Harvard Law Review Harvard Law Review Harvard Law Review

Class Actions

Microsoft Corp. v. Baker

Class certification decisions tend to be make-or-break for plaintiffs and defendants alike. It’s no surprise, then, that both have devised litigation tactics to help them secure the decisions they need.1×1. See, e.g., Campbell-Ewald Co. v. Gomez, 136 S. Ct. 663, 672 (2016) (rejecting a defendant-side tactic and comparing it to a similar “gambit” the Court rejected in an earlier case). Last Term, in Microsoft Corp. v. Baker,2×2. 137 S. Ct. 1702 (2017). the Supreme Court rejected a tactic that let plaintiffs engineer the right to appeal adverse class certification orders by volunteering to have their individual claims dismissed with prejudice.3×3. Id. at 1707. The Court’s desire to prevent plaintiffs from undermining the policies embodied in 28 U.S.C. § 1291 and Rule 23(f) of the Federal Rules of Civil Procedure was reasonable enough. Even so, the Court’s reasoning was puzzling — and all the more so because the Court might have reached the same result on neater doctrinal grounds. The Court’s opinion makes more sense, perhaps, if the Court was also attempting to avoid the Article III issue lurking in the case, as well as unsettled questions about when Article III issues may safely be avoided. If that diagnosis is right, however, Microsoft suggests that the Court might find itself unable to postpone questions about class action standing in future Terms.

Microsoft released the Xbox 360 in 2005.4×4. In re Microsoft Xbox 360 Scratched Disc Litig., No. C07–1121, 2009 WL 10219350, at *1 (W.D. Wash. Oct. 5, 2009). Soon after, some Xbox 360 owners discovered that their machines left scratches on the game discs inside.5×5. Id. In 2007, a group of owners filed a putative class action lawsuit against Microsoft in federal district court, alleging the existence of a design defect.6×6. Id. at *2. In that case, the district court denied class certification on the ground that individual questions of consumer protection law, causation, and damages predominated over common questions of law and fact.7×7. See id. at *5–7. The Ninth Circuit declined to review the district court’s order, and the plaintiffs settled their claims.8×8. Microsoft, 137 S. Ct. at 1710 (citing Baker v. Microsoft Corp., 851 F. Supp. 2d 1274, 1276 (W.D. Wash. 2012)). In 2011, Seth Baker and several other Xbox 360 owners (Baker) filed another putative class action against Microsoft in the same district court, alleging the same design defect.9×9. Id. This time, the district court granted Microsoft’s motion to strike Baker’s class allegations10×10. Id. at 1710–11 (citing Baker, 851 F. Supp. 2d at 1280–81).  — the equivalent of denying class certification11×11. Id. at 1711 n.7 (“An order striking class allegations is ‘functional[ly] equivalent’ to an order denying class certification and therefore appealable under Rule 23(f).” (quoting Scott v. Family Dollar Stores, Inc., 733 F.3d 105, 110 n.2 (4th Cir. 2013))).  — on the ground that comity required it to defer to its own previous denial of class certification.12×12. Id. at 1710–11. To reach that conclusion, the district court rejected Baker’s argument that an intervening Ninth Circuit decision constituted a change in law that made deference inappropriate.13×13. Id. Baker petitioned the Ninth Circuit for permission to appeal under Rule 23(f), but the court denied his petition.14×14. Id. at 1711.

At that point, Baker did something artful. Rather than litigate his individual claims to final judgment or pursue other familiar options,15×15. The Court noted that Baker might also have settled his claims (like the plaintiffs in the first suit) or sought interlocutory appeal under 28 U.S.C. § 1292(b) (2012). Id. Baker volunteered to have his individual claims dismissed with prejudice to create an appealable final decision.16×16. See id. Baker and Microsoft filed a joint stipulation and proposed order dismissing Baker’s claims with prejudice.17×17. Stipulation and Order Granting Plaintiffs’ Motion to Dismiss with Prejudice, Baker v. Microsoft Corp., 851 F. Supp. 2d 1274 (W.D. Wash. 2012) (No. 11-cv-00722). The parties disagreed, however, about the effect the dismissal would have if granted — in particular, about whether Baker would have the right to appeal the district court’s order striking his class allegations.18×18. Id. at 2. The district court granted the joint motion and ordered Baker’s claims dismissed with prejudice.19×19. See Microsoft, 137 S. Ct. at 1711. Baker appealed and challenged the district court’s order to strike, but not its dismissal order.20×20. Id.

