It is a fundamental principle of American law that every person is entitled to his or her day in court. Multiple victims of air disasters, multiple stockholders of companies that have committed securities violations, and multiple holders of rights in pensions, normally may all bring their own suits even if the defendant engaged in a single course of action that affected everyone similarly. One formal exception to this principle is the class action, recognized in federal court under Fed.R.Civ.P. 23. The vigor with which the defense bar has often opposed class certifications might cause one to think that defendants prefer to take their cases one at a time, but that would be too simplistic a view. In fact, the existence and incidence of another exception to the general rule, the doctrine of virtual representation, suggests that defendants sometimes like the benefits of a group result--because it is usually defendants who argue that a new group of plaintiffs is barred from bringing an action since the plaintiff in an earlier suit was its "virtual representative."

This case requires us to explore what courts have actually meant when they have referred to virtual representation, and what if any independent significance that concept has. The district court dismissed the claims of American Airlines pilot Robert Tice and his eleven colleagues on the ground that they were bound by the result reached in a different lawsuit brought by different pilots, in which essentially the same American Airlines policy was challenged under the Age Discrimination in Employment Act (ADEA), 29 U.S.C. § 623(a)(1), (2). The Tice plaintiffs argue that this was error, because they were represented neither formally nor "virtually" in the earlier case, Johnson v. American Airlines, Inc., 745 F.2d 988 (5th Cir.1984), and it would thus violate due process to preclude them from bringing their own suit. American has cross-appealed, claiming that the district court abused its discretion in refusing to transfer this case to the Northern District of Texas pursuant to 28 U.S.C. § 1404(a).

* At the center of this case is the way that American Airlines has responded to the Federal Aviation Administration (FAA) rule providing that no person may serve as a pilot if that person has reached the age of 60. See 14 C.F.R. § 121.383(c); Baker v. FAA, 917 F.2d 318 (7th Cir.1990). This rule covers only the positions of pilot and copilot, and it thus does not prevent former pilots from serving as flight officers after their 60th birthday. The flight officer, third in command of commercial passenger aircraft, monitors the aircraft's fuel, electrical, hydraulic and other systems before, during and after the flight, but does not pilot the aircraft. The reason why American does not permit people who they determine are too old to be pilots to bid for or hold the flight officer position is because it has a rigid "up-or-out" policy under which all flight officers must be eligible to advance to the position of pilot. See Murnane v. American Airlines, Inc., 667 F.2d 98, 99 (D.C.Cir.1981). That rule permits American to use the flight officer slot as a training ground for future pilots--a practice that would obviously be impossible if individuals deemed ineligible for the pilot's job were to occupy those positions.

American has successfully defended its policy twice before. The first time was in the Johnson litigation mentioned above, a case brought by 22 pilots as a group action authorized by the ADEA, see 29 U.S.C. § 626(b) (incorporating 29 U.S.C. § 216(b) of the Fair Labor Standards Act), in which the Fifth Circuit upheld a jury verdict that American's up-or-out policy qualified as a bona fide occupational qualification (BFOQ) for purposes of the ADEA. See also Murnane, 667 F.2d at 98 (upholding a related aspect of the then-extant version of American's up-or-out policy under which it would not hire anyone over the age of 30 to be a flight officer). The second time was in an action brought by the Equal Employment Opportunity Commission (EEOC) under the ADEA on behalf of a class of named pilots age 40 and over who applied for and were denied employment with American as flight officers because, under the current version of the up-or-out policy, they would not have enough time to progress to the rank of Captain and work there long enough to satisfy American. See EEOC v. American Airlines, Inc., 48 F.3d 164 (5th Cir.1995) ( EEOC).

