upon the prevailing attorneys' hourly rate whether that rate be too high or too low. The regular billing rate of plaintiff's attorneys does not establish the prevailing market rate in the relevant community of Madison, Wisconsin for attorneys of comparable skill and experience.

This passage, like some other discussion in the opinion, suggests that the district judge believes that the approach of Blum v. Stenson, 465 U.S. 886, 104 S. Ct. 1541, 79 L. Ed. 2d 891 (1984), governs all fee-shifting cases.

Blum presented the question whether litigants who engage attorneys employed by legal services organizations are limited to recovering from their adversaries the hourly wages the organizations pay these lawyers. The Court concluded that the ordinary rate for private counsel in the locale, rather than the discounted rate that some attorneys are willing to accept from public-interest organizations, is the appropriate measure in fee-shifting cases. The district judge has concluded that Blum is a universal measure: if plaintiff gets the market average when he hires an attorney with bargain basement fees, then he gets the market average when he hires an attorney with high fees. But Blum does not establish a principle of averaging. It is concerned with two situations: first, the lawyer who does all of his work on contingent fee and thus lacks a regular market rate; second, the lawyer who sacrifices income to assist a favored group of clients:

Some lawyers dedicate their professional lives to causes they find admirable and worthy of support--to legal services for the poor, to the representation of unions. These lawyers are making contributions to their favored causes, not in money but in time. Blum v. Stenson, 465 U.S. 886 [104 S. Ct. 1541, 79 L. Ed. 2d 891] (1984), holds that lawyers who donate their services at bargain rates to legal aid organizations may collect under § 1988 the fees they could obtain if the charitable element were removed. Likewise, Save Our Cumberland Mountains, Inc. v. Hodel, 857 F.2d 1516 (D.C.Cir.1988) (in banc), holds that lawyers who reduce their hourly rates when providing services to environmental plaintiffs may collect the market rate for the time--the rate that the solvent defendants would have paid for work of like quality. These cases, like our own opinion in [In re Continental Illinois Securities Litigation, 962 F.2d 566 (7th Cir.1992) ], hold that the market rate of legal time is the opportunity cost of that time, the income foregone by representing this plaintiff. Using opportunity cost as the measure of legal services means that the value of the lawyer's gift inures to the favored cause, and not to the adversary in litigation.

Barrow v. Falck, 977 F.2d 1100, 1105 (7th Cir.1992). There is no broader principle of averaging. A client who retains a lawyer with an hourly rate of $100, when the average in the community is $150, is entitled to collect from his adversary only $100 for each hour reasonably expended. Ibid. And lawyers who fetch above-average rates are presumptively entitled to them, rather than to some rate devised by the court. Continental Illinois Securities, 962 F.2d at 568; Pressley v. Haeger, 977 F.2d 295, 299 (7th Cir.1992).

Recall that the objective is to find the reasonable fee for the work. When the lawyers sell their time in the market, the market provides the starting point: the lawyer's hourly rate. Eddleman v. Switchcraft, Inc., 965 F.2d 422, 424-25 (7th Cir.1992). Lawyers do not come from cookie cutters. Some are fast studies and others require extra preparation. Some are more nimble on their feet and apt to achieve better results at trial. Some have deeper insight and in a few hours may find ways to prevail (or to curtail costly discovery) that will elude their colleagues. Clients are willing to pay more, per hour, for these better lawyers. A $225 per hour lawyer may end up costing less than a $150 lawyer for the same result or may produce better results for the same total bill. Markets recognize these truths; judges must too. Only an assumption that all lawyers are identical could support the averaging approach, under which all lawyers in a division of the court receive the same hourly fee. Perhaps it would all come out in the wash if the district judge multiplied the average hourly rate by the number of hours an average lawyer would have taken to achieve the same result. If the $225 per hour fee reflects extra productivity, then the judge, after reducing the hourly rate to $150, could increase the number of hours to which it is applied. But such a process of estimating every component of the cost of litigation has nothing to recommend it. Far better to use what the market supplies: actual hours (if reasonably invested), multiplied by actual rates.