The Ninth Circuit reversed.21×21. Baker v. Microsoft Corp., 797 F.3d 607, 609 (9th Cir. 2015). Writing for the panel, Judge Rawlinson22×22. Judge Rawlinson was joined by Judge Hawkins. Judge Bea concurred in the result, but he wrote separately to argue that the court should have provided lower courts with more guidance on how and when comity requires deference. See id. at 616–23 (Bea, J., concurring in the result). held that the court had jurisdiction under 28 U.S.C. § 1291,23×23. Id. at 612. which gives circuit courts jurisdiction over “all final decisions of the district courts.”24×24. 28 U.S.C. § 1291 (2012). Because the stipulated-to dismissal was with prejudice and not part of a settlement agreement, Ninth Circuit precedent dictated that it was “a sufficiently adverse — and thus appealable — final deci-sion” under § 1291.25×25. Baker, 797 F.3d at 612 (quoting Berger v. Home Depot USA, Inc., 741 F.3d 1061, 1065 (9th Cir. 2014)). Judge Rawlinson then reviewed the district court’s order to strike Baker’s class allegations and held that, in light of developments in Ninth Circuit law since the first Xbox lawsuit, the district court abused its discretion by granting Microsoft’s motion to strike.26×26. See id. at 612–16. The court took no position on whether class certification was appropriate on remand; it maintained only that the district court should have taken a fresh look at the class allegations.27×27. See id. at 615.

The Supreme Court reversed and remanded.28×28. Microsoft, 137 S. Ct. at 1715. Writing for the Court, Justice Ginsburg29×29. Justice Ginsburg was joined by Justices Kennedy, Breyer, Sotomayor, and Kagan. Justice Gorsuch took no part in the consideration or decision of the case. held that the Ninth Circuit lacked jurisdiction under § 1291.30×30. Microsoft, 137 S. Ct. at 1706–07. The Court declined to decide whether the Ninth Circuit had jurisdiction under Article III. See id. at 1712. To reach that conclusion, the Court relied on two main premises. First, the Court explained that § 1291’s “final decision” requirement “is to be given a practical rather than a technical construction”31×31. Id. at 1712 (quoting Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 171 (1974)).  — a construction geared toward “achieving a healthy legal system.”32×32. Id. at 1715 (quoting Cobbledick v. United States, 309 U.S. 323, 326 (1940)). Second, the Court explained that there were “[t]wo guides” — Coopers & Lybrand v. Livesay33×33. 437 U.S. 463 (1978). and Rule 23(f) — that should inform and control the Court’s practical construction of § 1291 in Microsoft.34×34. Microsoft, 137 S. Ct. at 1707.

The Court started with its first guide to § 1291: Coopers & Lybrand. Before Coopers & Lybrand, several circuit courts had developed a “death-knell” doctrine, which gave them § 1291 jurisdiction over orders denying class certification when, but only when, those orders left plaintiffs without adequate economic incentive to pursue their individual claims.35×35. See id. In Coopers & Lybrand, the Supreme Court rejected the death-knell doctrine, holding that orders denying class certification are not final under § 1291, even if they effectively end the litigation.36×36. See id. at 1707–08 (reiterating that class certification orders are “inherently interlocutory,” id. at 1707 (quoting Coopers & Lybrand, 437 U.S. at 477)). The Microsoft Court drew three basic lessons from Coopers & Lybrand. First, § 1291’s finality requirement is incompatible with a doctrine that effectively permits multiple interlocutory appeals.37×37. See id. at 1707. Second, § 1291 should be interpreted to preserve the policy compromises embodied in the existing rules governing interlocutory appeals.38×38. See id. at 1708. For example, in Coopers & Lybrand, the concern was that the death-knell doctrine sidestepped the “two-tiered ‘screening procedure’” set out in the Interlocutory Appeals Act of 1958 (current version at 28 U.S.C. § 1292(b) (2012)). See Microsoft, 137 S. Ct. at 1708 (quoting Coopers & Lybrand, 437 U.S. at 474). Third, § 1291 should not be interpreted to permit one party but not the other to seek interlocutory review.39×39. See Microsoft, 137 S. Ct. at 1708.

