Damages Classes Under Rule 23(b)(3)

Damages Classes Under Rule 23(b)(3)

Rule 23(b)(3) permits a class action when “the court finds that the questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy.”¹ The court must make both findings on the proponent's proof, under the burden described at the start of this chapter. Predominance asks how much of the case common proof can resolve. Superiority asks whether a class suit resolves it better than the alternatives. The Rule closes with a list of matters pertinent to both findings; Amchem calls the list nonexhaustive and treats it as guiding a court's “close look” at the two criteria.²

The 1966 drafters built (b)(3) for cases that fit neither mandatory category. Amchem called it “the most adventuresome” innovation of the 1966 amendments: it authorized damages class actions that bind every member who does not elect exclusion, and it replaced the “spurious” class action, a permissive-joinder or opt-in device.³ The drafters framed it for situations in which “class-action treatment is not as clearly called for” as under (b)(1) and (b)(2), where a class suit “may nevertheless be convenient and desirable.”⁴

The word “convenient” carries the logic. The mandatory categories answer problems familiar from the joinder rules: competing demands on insufficient resources and remedies that cannot be divided among members. A (b)(3) member faces neither problem. Each member could sue alone and win an individual judgment, so only the efficiency of aggregate treatment holds the class together, most clearly for small claims that would not merit prosecution on their own.⁵ Small recoveries give no one an incentive to bring a solo action, and a class suit fixes that by aggregating “relatively paltry potential recoveries into something worth someone's (usually an attorney's) labor.”⁶

Efficiency is a thinner justification for binding absent people than indivisibility, so the Rule demands more in return. A (b)(3) class carries added protections: the predominance and superiority findings, individual notice, and the right to leave the class.⁷ The court must direct “the best notice practicable under the circumstances, including individual notice to all members who can be identified through reasonable effort,” and the expense of mailing does not excuse that notice.⁸ A member who opts out is not bound by the judgment and cannot later take advantage of it.⁹ Because these protections are costly and burdensome, class counsel try to fit an action within (b)(1) or (b)(2) and avoid them. Dukes limits the tactic: an action seeking both equitable relief and damages such as back pay cannot be certified solely under (b)(2) when the damages claims are not merely incidental.¹⁰

In practice, antitrust plaintiffs who seek damages usually have only (b)(3) available, and courts sometimes certify an antitrust action under both (b)(2) and (b)(3) when injunctive and monetary relief are both appropriate.¹¹ The category more readily takes an antitrust action by investors in securities than a mass tort, especially a products-liability claim, in part because of manageability.¹² Because class certification is most commonly sought under (b)(3), predominance may be the single biggest hurdle a proponent faces.¹³ That finding comes first.

Predominance

Rule 23(b)(3) allows a damages class only if “the questions of law or fact common to class members predominate over any questions affecting only individual members.”¹⁵ The finding tests whether the proposed class is “sufficiently cohesive to warrant adjudication by representation.”¹⁶ Cohesion matters because nothing else binds a (b)(3) class. No limited fund caps the members' recoveries and no indivisible injunction ties their fates; each member could sue alone, and the only reason to combine them is that one proceeding is more efficient than many.¹⁷ That efficiency exists only if trying the common questions once settles most of what matters. When exposure, causation, reliance, or damages must still be proved member by member, a classwide trial leaves the real work undone and bundles claims that differ in substance under one caption.

Only questions that exist before a settlement count. In Amchem Products, Inc. v. Windsor, the district court found predominance in a proposed settlement class of asbestos-exposed persons because the members shared the experience of exposure and an interest in prompt, fair compensation, and the settling parties added that the settlement's fairness was itself a common question.¹⁸ The Supreme Court rejected both arguments. Predominance “trains on the legal or factual questions that qualify each class member's case as a genuine controversy,” and those questions preexist any settlement.¹⁹ The fairness review that Rule 23(e) requires protects members from unjust settlements, but it does not supply the cohesion that makes representative action legitimate to begin with.²⁰ The mechanism is plain: every proposed settlement is a bargain its sponsors consider fair, so counting fairness as a common question would give every settlement class a common question and strip predominance of meaning in that setting.

Amchem also separates predominance from commonality. Shared exposure to asbestos might satisfy Rule 23(a)(2), the Court said, but “the predominance criterion is far more demanding.”²¹ Plaintiffs who show commonality must therefore make a further showing that common questions predominate.²² The two inquiries do different work. Commonality asks whether the class has at least one question whose answer moves every claim, and one is enough. Predominance sets the common questions against the individual ones and weighs them, so a class can share a genuine common question and still fail because the individual questions outweigh it.