Our recent cases have stressed that the best measure of the cost of an attorney's time is what that attorney could earn from paying clients. For a busy attorney, this is the standard hourly rate. If he were not representing this plaintiff in this case, the lawyer could sell the same time to someone else. That other person's willingness to pay establishes the market's valuation of the attorney's services. Notice the assumption built into this way of putting things: if clients willingly pay $X for the time an attorney sells in the market, they would have paid $X for the time sunk in this litigation. Gusman's lawyers ask us to turn this assumption into a rule of law. They receive $225 or $200 or $140 from paying clients in both Chicago and Madison; it follows, they believe, that these are the only rates judges may use in fee-shifting cases. Not so--which is why, in cases such as Eddleman, Continental Securities, Barrow, and Pressley, we have referred to the attorney's billing rate as a presumptive rather than a dispositive figure. Consider two cases: (1) An attorney who ordinarily works 2,000 hours in a year sells 1,900 of these hours to clients who pay $250 per hour and devotes the other 100 hours to civil rights litigation in which the court will fix the fee; (2) An attorney who ordinarily works 2,000 hours in a year sells 100 of these hours to clients who pay $250 per hour and devotes the other 1,900 to civil rights litigation in which the court will fix the fee. The willingness of clients to buy almost all of the attorney's time in Case 1 at $250 per hour implies that he could have sold the rest at that rate or close to it. Any discount needed to fill the remainder of his calendar would not have been steep. Thus $250 per hour is a good estimate of the opportunity cost of the civil rights case. In Case 2 the inference that the attorney could have sold 1,900 hours at $250 per hour is not nearly so strong. Selling the first 5% of one's time is much easier than filling the remaining 95% of the timesheet. Too, most work in fee-shifting cases is compensated only if the client prevails. Even a lawyer who wins 80% of his cases is working, in effect, for 80% of the stated hourly wage. (The possibility of a multiplier to raise the effective rate back to 100% was scotched by Burlington v. Dague, --- U.S. ----, 112 S. Ct. 2638, 120 L. Ed. 2d 449 (1992).) Unless this lawyer is donating the other 20% of the value of his time to a favored cause, his behavior tells us that his market-clearing price is less than $250 per hour, perhaps considerably less. Thus we do not require district judges to leap directly from the willingness of some persons to pay $X to Lawyer Y that $X is "the" hourly rate of Lawyer Y. But this is, as we have said in many other cases and reiterate today, the starting point.

A judge who departs from this presumptive rate must have some reason other than the ability to identify a different average rate in the community. A judge might say, for example, that the lawyers did not display the excellence, or achieve the time savings, implied by their higher rates. A judge might conclude that the plaintiff did not need top-flight counsel in a no-brainer case. But no claim along these lines has been made. Unisys conceded that the hours devoted to the case were appropriate. Because Unisys is still contesting its liability, it cannot characterize this case as a pushover that could have been assigned to junior lawyers. And a party that hired pricey big-city lawyers to defend itself is in no position to contend that only small-town lawyers, at small-town rates, were appropriate. (Beck, Chaet, Loomis, Molony & Bamberger of Milwaukee represented Unisys at trial, and has been joined on appeal by Seyfarth, Shaw, Fairweather & Geraldson of Chicago.) Because of the possibility that plaintiff's lawyers' practice looks more like our hypothetical Case 2 than like Case 1, we do not direct the district court on remand to increase the hourly rates to those counsel have claimed. We do conclude, however, that these are the rates presumptively appropriate, and that Unisys must establish a good reason why a lower rate is essential to assess a "reasonable attorney's fee."

The judgment on the merits is affirmed. The order awarding attorneys' fees is vacated, and the case is remanded for further proceedings consistent with this opinion.

MILLER, District Judge, concurring in part and dissenting in part.

I concur wholeheartedly with respect to the principal judgment awarded to the plaintiff, but respectfully dissent with respect to the attorney fee award. While I admire the majority opinion's effort to breathe greater reason and predictability into the fee-setting process, I do not believe that the district court abused its discretion in awarding fees at a rate lower than was sought, and do not believe that a remand would be workable in any event.

Mr. Gusman obtained his legal services from the firm of Davis, Miner, Barnhill & Galland, a firm based in Chicago with an office in Madison. We are told that Davis, Miner bills for its attorneys' time at the same rates regardless of whether the services are performed in Chicago or Madison. As I read the district court's opinion, it is the tension between billing rates in Chicago and Madison that gives rise to this fee dispute: the plaintiff wishes Unisys to be ordered to compensate his attorneys at Chicago/Madison rates, while the district court believed it to be appropriate to order Unisys to compensate the plaintiff's attorneys at Madison rates. The majority opinion dismisses this issue with the observation that Unisys having "hired pricey big-city lawyers to defend itself is in no position to contend that only small-town lawyers, at small-town rates, were appropriate."