The Court then turned to its second guide to § 1291: Rule 23(f). “After Coopers & Lybrand,” the Court explained, “a party seeking immediate review of an adverse class-certification order had no easy recourse.”40×40. Id. No class action–specific rule existed. Litigants could attempt to convince two courts to hear their appeal under § 1292(b); they could litigate their individual claims to final judgment; or, in rare cases, they could seek review through a writ of mandamus. See id. Rule 23(f) was developed and ratified in this post–Coopers & Lybrand landscape, and embodied a “careful[ly] calibrat[ed]” compromise between the benefits of immediate review and the costs of interlocutory appeals.41×41. See id. at 1709. Specifically, Rule 23(f) permits both parties to seek review of adverse class certification orders, and it gives circuit courts “unfettered discretion” to grant review.42×42. See id. (quoting Fed. R. Civ. P. 23(f) advisory committee’s note to 1998 amendment). The Court also observed that the Rules Committee declined to provide appeals as of right and embraced the view that “death-knell” considerations are best accounted for by circuit courts exercising their discretion.43×43. See id. at 1709–10.

Last, the Court emphasized its pragmatic approach to § 1291. It first explained that the “final-judgment rule, now codified in § 1291,”44×44. Id. at 1712. ensures that “the whole case and every matter in controversy in it [is] decided in a single appeal.”45×45. Id. (quoting McLish v. Roff, 141 U.S. 661, 665–66 (1891)). At bottom, however, that rule was designed to achieve the appropriate balance between the work of trial and appellate courts, to minimize the costs of interlocutory appeals, and to “promote[] the efficient administration of justice.”46×46. Id. Given these goals, the Court reasoned, “finality is to be given a practical rather than a technical construction.”47×47. Id. (quoting Eisen v. Carlisle & Jacquelin, 417 U.S. 156, 171 (1974)). The Court explained that it has often rejected “efforts to stretch § 1291 to permit appeals of right that would erode the finality principle and disserve its objectives.”48×48. Id. In addition, the Court reasoned, Congress provided a rulemaking process to determine what will count as “final decisions” under § 1291 — and thus the Court should avoid changes created by “judicial decisions . . . or inventive litigation ploys.”49×49. Id. at 1714.

Given all that, the Court concluded that § 1291 should not be interpreted to technically permit what Coopers & Lybrand and Rule 23(f) effectively forbid. In other words, the Court’s practical approach to finality, together with its understanding of the careful calibration embodied in Rule 23(f), required it to conclude that the “final-judgment rule is not satisfied whenever a litigant persuades a district court to issue an order purporting to end the litigation.”50×50. Id. at 1715. And in Microsoft, Justice Ginsburg wrote, “the voluntary dismissal essayed by [Baker] does not qualify as a ‘final decision’ within the compass of § 1291.”51×51. Id. at 1707. The alternative, she reasoned, would reintroduce the very problems Coopers & Lybrand eliminated and permit litigants to rewrite the policy embodied in Rule 23(f).52×52. See id. at 1713–15. For example, the “voluntary-dismissal tactic,” id. at 1713, at issue in Microsoft promises to generate repeated appeals; it permits plaintiffs alone to immediately appeal adverse class certification rulings; and it removes the circuit court’s discretion to hear appeals. See id. at 1713–15. In closing, emphasized that “[i]t is not the prerogative of litigants or federal courts to disturb th[e] settlement” embodied in Rule 23(f).53×53. Id. at 1715. The Court also rejected Baker’s argument that United States v. Procter & Gamble Co., 356 U.S. 677 (1958), permitted the voluntary dismissal tactic. See Microsoft, 137 S. Ct. at 1715 n.11. The Court reasoned that Procter & Gamble involved neither class certification, id., nor the plaintiff’s “consent to [the] judgment against [it],” id. (second alteration in original) (quoting Procter & Gamble, 356 U.S. at 681).

