Eric Rothner D/B/A Chicago Game Co., and D/B/A Bell Vending, Inc. v. City of Chicago, a Municipal Corporation

EASTERBROOK, Circuit Judge,

dissenting.

Two other appellate courts have considered whether a district judge’s order remanding a case to state court on account of waiver implied from the course of litigation may be reviewed by the court of appeals. Both have answered “no”. In re Weaver, 610 F.2d 335 (5th Cir.1980); Schmitt v. Insurance Co. of North America, 845 F.2d 1546 (9th Cir.1988). Today we create a conflict. Circuits should not strike off on their own on jurisdictional questions without powerful reasons. Newman-Green, Inc. v. Alfonzo-Larrain, — U.S. -, -, 109 S.Ct. 2218, 2220 , 104 L.Ed.2d 893 (1989); FDIC v. Elefant, 790 F.2d 661, 665-66 (7th Cir.1986). We don’t have one, for 28 U.S.C. § 1447 (d) supports the Fifth and Ninth Circuits as strongly as a text ever does: “An order remanding a case to the State court from which it was removed is not reviewable on appeal or otherwise ... ”.

Section 1147(d) fortifies principles that defer review of orders allocating cases among forums. Orders transferring cases from one court to another or declining to do so, orders compelling arbitration, and the like are not appealable because they do not conclude the litigation. E.g., Lauro Lines s.r.l. v. Chasser, — U.S. -, 109 S.Ct. 1976 , 104 L.Ed.2d 548 (1989); Van Cauwenberghe v. Biard, 486 U.S. 517 , 108 S.Ct. 1945, 1952-54 , 100 L.Ed.2d 517 (1988); Gulfstream Aerospace Corp. v. Mayacamas Corp., 485 U.S. 271 , 108 S.Ct. 1133, 1136-38 , 99 L.Ed.2d 296 (1988). That remand orders in particular are not appeal-able has been settled since 1872. See Insurance Co. v. Comstock, 83 U.S. (16 Wall.) 258, 270 , 21 L.Ed. 493 (1872); Railroad Co. v. Wiswall, 90 U.S. (23 Wall.) 507 , 23 L.Ed. 103 (1874), neither of which depended on the predecessors to § 1447(d). See also, e.g., McLaughlin v. Hallowell, 228 U.S. 278, 286 , 33 S.Ct. 465, 468 , 57 L.Ed. 835 (1913). Remands may be reviewed, if at all, by mandamus, an extraordinary remedy requiring a correspondingly extraordinary justification — such as lack of power to act. Error in the exercise of *1420 admitted power won’t do. Kerr v. United States District Court, 426 U.S. 894 , 402-03, 96 S.Ct. 2119, 2123-24 , 48 L.Ed.2d 725 (1976).

My colleagues lack a powerful argument driven by a statute; they have instead an ingenious argument to avoid a statute. Novel as well as ingenious: neither party to this case nor any other judge has advanced it. The argument goes in four steps.

First, the Supreme Court held in Thermtron Products, Inc. v. Hermansdorfer, 423 U.S. 336 , 96 S.Ct. 584 , 46 L.Ed.2d 542 (1976), that § 1447(d) does not mean what it says, that it forbids appellate review only when the remand was authorized by § 1447(c), which speaks of cases removed “improvidently and without jurisdiction”. Because the majority in Thermtron believed that the sets “remands authorized by § 1447(c)” and “lawful remands” are identical, Thermtron would require us to decide whether cases ever may be remanded on account of waiver. If yes, appellate review is barred in every case. Gravitt v. Southwestern Bell Telephone Co., 430 U.S. 723 , 97 S.Ct. 1439 , 52 L.Ed.2d 1 (1977). Courts have been remanding on account of waiver for more than a century, without a peep from Congress. The Removal Cases, 100 U.S. 457, 473 , 25 L.Ed. 593 (1879) (dictum); Karl Koch Erecting Co. v. New York, 838 F.2d 656 (2d Cir.1988); Texas Wool & Mohair Marketing Ass’n v. Standard Insurance Co., 175 F.2d 835 (5th Cir.1949); Atlanta, Knoxville & Northern Ry. v. Southern Ry., 131 F. 657, 660-63 (6th Cir.1904) (Lurton, J.). The majority does not deny that waiver is a lawful ground of remand. So Thermtron does not authorize appellate review in our case.

Step Two. In Carnegie-Mellon University v. Cohill, 484 U.S. 343 , 108 S.Ct. 614 , 98 L.Ed.2d 720 (1988), over the dissent of the author of Thermtron, the Court broke the link between “lawful remands” and “remands authorized by § 1447(c)”. Carnegie-Mellon left us with three categories of grounds for remand: (1) those authorized by § 1447(c) and beyond the power of appellate review; (2) those not authorized by § 1447(c) but nonetheless lawful in principle and reviewable for error in execution (the example from Carnegie-Mellon is a remand of a pendent state claim after deciding the federal claim that authorized the removal); (3) those not authorized by § 1447(c) or anything else, and subject to automatic reversal (the example from Thermtron is a remand of a suit the district court thinks it is too busy to hear). After Carnegie-Mellon we must decide whether waiver makes removal “improvident” so that § 1447(c) authorizes remand. For the century prior to Carnegie-Mellon — when only grounds given by § 1447(c) allowed remand — the conclusion that remand for waiver was ever lawful would show that § 1447(c) supplies the power. Camegie-Mellon did not change the source of the power to remand on account of waiver. I would reach this point and stop, dismissing the appeal. My colleagues go on to

Step Three. When is removal “improvident” within the meaning of § 1447(c)? One attractive possibility is that removal is “improvident” whenever a case is “not removable on the day the petition is filed”. Waiver fits that bill, quite unlike a case such as Camegie-Mellon in which removal was proper and remand was justified by subsequent events. The majority takes a different tack, however, equating “improvident” with defective procedures. Section 1446 establishes hoops through which a petitioner must jump, and my colleagues conclude that only the failure to take a step required by § 1446 makes removal “improvident” under § 1447(c).