The Tice plaintiffs, however, did not participate in those earlier cases, and the claims they present are somewhat different: because they were already American pilots, they are not complaining about a failure to be hired. Instead, they were forced to retire when they turned 60, because the up-or-out policy prevented them from "downbidding," which is the mechanism by which they might transfer back to the flight officer level. (In the absence of American's policy, downbidding would probably have been easy for them, because the flight officer position, while not affected by the FAA's rule that commercial pilots and copilots be under 60, is on the same union seniority list as that maintained for pilots.) The Tice plaintiffs point out that American permits or requires pilots below the age of 60 to downbid to flight officer in a variety of situations, and, according to the Tice plaintiffs, all other major U.S. airlines now permit their pilots who turn 60 to downbid in this manner. The "downbidding" claim never arose in the Murnane or EEOC litigation, and the Tice plaintiffs stress that they were American employees, either working as pilots or in line to become pilots, at the time those cases were pending, and thus they were unaffected by the initial hiring rules. Even though the claims of the Tice plaintiffs are quite similar to those of the Johnson plaintiffs, the Tice plaintiffs stress that they had not yet been subjected to the ban on downbidding at the time of that case. In fact, they were actually excluded from the class eligible to join that litigation as plaintiffs because they were too young at the relevant time. Furthermore, their interests at the time diverged from those of the Johnson plaintiffs (forcing out older workers can be thought to help younger workers, on at least a surface level), which suggests that the Johnson plaintiffs could not have adequately represented them.

Shortly after the plaintiffs filed their complaint, American moved for a transfer of venue to the U.S. District Court for the Northern District of Texas under 28 U.S.C. § 1404(a), arguing that most witnesses and evidence were located there, including a flight simulator at American's facility at the Dallas-Fort Worth Airport (DFW) that could not be transported out of Texas, that the principal situs of American's operations is in the Northern District of Texas, and that more of the named plaintiffs resided in Texas than in Illinois. American pointed out as well that the Johnson case, which eventually wound up in Texas, had also begun in Illinois and had been transferred for many of the same reasons American was urging here. The Tice plaintiffs responded that almost the same number of plaintiffs were based at O'Hare Airport, in Chicago, as at DFW, that most documents had already been produced, and that American engaged in enough business in Illinois through its extensive operations at O'Hare, American's largest international airport, to warrant respecting their choice of forum.

The court denied American's motion, stating its reasons in open court. It found that from the standpoint of convenience to parties and witnesses the advantages of the two districts were relatively evenly balanced. The court also expressed skepticism that American's flight simulator at DFW was truly unique, or (more importantly) that such a machine would be useful in any event at a jury trial. When American renewed the motion after another similar case was filed in the Northern District of Texas, Roddie v. American Airlines, Inc., 4:96-CV-548-A (N.D.Tex. Sept. 9, 1996), the court denied it again.

American then moved for judgment on the pleadings, under Fed.R.Civ.P. 12(c), on the ground that the doctrine of claim preclusion applied here. The court (Judge Ann C. Williams taking over for Judge Brian Barnett Duff) granted the motion. The district court found that under federal principles of preclusion American had to show three things: (1) a final judgment on the merits in an earlier action, (2) an identity of the cause of action in both the earlier and the later suit, and (3) an identity of parties or their privies in the two suits. The first element was undisputed, and the court found that for the second element the question in both cases was whether American's policy violated the ADEA. That left the third element, which is central to this appeal. The court recognized that the Tice plaintiffs were not formal parties to the Johnson action and that "privity" could exist only if the doctrine of virtual representation applied. (It also noted that neither Murnane nor EEOC precluded the Tice plaintiffs' claim, because of the differences between the class of people seeking to be hired and the class of people seeking to avoid compulsory retirement.) After noting that no express analytical framework was available to delineate clearly when virtual representation should or should not be applied, the district court, guided mainly by earlier decisions from the Northern District of Illinois, concluded that the Tice plaintiffs had been "virtually represented" by the Johnson plaintiffs.

We agree with the district court on one basic point: the doctrine of virtual representation is amorphous. Indeed, in our view the term itself illustrates the harm that can be done when a catchy phrase is coined to describe a perfectly sensible result. The phrase takes on a life of its own, and before too long, it starts being applied to situations far removed from its intended and proper context. In the case of "virtual representation," the concept had its origin in the field of probate proceedings, in which "it is often necessary to establish a procedure that will bind persons unknown, unascertained, or not yet born." 18 Charles Alan Wright et al., Federal Practice and Procedure § 4457 at 494 (1981) (hereinafter Wright). In that narrow setting, courts would find an identity of interests between the representatives who participated in the litigation and other individuals whose interests were clearly aligned with those of the actual litigants. In the argot of res judicata law, the technical nonparties were treated as parties to the first suit for purposes of assessing its preclusive force.