Justice Thomas54×54. Justice Thomas was joined by Chief Justice Roberts and Justice Alito. wrote an opinion concurring in the judgment, in which he accepted the Court’s conclusion but rejected its reasoning.55×55. Microsoft, 137 S. Ct. at 1715–16 (Thomas, J., concurring in the judgment). Unlike the Court, Justice Thomas viewed Baker’s voluntary dismissal with prejudice as a “final decision” under § 1291.56×56. See id. at 1716. After all, he reasoned, the Court’s own precedent dictates that a district court’s decision is “final” when it “ends the litigation on the merits and leaves nothing for the court to do but execute the judgment,”57×57. Id. (quoting Catlin v. United States, 324 U.S. 229, 233 (1945)). and in this case the district court’s order did just that — it dismissed Baker’s individual claims with prejudice and “direct[ed] the Clerk to enter Judgment . . . and close th[e] case.”58×58. Id. (quoting Appendix to Petition for a Writ of Certiorari at 39a, Microsoft, 137 S. Ct. 1702 (No. 15-457)). To conclude otherwise, Justice Thomas continued, the Court had to “rely[] not on the text of § 1291 or th[e] Court’s precedents about finality, but on Rule 23(f) . . . [which] says nothing about finality.”59×59. Id. Justice Thomas agreed that Baker’s tactic would frustrate the purpose of Rule 23(f), but he denied that this was reason enough to “warp our understanding of finality under § 1291.”60×60. Id.

In Justice Thomas’s view, the Court should have based its judgment on Article III.61×61. Id. He noted first that the jurisdiction of federal courts extends only to “Cases” and “Controversies”62×62. Id. (quoting U.S. Const. art. III, § 2). where the parties have “actual” and “concrete” interests in conflict.63×63. Id. at 1716–17 (quoting Campbell-Ewald Co. v. Gomez, 136 S. Ct. 663, 669 (2016)). Justice Thomas then explained that these limits “den[y] federal courts the power to decide questions that cannot affect the rights of litigants.”64×64. Id. at 1717 (quoting Lewis v. Continental Bank Corp., 494 U.S. 472, 477 (1990)). When Baker volunteered to dismiss his claims with prejudice, Justice Thomas reasoned, he “consented to the judgment against [him] and disavowed any right to relief.”65×65. Id. The upshot, he continued, was that Baker and Microsoft “were no longer adverse to each other,”66×66. Id. and thus a court could not “affect the[ir] rights” in the relevant sense.67×67. Id. (alteration in original) (quoting Lewis, 494 U.S. at 477). Justice Thomas rejected the argument that Baker’s interest in reversing the district court’s order striking his class allegations sufficed for Article III jurisdiction. Baker sought only a “procedural mechanism” for litigating his individual claims.68×68. See id. And because those individual claims had become moot, Justice Thomas wrote, Baker had no further interest in the class mechanism.69×69. See id. Justice Thomas also rejected the argument that the Ninth Circuit could hear the appeal because Baker could “revive” his claims if he won on the grounds that there must be a controversy “at all stages of review.” Id. (quoting Campbell-Ewald Co. v. Gomez, 136 S. Ct. 663, 669 (2016)).

In Microsoft, the Court took a defensible stand against a litigation tactic that undercut § 1291 and Rule 23(f). Even so, parts of the Court’s opinion are puzzling — in particular, the opinion fails to acknowledge an important line of cases on finality, and it leaves the effect of the district court’s dismissal order uncertain. These puzzles call for explanation, not least because the Court might have reached the same basic result on neater doctrinal grounds. One plausible explanation, perhaps, is that the Court made these choices to avoid two issues: first, the divisive Article III question that Justice Thomas’s concurring opinion addressed; and second, the unsettled questions about when, exactly, a court may avoid divisive Article III issues in the first place. If that’s right, it underscores a point other commenters have made about Microsoft: when questions about Article III standing in putative class actions return to the Court’s docket, the Court might be unable to avoid them.