Wal-Mart Stores, Inc. v. Dukes shows how closely the two inquiries sit. The partial dissent charged that the majority's emphasis on differences among class members “mimics” the predominance inquiry and, by asking whether differences “impede” common adjudication, duplicates the superiority question.²³ The majority answered that it weighed dissimilarities only to decide whether there is “[e]ven a single [common] question,” as Rule 23(a)(2) requires, and not, as (b)(3) requires, whether common questions predominate.²⁴ The opinions disagree about which inquiry a given fact belongs to, but both treat commonality and predominance as separate steps. Falcon raised a related concern, under typicality, about “across-the-board” classes, in which a woman claiming discrimination in hiring, for example, seeks to represent women who claim discrimination in promotion, termination, or pay.²⁵ On an exam, keep the steps apart: first ask whether any common question has a common answer, then ask whether that question, weighed against the individual ones, is large enough to justify a class trial.

Neither the Rule nor the Advisory Committee defines “predominate,” and courts weigh the two kinds of questions in several ways.²⁶ Some ask whether resolving the common questions at least signals the beginning of the end of the litigation. Others ask whether it materially advances the litigation as a whole. Still others ask how much individualized proof remains, so that a class fails when plaintiffs “must still introduce a great deal of individualized proof or argue a number of individualized legal points to establish most or all of the elements of their individual claims.”²⁷ Whatever the formulation, predominance is practical; it requires no complete identity of legal or factual claims.²⁸ A court may also count a question the defendant concedes, and at least one court has held that conceded classwide liability still counts when common questions are weighed against individual ones.²⁹

Predominance comes most easily where one course of conduct by the defendant generates every claim. Amchem observed that the test is “readily met in certain cases alleging consumer or securities fraud or violations of the antitrust laws.”³⁰ Those cases look upstream at what the defendant did, such as one fraud or one anticompetitive scheme, and not downstream at what happened to each member, so the central question has the same answer for everyone.³¹ Allegations about the existence, scope, and effect of an antitrust conspiracy have been held to satisfy commonality, and the individual damages that remain can be handled by bifurcating liability from damages, appointing a special master, or relying on the defendant's records.³² Antitrust fights over predominance therefore usually turn on whether plaintiffs can offer common proof of injury or only “highly individualized, fact-intensive inquir[ies]” into factors unique to each member.³³

Securities fraud shows predominance turning on a single element. Each investor ordinarily must prove reliance on the alleged misrepresentation, and individual proof of reliance would often be an insuperable barrier to certification.³⁴ The Supreme Court eased the problem first for omissions: Affiliated Ute upheld a presumption of reliance in Rule 10b-5 omissions cases, so plaintiffs whose allegations consist of omissions only can show predominance.³⁵ The presumption does not fit claims that rest on misrepresentations, entirely or in significant measure.³⁶ For those claims the fraud-on-the-market presumption does the work: plaintiffs who prove that their shares traded on an efficient market can presume that traders relied on the integrity of the market price.³⁷ Plaintiffs would have to make that showing on the merits at trial anyway, which is why securities fraud is perhaps the most common setting in which a court considers a merits question at the certification stage.³⁸

The two versions track a line the Advisory Committee drew. When members were exposed to differing representations, individual questions of reliance may predominate, and defendants routinely argue that reliance turns on what each plaintiff was told, what information each had, and whether each actually and reasonably relied.³⁹ A fraud carried out through similar misrepresentations to many people can still be an appealing class case even if damages must later be determined person by person, while material variation in the representations, or in the kinds or degrees of reliance, may make a class unsuitable.⁴⁰ Courts are particularly reluctant where the complaint rests on oral misrepresentations, whereas similar misstatements repeated across a series of written offering documents may be managed by dividing the class into subclasses.⁴¹

Individualized damages draw a common predominance objection, and the answer depends on how much of the case remains common. Tyson Foods, Inc. v. Bouaphakeo made clear that individualized damages do not automatically defeat predominance: “When one or more of the central issues in the action are common to the class and can be said to predominate, the action may be considered proper under Rule 23(b)(3) even though other important matters will have to be tried separately, such as damages.”⁴² Comcast Corp. v. Behrend marks the limit. There the Court overturned certification of an antitrust class because the plaintiffs fell “far short of establishing that damages are capable of measurement on a classwide basis,” and “[q]uestions of individual damage calculations will inevitably overwhelm questions common to the class.”⁴³ A majority of courts since Comcast have held that it does not foreclose certification of classes whose damages differ by member.⁴⁴