Branching out from those roots, the term "virtual representation" began to be referred to as a doctrine. As such, its scope broadened in a line of cases from the Fifth Circuit, most notably Aerojet-General Corp. v. Askew, 511 F.2d 710 (5th Cir.1975), in which the court explained the theory as follows:

Under the federal law of res judicata, a person may be bound by a judgment even though not a party if one of the parties to the suit is so closely aligned with his interests as to be his virtual representative.

Because, under the Aerojet conception, virtual representation offers one way to bind a nonparty to the results of an earlier case, it is apparent that this concept, or doctrine, stands at the intersection of two important principles of civil litigation: the right, grounded both in due process and in the property right represented by the lawsuit chose in action, of each individual to assert her own claim; and the need of litigants and the judicial system alike for finality of decisions after a full and fair airing of a matter.

The difficult question here is to decide what an idea of "virtual" representation legitimately can add to the law of preclusion that is not already captured by a flexible inquiry into what used to be called "privity." As this court noted in In Matter of L & S Industries, Inc., 989 F.2d 929 (7th Cir.1993), "privity" is now seen as "a descriptive term for designating those with a sufficiently close identity of interests." Id. at 932. Quoting with approval from Wright § 4449, at 418-19, the court acknowledged that "the privity label simply expresses a conclusion that preclusion is proper." Id. at 933. In a similar vein, the Second Circuit cautioned against an overly formalistic approach to privity in Chase Manhattan Bank, N.A. v. Celotex Corp., 56 F.3d 343 (2d Cir.1995), observing that "[w]hether there is privity between a party against whom claim preclusion is asserted and a party to prior litigation is a functional inquiry in which the formalities of legal relationships provide clues but not solutions." Id. at 346. See also In re Schimmels, 127 F.3d 875, 881 (9th Cir.1997). Indeed, in Schimmels the court at one point implied that "virtual representation" just describes a form of privity. Id., citing Aerojet.

We think the term "virtual representation" has cast more shadows than light on the problem to be decided. As a matter of fact, a finding that nonparties were virtually represented in earlier litigation has rarely been used actually to bar litigation. As far as we can tell, no published opinion by this court has done so, although one unpublished order has, see Henderson v. Stone, 930 F.2d 25 (table), 1991 WL 54855 (7th Cir.1991), and the doctrine was of indirect relevance in another order. Goodluck v. City of Chicago, 70 F.3d 1274 (table), 1995 WL 687637 (7th Cir.1995). The Wright treatise observes that "[a]ll of the cases that in fact preclude relitigation by a nonparty have involved several factors in addition to apparently adequate litigation by a party holding parallel interests." Wright, § 4457 (1998 Supp.) at 420. Examples of these additional factors include control or participation in the earlier litigation, acquiescence, deliberate maneuvering to avoid the effects of the first case, or the close relationship between the parties to the various cases. E.g., Tyus v. Schoemehl, 93 F.3d 449, 454-56 (8th Cir.1996) (adding that the doctrine is more appropriate in public law cases); NAACP v. Metropolitan Council, 125 F.3d 1171, 1175 (8th Cir.1997), vacated, --- U.S. ----, 118 S. Ct. 1162, 140 L. Ed. 2d 173, reinstated, 144 F.3d 1168 (8th Cir.1998); Bittinger v. Tecumseh Prod. Co., 123 F.3d 877, 881 (6th Cir.1997); Collins v. E.I. DuPont de Nemours & Co., 34 F.3d 172, 175-78 (3d Cir.1994); see also Wright, et al., 18 Federal Practice and Procedure: Jurisdiction § 4457, at 502.