Start with what’s puzzling. First, Justice Ginsburg’s opinion does not even mention — let alone grapple with — Catlin v. United States.70×70. 324 U.S. 229 (1945). In Catlin, the Court held that a “final decision” under § 1291 is “generally . . . one which ends the litigation on the merits and leaves nothing for the court to do but execute the judgment.”71×71. Id. at 233. There are exceptions to the Catlin rule, to be sure, but they tend to be rare and explicit.72×72. See, e.g., Mohawk Indus., Inc. v. Carpenter, 558 U.S. 100, 106–07 (2009) (explaining the collateral order doctrine, which expands § 1291 to cover a “small category,” id. at 106, of decisions that are not litigation-ending (quoting Swint v. Chambers Cty. Comm’n, 514 U.S. 35, 42 (1995))). What’s more, the Court has continued to stress that “final decisions” are those through “which [the] district court disassociates itself from a case.”73×73. Gelboim v. Bank of Am. Corp., 135 S. Ct. 897, 902 (2015) (quoting Swint, 514 U.S. at 42). In 2015, for instance, Justice Ginsburg cautioned that although § 1291 is often given a “practical rather than a technical construction,”74×74. Id. (quoting Mohawk Indus., Inc., 558 U.S. at 106). its “core application is to rulings that terminate an action.”75×75. Id.

Second, though perhaps related, Justice Ginsburg’s opinion discussed the district court’s dismissal order as though it misfired — as though it tried but failed to extinguish Baker’s individual claims. To start, the Court concluded that it lacked § 1291 jurisdiction over the dismissal order, not just the order to strike Baker’s class allegations.76×76. See Microsoft, 137 S. Ct. at 1707 (“We hold that the voluntary dismissal essayed by respondents does not qualify as a ‘final decision’ within the compass of § 1291.” (emphasis added)). The Court then suggested that this was because — appearances notwithstanding — the dismissal order did not end the litigation. For example, the Court stated that Baker “essayed” to have his claims dismissed with prejudice (not that he did),77×77. Id. and that the district court’s order “purport[ed]” to end the litigation (not that it did).78×78. Id. at 1715; see also id. at 1712 (“[W]e hold that § 1291 does not countenance jurisdiction by these means . . . .” (emphasis added)). To be clear, these word choices don’t strictly require the misfire interpretation. But it would explain the Court’s holding and its curious hint that the district court might not have done what it appeared to do. In addition, the misfire interpretation would explain why the Court did not engage with Catlin: if the district court’s order merely purported to end the litigation, then the Court’s holding is a straightforward consequence of Catlin, not a striking departure from it.

To make these puzzles more vivid, notice that the Court might have reached the same basic result on neater doctrinal grounds. For example, the Court might have held that the district court’s dismissal order was final under the Catlin test and § 1291, but that its order striking Baker’s class allegations did not merge in that final judgment, and hence was not reviewable on appeal. In general, interlocutory orders are not reviewable until there is a “final decision,” at which point once-interlocutory orders “merge in” the final judgment.79×79. John’s Insulation, Inc. v. L. Addison & Assocs., Inc., 156 F.3d 101, 105 (1st Cir. 1998) (collecting cases); see also, e.g., Shannon v. Gen. Elec. Co., 186 F.3d 186, 192 (2d Cir. 1999) (“When a district court enters a final judgment in a case, interlocutory orders . . . typically merge with the judgment for purposes of appellate review.”). This “merger rule” is thought to strike the right balance between appellate and trial courts, and between comprehensive and consolidated review.80×80. See, e.g., Shannon, 186 F.3d at 192. But the rule has exceptions. Most relevant here, circuit courts have often held that plaintiffs can’t secure immediate review of adverse interlocutory orders — including adverse class certification orders — by giving up and waiting for the district court to dismiss their claims for failure to prosecute.81×81. See, e.g., Huey v. Teledyne, Inc., 608 F.2d 1234, 1240 (9th Cir. 1979) (“Where . . . the denial of class certification caused the failure to prosecute, that ruling does not merge in the final judgment for purposes of appellate review . . . .”); see also Bowe v. First of Denver Mortg. Inv’rs, 613 F.2d 798, 801 (10th Cir. 1980) (quoting Huey approvingly). But see Gary Plastic Packaging Corp. v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 903 F.2d 176, 178–79 (2d Cir. 1990) (holding that an adverse class certification order merged despite dismissal for failure to prosecute). As then-Judge Sotomayor explained, applying the merger rule in such cases would “severely weaken[]” the “policy against piecemeal litigation and review” and “would in effect provide a means to avoid the finality rule embodied in . . . § 1291.”82×82. Shannon, 186 F.3d at 192 (quoting Palmieri v. Defaria, 88 F.3d 136, 139 (2d Cir. 1996)). In Shannon, then-Judge Sotomayor also cast doubt on Gary Plastic. See id. at 192–93. These concerns echo the Court’s concerns in Microsoft, of course, and might have underwritten an analogous exception to the merger rule.