It is well-settled that a fee arrangement between the prevailing party and counsel does not determine the "reasonable" attorney fee for which the losing defendant will be held responsible; the private arrangement is relevant, but not dispositive. Blanchard v. Bergeron, 489 U.S. 87, 93-94, 109 S. Ct. 939, 944-45, 103 L. Ed. 2d 67 (1989); Kirchoff v. Flynn, 786 F.2d 320 (7th Cir.1986). It is difficult to see why the fee arrangement between the losing party and its counsel is entitled to greater preclusive effect. Further, the record indicates that Unisys' Milwaukee trial lawyer was compensated at an hourly rate of $150.00, the same rate as that awarded to Mr. Gusman's principal attorneys; thus, the compensation comparison favors the district court's determination, unless we are to hold that the plaintiff's lawyers are worth half again more simply because they won, a point for which I can find no authority. Cf. Samuels v. American Motors Sales Corp., 969 F.2d 573, 579 (7th Cir.1992) (district court noted defense counsel's $130 hourly rate in process of reducing plaintiff's counsel's hourly rate from requested $250 to $125).

Calculation of a reasonable fee award begins with the familiar "lodestar" calculation required by Hensley v. Eckerhart, 461 U.S. 424, 433-434, 103 S. Ct. 1933, 1939-1940, 76 L. Ed. 2d 40 (1983), whereby the number of hours reasonably expended is multiplied by a reasonable hourly rate. In Jardien v. Winston Network, Inc., 888 F.2d 1151, 1159 (7th Cir.1989), this court stated that, "[i]n a properly functioning market, a 'reasonable fee' should approximate the market rate in the relevant community," and affirmed the award by the district judge, who was "better situated than this court to assess the reasonableness of an hourly rate, because he can consider both the experience of the attorneys involved and the rates charged by other Chicago lawyers." Id. The district judge did that in this case, but had he done no more he would have erred.

More recent cases have indicated that the attorney's regular billing rate must be considered, as well. In Matter of Continental Illinois Securities Litigation, 962 F.2d 566, 568 (7th Cir.1992), a common fund class action suit, the district court reduced the prevailing attorneys' hourly rates, not on the basis of affidavits establishing the market rate for attorneys' services in the community, but on the basis of a belief that higher-priced legal talent simply was not needed in the case. This court, noting that the defendants themselves had hired "a crowd of pricey lawyers to defend the case", disagreed and reversed. Although the court stated that the district court's function was to decide "what the lawyer would receive if he were selling his services in the market rather than being paid by court order" rather than to "determine the equivalent of the medieval just price", nothing in the opinion suggests that the court must or may blind itself to the reasonableness of the requested hourly rate when laid beside rates for other attorneys of comparable skill in the community.

In Pressley v. Haeger, 977 F.2d 295 (7th Cir.1992), the district court was faced with fee petitions from three counsel of record, each of whom stated that his hourly rate was $182; the petitions were supported by an affidavit of a fourth attorney attesting that $182 was the prevailing hourly rate for litigation of that sort and quality in Chicago, where the suit was tried. The defense did not contest that evidence of the prevailing market rate within the community, but argued that those who were not lead counsel should receive less than those who were; the district court agreed. This court reversed and directed entry of a fee award calculated at hourly rates of $182 for each, reminding the district court that "[f]ee litigation under § 1988 depends on what the market rate is rather than what litigants and judges think it ought to be." 977 F.2d at 299. Again, no evidence was submitted to suggest that the market rate in the community in which the legal services were rendered was anything other than that sought by plaintiff's counsel.

Barrow v. Falck, 977 F.2d 1100 (7th Cir.1992), involved a central Illinois attorney who performed legal services for the plaintiff in central Illinois. The attorney sought a fee award based on an hourly rate of $135, and submitted another lawyer's affidavit to the effect that the market rate in central Illinois for attorneys with plaintiff's counsel's skill and experience was $135 per hour. This court reversed the district court's award of fees at the requested hourly rate. The record also reflected that a union paid the plaintiff's attorney for his work at hourly rates ranging from $80 to $110. Stating that judges "must stick to the market rate for the attorneys' time--that is to say, the opportunity costs of their time, the rate they could receive in other engagements," this court held that the market value of the attorney in question was, in light of his demonstrated billing history, no more than $110. 977 F.2d at 1105.