In re Hannaford Brothers Data Security Breach Litigation shows the failing side of that line. The district court found numerosity, commonality, typicality, and adequacy satisfied, with common questions about whether the defendant's conduct was negligent or a breach of contract and whether it caused the theft of customers' data. It nonetheless denied (b)(3) certification, because customers' economic harm varied with their different mitigation efforts and fees and because the plaintiffs offered no expert to testify to a lump-sum damages award.⁴⁵ The proponent carries the burden, so a plaintiff who can name a common wrong but offers no common way to prove the harm leaves the court nothing to weigh against the individual questions. A relevant consideration under (b)(3) is whether plaintiffs have proposed a viable plan to try the case, and an increasing number of courts require a “trial plan” that describes the issues likely to arise at trial and tests whether they are susceptible of classwide proof.⁴⁶

Mass torts are where predominance most often fails. The Advisory Committee for the 1966 revision said that a mass accident case is “ordinarily not appropriate” for class treatment, because significant questions of damages, liability, and defenses are likely to affect individuals in different ways.⁴⁷ Amchem declined to turn that caution into a ban: the text of the Rule does not categorically exclude mass tort cases, and district courts have certified them in increasing numbers since the late 1970s. But the Committee's warning still counsels caution “when individual stakes are high and disparities among class members great.”⁴⁸ The asbestos class failed on exactly those terms. Given the greater number of questions peculiar to the several categories of class members, and to individuals within each category, and the significance of those uncommon questions, an overarching dispute about the health consequences of asbestos exposure could not satisfy predominance.⁴⁹

The cloud case is the classic mass-tort personal-injury situation, with potentially hundreds of individualized issues: the extent and timing of each person's exposure, whether the exposure caused that person's particular injuries, and the amount of damages. Those issues typically predominate over the common ones, so certification under (b)(3) is inappropriate and a class action is not superior to individual suits.⁵⁰

Step three usually decides the case. A securities class that can invoke the fraud-on-the-market presumption and an antitrust class with a classwide measure of overcharge pass it. A data-breach class with no classwide damages evidence, or a toxic-exposure class whose members were exposed in different ways, does not.

Superiority

Rule 23(b)(3) pairs predominance with a second finding: the court must find that “a class action is superior to other available methods for fairly and efficiently adjudicating the controversy.”⁵³ “Superior” is a comparative word, so a court cannot make the finding by studying the proposed class alone. It must ask what would become of these claims if it refused certification, and it certifies only when a class resolves them better than the realistic alternatives. Predominance asks whether the claims can be tried together; superiority asks whether trying them together is the best way to proceed. A class that clears the first test can fail the second when another procedure does the job at lower cost to everyone.

A court should name and weigh the competing procedures, which include:

• individual actions by each member;
• joinder of the members as parties⁵⁴;
• consolidation of separate suits under Rule 42(a), or a series of statewide or more narrowly defined class actions⁵⁵;
• multidistrict proceedings for claims already filed⁵⁶; and
• another forum, such as a bankruptcy proceeding.⁵⁷

Each device is measured against the class on the two measures the Rule names, fairness and efficiency.

The comparison is required because a class action has a price. It departs from the conventional rule that each individual may control any litigation that threatens to impose liability or end a claim, and only the adequacy of the representation justifies the departure.⁵⁸ That price is worth paying only when the class supplies what the alternatives cannot: a way to resolve claims that would otherwise go unresolved or be resolved wastefully.

The paradigm is the negative-value suit, in which each member's claim is worth less than the cost of litigating it. Amchem described the policy at the core of the class action as overcoming the problem that small recoveries give no one an incentive to sue alone, and it said the class solves that problem by aggregating “the relatively paltry potential recoveries into something worth someone's (usually an attorney's) labor.”⁵⁹ In Eisen v. Carlisle & Jacquelin the named plaintiff's individual stake in the damages he sought was only $70, and the Court observed that no competent attorney would undertake the complex antitrust action to recover so inconsequential an amount: “Economic reality dictates that petitioner's suit proceed as a class action or not at all.”⁶⁰ That sentence states the superiority comparison in its simplest form. Individual actions, the only alternative, are empty because no rational member files. Without the class, the defendant keeps whatever it gained from thousands of small losses, and no court decides whether the conduct was lawful.