Indeed, extending the “no merger” rationale to Microsoft seems to have some clear advantages,83×83. For example, an equitable exception to the merger rule is perhaps a less controversial outlet for the Court’s policy concerns than § 1291’s text, the exception would buttress Rule 23(f), and it would fully accommodate Catlin. and there is good reason to think it was on the Court’s radar.84×84. Several of the cases constituting the circuit split discussed the question in those terms. See, e.g., Camesi v. Univ. of Pittsburgh Med. Ctr., 729 F.3d 239, 244–45 (3d Cir. 2013); Gary Plastic, 903 F.2d at 178–79; Huey, 608 F.2d at 1239–40. But it is worth noting that the principal briefs did not frame the issue that way. Why, then, might the Court have declined to adopt it? There are several possible reasons,85×85. For example, perhaps the Court thought that, in the class action context, the gamesmanship involved in failing to prosecute was different enough from the gamesmanship involved in seeking voluntary dismissal. Cf. John’s Insulation, Inc. v. L. Addison & Assocs., Inc., 156 F.3d 101, 107 (1st Cir. 1998) (distinguishing dismissals for failure to prosecute from voluntary dismissals on policy grounds in non–class litigation). of course, and reverse engineering is a difficult business. Even so, one intriguing hypothesis is that the Court’s actual rationale enabled it to better avoid the difficult and divisive Article III issue in the case: namely, if Baker’s individual claims were dismissed at his invitation, did he have Article III standing to challenge the district court’s order striking his class allegations? The Court has brushed up against related questions in recent cases,86×86. See, e.g., Genesis Healthcare Corp. v. Symczyk, 569 U.S. 66, 73 (2013) (holding that “respondent’s [FLSA collective action] suit became moot when her individual claim became moot”). But see id. at 84–87 (Kagan, J., dissenting) (disputing the Court’s assumption that respondent’s individual claim was moot and criticizing the conclusion the Court drew from it). and those encounters suggest a Court divided on issues that could have important and hard-to-predict implications for plaintiffs in class actions and similar suits.87×87. The principal question is whether the Court will revisit Deposit Guaranty National Bank v. Roper, 445 U.S. 326 (1980), and United States Parole Commission v. Geraghty, 445 U.S. 388 (1980), where it recognized that a “proposed [class] representative retains a ‘personal stake’ in obtaining class certification sufficient to assure that Art. III values are not undermined,” id. at 404, even after the named plaintiff’s individual claims have become moot. See, e.g., Archis A. Parasharami, Supreme Court Rejects End Runs Around Rule 23(f) by Use of “Voluntary Dismissal” Tactic, Class Defense Blog (June 12, 2017), https://www.classdefenseblog.com/2017/06/supreme-court-rejects-end-runs-around-rule-23f-use-voluntary-dismissal-tactic/ [https://perma.cc/3MBH-KFLG]; Adam Zimmerman, Some Thoughts on Microsoft v. Baker, Summary Judgments (June 13, 2017), http://summaryjudgments.lls.edu/2017/06/some-thoughts-on-microsoft-v-baker.html [https://perma.cc/6UPA-A3NH].

With that in mind, reconsider the Court’s options. If it held that the district court’s dismissal order misfired, it would make the Article III question in the case merely hypothetical. In other words, if Baker failed to get his individual claims dismissed, then the principal Article III question would have a false antecedent. In contrast, the “no merger” rationale does not secure that result: rather, it assumes that Baker had his individual claims dismissed, which would make the Article III question nonhypothetical. Does this difference matter to the Court? The thought that it might draws support from Campbell-Ewald Co. v. Gomez.88×88. 136 S. Ct. 663 (2016). In Campbell-Ewald, Justice Ginsburg — joined by the same four Justices that joined her in Microsoft — rejected a defendant-side class action gambit without reaching an Article III question that was only “hypothetical.”89×89. Id. at 672. In Campbell-Ewald, the Court held that defendants in putative class actions could not moot plaintiffs’ claims by simply offering to make the named plaintiff whole. See id. The Court postponed judgment on a related Article III question that was “appropriately reserved for a case in which it is not hypothetical.” Id. (emphasis added). The proposed explanation, then, is that the five-Justice coalition in Campbell-Ewald and Microsoft was prepared to curtail certain class action tactics without reaching Article III issues, so long as those issues could be rendered hypothetical. And in Microsoft, that position led the Court to prefer its rationale to a “no merger” alternative.