These factors are all merely heuristics, however, shortcuts that courts use to determine the answer to the real (fact-specific) question--whether there was (or should be implied at law) the kind of link between the earlier and later plaintiffs that justifies binding the second group to the result reached against the first. See also McNealy v. Caterpillar, Inc., 139 F.3d 1113, 1116-17 (7th Cir.1998) (preclusion improper where the issues litigated in the two actions are different). This is, of course, the same question we and other courts have already identified as the crux of the privity inquiry. A proper functional analysis of privity, focusing on the general question whether the earlier parties were in some sense proper agents for the later parties, would therefore support preclusion in the cases that have used the lingo of virtual representation. Conversely, if a relationship between a nonparty and an earlier litigant does not satisfy this analysis, serious due process problems would arise if the earlier nonparty were barred from her own day in court.

Our conclusion is fortified by the way the Supreme Court handled an analogous problem in Richards v. Jefferson County, 517 U.S. 793, 116 S. Ct. 1761, 135 L. Ed. 2d 76 (1996). Like our case, Richards involved two rounds of litigation directed at the same policy--there, the validity of an occupation tax imposed by Jefferson County, Alabama. The first case had been brought in Alabama state court by the acting director of finance for the city of Birmingham and the city itself, and later had been consolidated for trial with a separate suit brought by three county taxpayers. Eventually, the case ended up in the Supreme Court of Alabama, which upheld the tax. The second case was brought (again in state court because of the Tax Injunction Act, 28 U.S.C. § 1341) by a class of all nonfederal employees subject to the county's tax, who claimed that the tax violated the U.S. Constitution's due process and equal protection clauses, as well as similar Alabama constitutional guarantees. The Supreme Court of Alabama agreed with the lower courts that the second suit was barred by the adjudication in the first, because there was a "substantial identity of parties" and the same cause of action was presented. The U.S. Supreme Court reversed, finding that the state supreme court's holding violated the federal due process rights of the second group of plaintiffs.

The Court began by reiterating principles that are of equal relevance here:

[I]n Anglo-American jurisprudence ... one is not bound by a judgment in personam in a litigation in which he is not designated as a party or to which he has not been made a party by service of process. This rule is part of our deep-rooted historic tradition that everyone should have his own day in court. As a consequence, a judgment or decree among parties to a lawsuit resolves issues as among them, but it does not conclude the rights of strangers to those proceedings.

517 U.S. at 798, 116 S. Ct. 1761 (internal quotations and citations omitted). The Court also acknowledged that formal party status is not always necessary to bind a party, and that the term "privity" now extends beyond traditional formal relationships such as guardian and ward or trustee and beneficiary. Id. Nevertheless, the Court found, the Richards plaintiffs could not be barred by the earlier Alabama litigation. It squarely rejected the Alabama Supreme Court's conclusion that res judicata could apply because the later plaintiffs were adequately represented in the earlier suit, not because it disagreed with the Alabama court's assessment of the nature of the earlier representation, but because this fact alone was not enough to bar the second suit. The parties in the earlier case did not provide the later plaintiffs with any notice that a suit was pending that would conclusively resolve their legal rights. The earlier suit was not on behalf of a class that included the later plaintiffs, and the fact that the city participated in the earlier case did not expand its preclusive scope either. Id. at 801-02, 116 S. Ct. 1761.

Richards therefore stands as an example both of the functional approach that is required for privity analysis and of the importance the Court attaches to assuring each person his or her own day in court. From this functional standpoint, a conclusion that the interests of the nonparties were sufficiently aligned with the earlier litigants would be impossible unless, among other things, the earlier litigants would have been adequate representatives of the later litigant(s). See id. at 800-01, 116 S. Ct. 1761; Wright, et al., 18 Federal Practice and Procedure: Jurisdiction § 4457, at 495, 502 (1981 & Supp.1998); id. at 421-22 (Supp.1998). As we noted in Ahng v. Allsteel, Inc., 96 F.3d 1033, 1037 (7th Cir.1996), a case in which we also refused to preclude later litigation because of virtual representation, "the doctrine of 'virtual representation' recognizes, in effect, a common-law kind of class action. It applies only when there is a practical identity of interests between the former litigant and the present one." (As we explain below, the fact that virtual representation looks like a class action but avoids compliance with Rule 23 is a weakness, not a strength, of the doctrine.) In properly certified class actions, parties who have adequate notice and (at a minimum) an opportunity to opt out of an earlier case may, if their interests are sufficiently aligned with one of the earlier parties, be bound by the results of that litigation. See, e.g., Hansberry v. Lee, 311 U.S. 32, 61 S. Ct. 115, 85 L. Ed. 22 (1940); and Phillips Petroleum Co. v. Shutts, 472 U.S. 797, 105 S. Ct. 2965, 86 L. Ed. 2d 628 (1985). This is the theory on which Rule 23 is based. The class action cases allowing preclusion after adequate notice and the opportunity to opt out recognize a form of consent that is enough to justify binding the later parties to the earlier result (again, when the other criteria such as identity of issue and interest are also satisfied).