The same reasoning holds in securities litigation, where a class action is often the only remedy when each plaintiff's damage is too small to warrant the expense of suit.⁶¹ It holds even when the numbers are enormous: in In re Domestic Air Transportation Antitrust Litigation the district court certified a class of an estimated 12.5 million unidentified ticket purchasers, reasoning that the individual claims were so small that most would never be pursued without a class.⁶² Size alone therefore does not defeat superiority when the alternative is that claims go unpursued, although manageability, taken up in a later section, is where size can cut the other way.

The comparison reverses when members hold large claims. Amchem noted that the text of (b)(3) does not exclude cases in which individual damages run high, but that the Advisory Committee had dominantly in mind “the rights of groups of people who individually would be without effective strength to bring their opponents into court at all.”⁶³ A member with a large claim has that strength and can find a lawyer, so courts are generally more reluctant to certify classes of such members.⁶⁴

How much members value controlling their own suits is the first of the four matters the Rule lists, and the next section takes it up.

The Amchem litigation shows a class failing the comparison even though the claims were individually significant. The Third Circuit thought that “[a] series of statewide or more narrowly defined adjudications, either through consolidation under Rule 42(a) or as class actions under Rule 23, would seem preferable.”⁶⁵ The Supreme Court affirmed, concluding that the class could not satisfy predominance.⁶⁶ The Third Circuit's reasoning still shows how the comparison runs: the question is always superior to what. In Eisen nothing could take the class's place; in Amchem narrower adjudications could. The Rule's phrase “fairly and efficiently” also makes fairness part of the measure alongside efficiency.

Not every proposed substitute counts as an alternative. In In re Hannaford Brothers Data Security Breach Litigation, a proposed refund program would have reimbursed fees customers paid to replace their cards, and the district court held that the program was not an alternative adjudication method contemplated by the Federal Rules for deciding between a class action and individual lawsuits.⁶⁷ The structure of the comparison supports that result: every device on the list decides claims through a neutral decisionmaker, while a refund program lets the party accused of the wrong decide what to repay. Courts have divided on how far to go. One court treated the defendants' refund and replacement programs as superior to class litigation, and another declined to weigh a defendant's private settlement program at all.⁶⁸

The four matters the Rule lists do not exhaust what a court may consider, and no single one is determinative.⁶⁹ Courts have also asked whether an alternative regulatory mechanism exists and whether a class would produce significant judicial efficiencies. Some have refused to find superiority where members who suffered no actual economic injury sought large statutory damages and the only benefit of certification was the pressure to settle that a threat of ruinous exposure creates.⁷⁰ Superiority also overlaps with manageability, which is one reason the category more readily takes an antitrust action by investors than a mass products-liability action.⁷¹

Individual Control and Pending Suits

Rule 23(b)(3) lists four matters pertinent to its predominance and superiority findings, and the first two ask how much the class members want, or have already exercised, control over their own claims. Factor (A) directs the court to weigh “the class members' interests in individually controlling the prosecution or defense of separate actions,” and factor (B) directs it to weigh “the extent and nature of any litigation concerning the controversy already begun by or against class members.”⁷³ Because the Rule introduces the list as matters pertinent to the court's findings, both factors feed the superiority comparison; neither is a separate hurdle for the proponent.⁷⁴

These factors belong in the analysis because of what a (b)(3) class takes from its members. The ordinary rule lets each person control any litigation that threatens to impose liability or end a claim, and a class action departs from that rule only on the strength of adequate representation.⁷⁵ A (b)(3) member could have sued alone, so a member who stays gives up the choice of lawyer, the shape of the strategy, and the timing of settlement. The price is easy to justify when the member would never have used those powers and hard to justify when the member would have. Factor (A) estimates which situation the court faces by asking what members with these stakes would realistically want. Factor (B) checks the estimate against conduct, because a member who has already hired a lawyer and filed suit has shown the interest directly.

A bare preference for control cannot carry the factor, because in any conceivable case some members will wish to assert their rights and others will not. Stockholders' derivative suits are brought almost invariably by minority stockholders to challenge action that the majority approves, and few taxpayer suits against a public expenditure lack taxpayers who would have preferred the expenditure; both routinely proceed as class suits.⁷⁶ The factor therefore asks about a practical interest, one strong enough that members would actually litigate on their own.