Blum v. Stenson, 465 U.S. 886, 895, 104 S. Ct. 1541, 1547, 79 L. Ed. 2d 891 (1984), teaches that "reasonable fees" under fee-shifting statutes "are to be calculated according to the prevailing market rates in the relevant community...." The majority is correct that Blum did not deal with counsel with historical billable rates, and thus the inquiry cannot stop there when such counsel is involved. But nothing in Continental Illinois Securities, Pressley, or Barrow suggests that counsel's billing history renders inquiry into the prevailing market rates unnecessary or improper. Those cases do not eliminate altogether the Blum Court's conclusion that fee awards "are to be calculated according to the prevailing market rates in the relevant community, regardless of whether plaintiff is represented by private or non-profit counsel." Blum v. Stenson, 465 U.S. at 895, 104 S.Ct. at 1547. They hold only that when determining a reasonable fee, the established billing history of the prevailing party's counsel is entitled to significant weight.

That, as I read it, is what the district court did here. The court noted the submission of other Davis, Miner billings and the rates assessed in those billings, and acknowledged that it was "certainly evidence to be considered by this Court as well. However, far more persuasive are the affidavits of Madison attorneys suggesting a relevant prevailing rate in the Madison area of $125 to $150 per hour." In other words, the court considered both the value placed on Davis, Miner's services by a broader Chicago/Madison market and the value placed on comparable services of comparable attorneys in the narrower Madison market, and found the latter evidence to be more persuasive in the context of this case.

In Chrapliwy v. Uniroyal, Inc., 670 F.2d 760 (7th Cir.1982), cert. denied, 461 U.S. 956, 103 S. Ct. 2428, 77 L. Ed. 2d 1315 (1983), the district court was found to have erred in reducing requested fees based on New York and Washington rates to those prevailing in South Bend, where the case was tried. This court held that in light of what then was perceived to be the complexity of pension litigation, nothing in the record suggested that South Bend counsel could have provided comparable legal services as easily, but recognized that other cases might lead to a contrary result:

If a high priced, out of town attorney renders services which local attorneys could do as well, and there is no other reason to have them performed by the former, then the judge, in his discretion, might allow only an hourly rate which local attorneys would have charged for the same service. On the other hand, there are undoubtedly services which a local attorney may not be willing or able to perform. The complexity and specialized nature of a case may mean that no attorney, with the required skills, is available locally.

* * * * * *

We think that a judge, in allowing an attorney's fee under Title VII or similar statute, has discretion to question the reasonableness of an out of town attorney's billing rate if there is reason to believe that services of equal quality were readily available at a lower charge or rate in the area where the services were to be rendered.

If, however, a party does not find counsel readily available in that locality with whatever degree of skill may reasonably be required, it is reasonable that the party go elsewhere to find an attorney, and the court should make the allowance on the basis of the chosen attorney's billing rate unless the rate customarily charged in that attorney's locality for truly similar services is deemed to require an adjustment.

670 F.2d at 768, 769. The ensuing decade's worth of opinions by this court and the Supreme Court does not appear to have undercut this reasoning. The record contains no suggestion that Mr. Gusman could not have found comparable legal representation at prevailing Madison rates. See Soto v. Adams Elevator Equipment Co., 941 F.2d 543, 553 (7th Cir.1991).

The majority opinion makes the attorney's established hourly rate (when an established rate exists) presumptively appropriate and then shifts to the defense the burden of overcoming that presumption. I believe that this approach suffers from several flaws.

First, the majority's approach distorts the market. It is one thing to say, as in Barrow, that an attorney's billing history within the market in which the services are performed is stronger evidence of the value of his time within that market than general evidence of attorney time in that market. It is quite another thing to allow the attorney's client base to define the market. The Davis, Miner client base reaches well beyond Madison or the Western District of Wisconsin; counsel agree that the bulk of the billings submitted in support of the fee petition were directed to clients based in Chicago. That an attorney's hour is worth $225 in Chicago is evidence, but certainly not conclusive proof, of the same hour's worth in Madison. The hour belongs to a Chicago-based law firm, but the court must decide the worth of that firm's hour to a Madison-based client. Eddleman v. Switchcraft, Inc., 965 F.2d 422, 424 (7th Cir.1992) ("The market rate is 'the rate that lawyers of similar ability and experience in the community normally charge their paying clients for the type of work in question.' ").

Further, the attorney's hourly rate within the community is a factual question that must be resolved before any presumption is to be triggered. A contingent fee attorney with a single client who pays $600 an hour for five hours of legal services each year cannot be entitled to a presumption that his "lodestar" award should be based on an hourly rate of $600. The majority opinion recognizes this through its "Case 1"/"Case 2" example. I fear, however, that we will find the real world closer to Case 2 than to Case 1. I do not believe that many plaintiffs' attorneys in employment discrimination cases sell 95% of their time on an hourly rate; I believe we will find that far more than one hour in twenty is devoted to work on contingency fee or fee-shifting statute cases.