Unless there is a properly certified class action, handled with the procedural safeguards both state and federal rules afford, normal privity analysis must govern whether nonparties to an earlier case can be bound to the result. There would be little point in having Rule 23 if courts could ignore its careful structure and create de facto class actions at will; indeed, the Supreme Court's recent decision in Amchem Products, Inc. v. Windsor, 521 U.S. 591, 117 S. Ct. 2231, 138 L. Ed. 2d 689 (1997), emphasizing the importance of following Rule 23's procedures for settlement classes, suggests that such an approach would be unauthorized. Instead, the appropriateness of preclusion will depend on how closely the two sets of interests coincide and the role the absentees played in the earlier litigation. At a minimum, the issue on which preclusion is sought must be common to both cases, and the claims or defenses of the two allegedly equivalent parties (earlier litigant, present litigant) must be the same. See McNealy, 139 F.3d at 1116-17. In addition, unless a formal kind of successor interest is involved (e.g., subsequent landowner, successor corporation), there should be some indication not only that the second party was aware that the first litigation was going on and that the earlier litigation would resolve its claims, cf. Shutts, 472 U.S. at 811-12, 105 S. Ct. 2965 (emphasizing the need for absentee parties to have the right at least to notice and an opportunity to opt out of litigation), but also that the second party either had participated or had a legal duty to participate. Finally, of course, the due process rights of absentees that the decisions in Hansberry, Shutts and Richards recognized must be respected.

Turning to the case before us, we conclude that the district court should have allowed the Tice plaintiffs to proceed with their suit. As the EEOC pointed out in its amicus brief in this court, the statute establishing the rights they are seeking to vindicate makes it quite clear that these rights are individual, not group-based The ADEA provides that "[n]o employee shall be a party plaintiff to any ... action [under it] unless he gives his consent in writing...." 29 U.S.C. § 216(b) (incorporated into the ADEA by 29 U.S.C. § 626(b)). American does not claim that the Tice plaintiffs ever gave their consent in writing to any of the earlier cases upon which it relies. Second, as we noted in Woodall v. Drake Hotel, Inc., 913 F.2d 447, 450 (7th Cir.1990), ADEA group actions are "opt-in" rather than "opt-out," as a result of the "consent in writing" requirement, and thus differ importantly from the class actions available under Rule 23. The additional protections of this system compared to Rule 23 "opt-out" class actions would be eviscerated were the Tice plaintiffs to be viewed as having been in privity with the earlier plaintiffs.

Beyond the specifics of the ADEA, there are too many differences between the Tice plaintiffs and the plaintiffs in the earlier cases to make preclusion appropriate. The same flaw exists here as we found in Ahng, namely, the dissimilarity of the two groups of plaintiffs. The EEOC and Murnane plaintiffs could not have represented the interests of the Tice group, because the class of persons complaining that they were not hired because of American's age policies is inherently different from the class of existing employees. What helps the one may directly hurt the other. Cf. Rutherford v. City of Cleveland, 137 F.3d 905, 910 & n. 5 (6th Cir.1998); General Tel. Co. of the Northwest v. EEOC, 446 U.S. 318, 331, 100 S. Ct. 1698, 64 L. Ed. 2d 319 (1980); Hill v. Western Elec. Co., 596 F.2d 99, 102 (4th Cir.1979). Furthermore, at the time of the Johnson case, the interests of American pilots over 60 (the plaintiffs there) diverged from the interests of American employees under the age of 60 (which the Tice plaintiffs were during the course of that litigation), since the younger pilots might well have wanted their elders forced out, freeing up more positions for themselves.