But why so much effort to render the Article III issue hypothetical? After all, couldn’t the Court have postponed the Article III issue on other grounds, even if it had created an exception to the merger rule? Yes, in principle. For instance, the Court might have reasoned that it simply didn’t have to reach the constitutional question to decide the case. That might even seem like an attractive position, given familiar reasons for avoiding constitutional questions.90×90. See, e.g., Ashwander v. Tenn. Valley Auth., 297 U.S. 288, 346–48 (1936) (Brandeis, J., concurring). The problem, however, is that Article III is different: to avoid a question about its constitutional jurisdiction, the Court would have to wade through unsettled questions about when it may depart from “the rule that Article III jurisdiction is always an antecedent question”91×91. Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 101 (1998).  — questions, that is, about when courts may prioritize some jurisdictional issues and bracket others.92×92. See id. at 93–94 (explaining that courts may not address the merits before subject matter jurisdiction); see also Ruhrgas AG v. Marathon Oil Co., 526 U.S. 574, 583–88 (1999) (explaining that courts may sometimes address personal jurisdiction before subject matter jurisdiction). See generally Scott C. Idleman, The Emergence of Jurisdictional Resequencing in the Federal Courts, 87 Cornell L. Rev. 1 (2001) (explaining and criticizing this line of cases). To be clear, the Court might have found its way through these questions. The idea that § 1291 jurisdiction may precede Article III jurisdiction in Baker’s situation is not foreclosed by sense or precedent. Even so, the question divides scholars,93×93. See, e.g., Heather Elliott, Jurisdictional Resequencing and Restraint, 43 New Eng. L. Rev. 725, 743–44 (2009) (noting that even critics of sequencing allow that statutory subject matter jurisdiction may be addressed first “so long as there is no underlying problem of Article III jurisdiction,” then proceeding to argue that sequencing should be permitted even when there might be an underlying Article III problem); Alan M. Trammell, Jurisdictional Sequencing, 47 Ga. L. Rev. 1099, 1116–35 (2013) (canvassing several positions, including the view — adopted by at least one circuit — that Article III jurisdiction has “absolute priority,” id. at 1125, even over other elements of subject matter jurisdiction, see id. at 1125–28). and the Justices’ stated views don’t suggest an obvious consensus position.94×94. For instance, compare Steel Co., 523 U.S. 83, where Justices Kennedy, Ginsburg, and Breyer indicated slightly different positions, with Ruhrgas, 526 U.S. 574, where Justice Ginsburg spoke for a unanimous Court. Justices Sotomayor and Kagan appear not to have addressed jurisdictional sequencing at length. Given that, there would be a nontrivial benefit to rendering the Article III question hypothetical: it would enable the Court to steer clear of it entirely.

In all, then, the Court’s somewhat curious opinion in Microsoft makes more sense if the Court wanted to dispense with Baker’s “voluntary-dismissal tactic,”95×95. Microsoft, 137 S. Ct. at 1713. but wished to do so without facing the difficult Article III question lurking in the case — or even the difficult questions about avoiding difficult Article III questions. This interpretation is (admittedly) speculative. If it’s on the right track, however, it would underscore a conclusion that several commentators have already drawn from Microsoft: that the Court appears increasingly ready to reconsider Article III standing for putative class representatives when it next gets the chance.96×96. See, e.g., Alison Frankel, Lurking in Latest SCOTUS Class Action Ruling: Long-Running Question of Standing, Reuters (June 12, 2017, 3:53 PM), http://www.reuters.com/article/us-otc-standing-idUSKBN1932DV [https://perma.cc/7DW8-4KKN]; Parasharami, supra note 87; Zimmerman, supra note 87. After all, if Microsoft sheds light on how difficult it might be for the Court to avoid the issue going forward, then the Court might get that chance sooner rather than later.