Amchem quotes the observation that individuals' interests in conducting separate lawsuits “may be so strong as to call for denial of a class action,” yet may instead be “theoretic rather than practical”: the class may be cohesive enough that prosecution through representatives would be unobjectionable, or the amounts at stake for individuals may be so small that separate suits would be impracticable.⁷⁷ The interest therefore rises and falls with the size of each member's stake. At the large-stakes pole, the Court noted that the text of (b)(3) does not exclude cases in which individual damages run high, but that the Advisory Committee had dominantly in mind “the rights of groups of people who individually would be without effective strength to bring their opponents into court at all.”⁷⁸ A member with a large claim has that strength, so courts are generally more reluctant to certify classes of such members.⁷⁹

Members with tiny claims stand at the opposite pole.

That result follows the policy at the core of the class device: aggregate small recoveries into something worth a lawyer's labor.⁸⁰ Because the inquiry is case-specific, a court makes a practical, realistic assessment of whether members would prefer to litigate separately or collectively.⁸¹ An answer that announces that large stakes defeat certification, or that small stakes guarantee it, skips that assessment.

Factor (B) turns from what members would want to what they have already done. The Rule directs the court to consider litigation already begun by or against class members but does not say which way that litigation points.⁸² Courts have accordingly reached opposite conclusions. Some treat similar pending suits as showing the superiority of a class, because the cases can be heard together, while others treat them as showing that plaintiffs already have enough incentive to sue without one.⁸³ The absence of suits is read both ways as well. It can favor a class when individuals have too little at stake to sue, and it has counseled denial in mass torts, where the Fifth Circuit in Castano worried that a court cannot assess predominance and superiority without a track record of prior trials.⁸⁴

The opposite readings come from what pending suits can reveal. A suit filed by a member shows that the member has the means and motive to proceed without a class, which lines factor (B) up with factor (A). A cluster of suits over the same events can instead show duplication, with the same witnesses deposed repeatedly and the same legal question decided over again by different judges, which a single class proceeding would end. The Rule's word “nature” points the court to the details that decide which message the litigation carries: who filed, in what forum, at what stage, and whether the suits are themselves class actions. If the other pending suits are class actions, a new class is much less likely to be superior, although even a multi-jurisdictional class action need not deter certification of a smaller class, particularly when the larger class has not been certified.⁸⁵ The forum matters too. In Gregory v. Finova Capital Corp. the plaintiffs sought a securities class against a bankrupt defendant, and the Fourth Circuit held the bankruptcy forum superior because letting the class action go forward would be “inefficient and needlessly duplicative” when an adversary proceeding would likely resolve the controversy in the normal course of the bankruptcy.⁸⁶

An exam stem that mentions pending suits is asking which direction they point, and the answer depends on who filed them and what they show.

Pending suits can also support aggregation while members keep a voice. Members' greater ability to participate in class litigation traces in part to cheap communication: low-cost social media gives motivated members access to information about their cases, a way to monitor their attorneys, and even direct contact with class counsel.⁸⁷ It traces in part to multidistrict aggregation of claims already filed, as in the NFL concussion and Gulf oil spill litigations, which operates independently of the class action.⁸⁸ A class action in that setting gathers up cases that members are already running, so the loss of control that factor (A) measures shrinks accordingly.

Forum Choice and Manageability

Rule 23(b)(3) ends its list of pertinent matters with two that turn from what the members want to what the court can actually do: “the desirability or undesirability of concentrating the litigation of the claims in the particular forum,” and “the likely difficulties in managing a class action.”⁹¹ Like factors (A) and (B), these matters bear on the predominance and superiority findings, so they weigh on that comparison and are not stand-alone requirements.⁹² Each answers a practical question the earlier factors leave open. A class of members who would never sue alone may be just the kind of group the class device exists to serve, and it may still be a group that no court can sensibly try, or one that belongs in a different court.

Factor (C) asks whether this court is a sensible place to resolve every member's claim at once. Aggregation puts the whole controversy before one judge, so the question is how well that court fits the case: where the events happened, where the witnesses and documents are, and where the parties live. Courts sometimes reject certification when parties, witnesses, and evidence are scattered across the country. They also frequently find concentration desirable despite that dispersion, either because the key events occurred in the forum state or because some inconvenience to some parties and witnesses was unavoidable wherever the case was heard.⁹³ The second line of cases explains why dispersion alone rarely decides the factor. Every forum burdens someone, so the real comparison is between this court and the realistic alternatives, and for members with tiny claims the alternative is often no suit at all.