Finally, the factors that have persuaded earlier courts to find preclusion, either under the rubric of virtual representation or otherwise, do not point in that direction here. There is no evidence of manipulative litigation practices, and we reject American's suggestion that the Tice plaintiffs' decision to retain the same counsel as the Johnson plaintiffs amounts to an improper practice. Collins, 34 F.3d at 178; Benson and Ford, Inc. v. Wanda Petroleum Co., 833 F.2d 1172, 1174 (5th Cir.1987) ; but see Tyus, 93 F.3d at 456-57. The Tice plaintiffs did not and could not have taken part in any of the earlier challenges to American's policy, nor did they acquiesce to being represented indirectly by those plaintiffs. Aside from the fact that both groups of pilots were employed by American, there is no evidence of any relationship between the Tice and Johnson plaintiffs. Cf. Bittinger, 123 F.3d at 881-82; Collins, 34 F.3d at 176-77. If American's position were taken to its logical conclusion, the fact that the plaintiff in Plessy v. Ferguson, 163 U.S. 537, 16 S. Ct. 1138, 41 L. Ed. 256 (1896), lost would have been enough to preclude the plaintiffs half a century later in Brown v. Board of Education, 347 U.S. 483, 74 S. Ct. 686, 98 L. Ed. 873 (1954). In light of the Supreme Court's continuing solicitude for each person's right to pursue her own case, we cannot strain preclusion principles this much.

For similar reasons we do not agree with American that the EEOC served as the Tice plaintiffs' representative in EEOC, supra. The EEOC brought that case to enforce the rights of 57 specifically named parties, none of whom is involved in the Tice litigation. It was therefore plainly not acting as the actual representative of the Tice plaintiffs. See 29 U.S.C. § 626(c)(1); cf. EEOC v. U.S. Steel Corp., 921 F.2d 489, 494-95 (3d Cir.1990). The argument that the EEOC was a virtual representative rests on the grounds we rejected above, as well as on its status as a public body. Its status as a public body does not change our earlier analysis--Richards did not accept a similar argument with respect to the city of Birmingham, 517 U.S. at 802 n. 6, 116 S. Ct. 1761, and we think the same result follows here. American's final argument, that issue preclusion should apply here even if claim preclusion does not, fails for all the reasons the earlier argument failed, because "[c]ollateral estoppel ... holds only between the same parties." United States v. Brocksmith, 991 F.2d 1363, 1367 (7th Cir.1993) (citations omitted); Baker v. General Motors Corp., 522 U.S. 222, 118 S. Ct. 657, 666 n. 11, 139 L. Ed. 2d 580 (1998).

Because we are reversing the district court's decision that the Tice plaintiffs' suit is precluded by the Johnson litigation (or other earlier suits), we must also reach American's cross-appeal challenging the district court's denial of its motion to change venue. American's problem is, in a phrase, the standard of review. "We give great deference to a district court's rulings on motions to transfer venue. Indeed, this court can only reverse a district court's determinations in this regard if we find a 'clear abuse of discretion.' " In re Chicago, Milwaukee, St. Paul & Pacific R.R. Co., 974 F.2d 775, 789 (7th Cir.1992), quoting Cote v. Wadel, 796 F.2d 981, 985 (7th Cir.1986). Even if we agreed with American that the Northern District of Texas would be a very convenient forum, that is not enough to support a finding that the district court abused its discretion in concluding that the Northern District of Illinois is also a convenient forum. In opposition to the factors to which American pointed, the district court gave some weight (as it was entitled to do) to the plaintiffs' choice of forum, to the relative ease with which American (an airline company) could transport its employees between Texas and Illinois, to the questionable relevance and need at trial for the Dallas flight simulator, and to the fact that essentially the same number of plaintiffs were employed at O'Hare as at DFW. Under the circumstances, we do not find reversible error in the district court's decision.

We therefore R EVERSE the district court's judgment for American and R EMAND for further proceedings consistent with this opinion. We A FFIRM the district court's ruling denying American's motion to transfer the case to the Northern District of Texas.