More troubling, though, is the fact-finding process by which courts and counsel will determine whether their case is closer to Case 1 or Case 2. The majority opinion shifts to Unisys the burden of overcoming a presumption, a burden likely to be met only by evaluating some billing history. Whether that evaluation is to focus on Davis, Miner's billing history as a whole, or the firm's billing for services provided in the Madison area, or the billing history of the Madison office, or the billing history of the attorneys who performed services for Mr. Gusman, the evidence will be available to Unisys only through discovery. The Hensley Court's hope that a "request for attorney's fees should not result in a second major litigation", 461 U.S. at 437, 103 S.Ct. at 1941, has proven forlorn, see, e.g., Nanetti v. University of Illinois at Chicago, 944 F.2d 1416 (7th Cir.1991), but this court should not relish the prospect of post-judgment depositions of prevailing plaintiffs' attorneys under subpoena to bring their firms' billing records for the last few years.

Perhaps these concerns could be addressed with only a slight modification to the majority's approach; the burden might be restored to the plaintiff seeking fees, where it ordinarily rests. Blum v. Stenson, 465 U.S. at 895-896 n. 11, 104 S. Ct. at 1547 n. 11; Tomazzoli v. Sheedy, 804 F.2d 93, 96 (7th Cir.1986); Zabkowicz v. West Bend Co., 789 F.2d 540, 548 n. 8 (7th Cir.1986). But if this test is applied, the district court cannot be said to have abused its discretion unless the hourly rate's reasonableness in the community is disregarded altogether. The attorneys' submissions do not indicate that any Madison-based client of Davis, Miner has paid an hourly rate of $225, although two Chicago-based clients paid such a rate in relation to litigation in Madison. One Madison-based client is shown to have paid an hourly rate of $200 for Ms. Siskind's services. Of the twenty-nine billing records submitted in support of the fee petition, these are the only three that relate to Madison; the balance relate to services performed with reference to Chicago litigation. These affidavits barely rebut the defendant's showing of the hourly rates charged by attorneys of comparable experience in the Madison community. A trial court errs by ignoring uncontested affidavits concerning the prevailing rate within the community. Henry v. Webermeier, 738 F.2d 188, 193 (7th Cir.1984). A district court should not be held to have abused its discretion by finding persuasive that evidence which it may not disregard.

Certainly, Chicago-based law firms should not be undercompensated for services provided to Chicago clients in Chicago litigation through fee awards based on a Madison fee rate not prevailing (or perhaps even available) in the Chicago market. Conversely, under Chrapliwy, a district court does not abuse its discretion by holding, in litigation that is not complex, that compensation for legal services performed in Madison for a Madison client should not be based on an hourly rate far above that which prevails in Madison. To so hold is not to say that outstanding attorneys may recover only an average rate; the district court awarded the highest reasonable rates prevailing in Madison. Instead, such an approach merely calculates the fee "according to the prevailing market rates in the community", Blum v. Stenson, 465 U.S. at 895, 104 S.Ct. at 1547, after giving due regard to the attorneys' customary billing rates as required by Continental Illinois Securities, Pressley, and Barrow.

Balancing higher rates based on a different market against lower rates prevailing in the community in which the services were performed admittedly is difficult. Reasonable minds can differ; had I served as the trial judge in this case, I may well have awarded the rates requested by Mr. Gusman's attorneys. This court's review, however, is highly deferential. Smith v. Great American Restaurants, Inc., 969 F.2d 430, 439 (7th Cir.1992). We defer to the district court's discretion in light of its superior understanding of the facts surrounding the litigation, the lack of overriding need for uniformity in fee awards, and in the interest of minimizing fee litigation. Estate of Borst v. O'Brien, 979 F.2d 511, 514 (7th Cir.1992). As long as the district court provides a concise but clear explanation for its reasons, an award of less than the requested hourly rate may be reversed only for an abuse of discretion, Nanetti v. Univ. of Illinois at Chicago, 944 F.2d at 1418, which occurs only if the district court reaches a conclusion that no evidence in record supports as rational. Littlefield v. McGuffey, 954 F.2d 1337, 1350 (7th Cir.1992). I believe the district court provided a clear explanation for its reasons, and can find no abuse of discretion.

Accordingly, I dissent from the majority's remand order. I would affirm the district court's judgment in its entirety.