The forum factor seldom carries a certification decision by itself. An answer should invoke it when the facts say something about where the events and evidence are located, and it should treat the point as one weight among several.

Manageability draws far more attention. Eisen v. Carlisle & Jacquelin shows the concern. The court of appeals there noted that distributing any ultimate recovery to the class would be formidable, though not necessarily insuperable, and it was “reluctant to permit actions to proceed where they are not likely to benefit anyone but the lawyers who bring them.” It directed further inquiry into “the mechanics involved in the administration of the present action.”⁹⁴ The worry is that a large aggregate recovery means little to members if administration consumes much of it. Manageability thus asks whether the bet that one proceeding can do the work of many can be collected on, which depends on whether the members can be reached, whether their claims can be proved together, and whether any recovery can get to the people it belongs to.

Reaching the members comes first. Because Rule 23(c)(2) requires notice to (b)(3) members, a court must sometimes try to notify a large class dispersed across the country with little or no information to guide it, and difficulty in notifying any significant percentage of the class may counsel against certification.⁹⁵ Notice is what lets a member decide whether to stay in or opt out, so a class whose members cannot be found binds people who never had a chance to leave. The notice rules themselves come up below with the certification order.

Size raises the question without answering it. In In re Domestic Air Transportation Antitrust Litigation the district court certified a class of an estimated 12.5 million unidentified ticket purchasers, reasoning that the individual claims were so small that most would go unpursued without a class. In In re Hotel Telephone Charges the Ninth Circuit held that a putative class of 40 million hotel guests was beyond a court's capacity to manage or effectively control.⁹⁶ The two outcomes show that the numbers are only evidence of difficulty. A court must still ask whether the members can be reached and the case run, and it must set the answer against the prospect that small claims would otherwise never be heard.

Proof is the second problem. Courts increasingly ask the proponent for a trial plan that describes the issues likely to arise at trial and tests whether they can be proved classwide.⁹⁷ Individualized damages do not automatically defeat certification, as Tyson Foods, Inc. v. Bouaphakeo confirmed, yet difficulty in calculating damages can still count as a manageability concern, particularly alongside other concerns.⁹⁸ In In re Hannaford Brothers Data Security Breach Litigation the plaintiffs offered no expert to testify to a lump-sum damages award, and the district court treated that gap, along with variation in members' mitigation efforts and fees, as a failure to carry their burden at certification.⁹⁹ A proponent who can name a common wrong but offers no workable way to prove the harm on a classwide basis leaves the court with a trial it cannot picture.

The hardest cases arise when someone tries to cure unmanageability with a shortcut, and Wal-Mart Stores, Inc. v. Dukes marks the limit. The Court held that the Ninth Circuit erred in trying to replace individualized backpay proceedings with “Trial by Formula,” which would have determined claims for a sample of class members and applied the result to the rest.¹⁰⁰ Once a plaintiff establishes a pattern or practice of discrimination, the employer may raise individual affirmative defenses and show that its action was lawful.¹⁰¹ An average applied across the class erases that second stage. Some members who receive the average would have lost on a defense the employer never got to present, and the employer pays them anyway. Because Rule 23 cannot be read to “abridge, enlarge or modify any substantive right” under 28 U.S.C. § 2072(b), the Court concluded that a class cannot be certified on the premise that the defendant will not be entitled to litigate its statutory defenses to individual claims.¹⁰² Manageability therefore measures whether the case can be tried the way the substantive law requires, and a plan that achieves feasibility by removing a defense has changed the claim. Tools that leave the defenses intact stay available but carry limits: special masters and separate juries for damages are generally used only in the settlement context, a special master appointed over a party's objection raises Seventh Amendment jury-trial concerns, and the use of multiple juries has been questioned on the same ground.¹⁰³

A class proposed only for settlement changes the manageability inquiry, but only part of it. A court asked to certify a settlement class need not ask whether the case would present intractable management problems if tried, since the proposal is that there be no trial.¹⁰⁴ The other requirements that protect absent members by blocking unwarranted or overbroad class definitions still apply and demand undiluted, even heightened, attention.¹⁰⁵ The trial-manageability point therefore drops out of a settlement-class analysis, while adequacy, predominance, and the rest of the Rule 23 safeguards stay in full force.