Step Four is identifying the requirements of § 1446. That statute does not mention waiver. Section 1446(b), though, insists that the petition be timely. Waiver rules originated in cases dealing with timely removal. Courts said “waiver” when they concluded that counsel had waited too long to remove, given what was happening in state court (maj. op. at 1412-16). Despite this history, the majority concludes that waiting too long to remove is not the kind of timeliness demanded by § 1446(b), so that delay (= waiver) cannot make removal *1421 “improvident” within the meaning of § 1447(c).

Steps One and Two follow higher authority. Steps Three and Four are of our own choosing. They are not demonstrably wrong-but they are not demonstrably right, either. Step Three, equating “improvident” with “failure to do what § 1446 demands”, does not flow from the text or history of § 1447(c). Section 1447(c) does not refer to § 1446; the removal provisions as a whole lack definitions of key terms; nothing in the legislative history of the statute mentions the relation, if any, between § 1446 and § 1447(c). Step Four, excluding waiver from the timeliness requirement of § 1446(b), also lacks support in text or history.

Without pretending assurance, I think we err in adopting Steps Three and Four. “Improvident” implies more than “procedurally defective”; in ordinary usage something is “improvident” if it ought not have been done. A case removed at a time when either substantive or procedural rules block that step has been removed “improvidently”. If so, Step Three is incorrect. More than just ordinary language undermines Step Three. Section 1446(d) requires a bond to accompany the petition for removal. If “improvident” means “out of compliance with § 1446”, then any case unaccompanied by a bond must be remanded (§ 1447(c) says that the court “shall remand” cases removed improvidently). Yet why should one day’s delay in posting a bond forfeit the right to remove? Many courts, including ours, hold that the bond may be supplied as soon as the omission comes to light, Tucker v. Kerner, 186 F.2d 79 (7th Cir.1950); see also, e.g., Climax Chemical Co. v. C.F. Braun & Co., 370 F.2d 616 (10th Cir.1966), National Quicksilver Corp. v. World Insurance Co., 139 F.2d 1 (8th Cir.1944), though under the majority’s position all such cases must be remanded. So too with any other technical defect-for example, failure to serve a copy of the original pleading with the petition to remand, § 1446(b), or failure to verify the petition, § 1446(a) although these are poor grounds on which to forfeit the privilege of removal if quickly rectified. 1

My colleagues rely on the 1988 revision of § 1447(c), Judicial Improvements and Access to Justice Act of 1988, Pub.L. 100-702, 102 Stat. 4642 (1988). This statute supports my approach rather than theirs. Consider the problem of timeliness. The new version refers to motions to remand “on the basis of any defect in removal procedure” and then distinguishes between cases removed without subject-matter jurisdiction, which may be remanded at any time before final judgment, and others, which may be remanded only if a motion is made within 30 days of the removal. Congress treated “defect in removal procedure” as synonymous with “improvidently and without jurisdiction” and redrafted the section only to create the different time limits. See H.R.Rep. No. 100-889 Part I, 72 (1988), U.S.Code Cong. & Admin.News 1988, p. 6033. Is a remand on the ground of untimely removal one based on a “defect in the removal procedure” under the new version of the statute? Yes, I hope, for it would be senseless to review such decisions on appeal while not reviewing other closely related judgments. Untimeliness in state court ought to be raised ASAP in federal court. Unless we treat timeliness as a “defect in the removal procedure”, no time *1422 limit will apply to bringing the point to the district court’s attention. Yet once we conclude that untimely removal is a “defect in the removal procedure”, it must follow that waiver too may be such a defect. For waiver inferred from conduct is a conclusion that the defendant waited too long in light of events taking place in state court. Waiver of the sort we are examining is a conclusion that the defendant’s time ran out in advance of the limit in the statute. So viewed, it is a defect in removal procedure covered by both the current and the former versions of § 1447(c). 2

No matter what one makes of the majority’s arguments for Steps Three and Four, however, the question remains: why struggle against § 1447(d)? We are not caught in a pinch between sound judicial administration and faithful execution of the statute. The policy behind § 1447(d) is that interlocutory review of decisions allocating cases among forums delays proceedings— this appeal has delayed proceedings — without resolving or even advancing the resolution of the merits. All of the rationales that constrict interlocutory review apply to orders allocating cases among tribunals, and § 1447(d) makes pellucid what ought to have been clear in its absence. Thermtron responded to the problem of the lawless judge, one who said in effect: “the law be damned, I can do what I want if there is no review”. It created a categorical rule limiting the reasons for remand. Carnegie-Mellon responded to the problem that the adjudication of claims properly pendent to a federal claim might be deferred or jeopardized if remand were impossible, and nothing in § 1447(c) says that “improvidently and without jurisdiction” are the only grounds of remand. Steps One and Two serve understandable functions. No similarly strong reason supports Steps Three and Four. Although courts strain against the limits of their authority when the reasons seem pressing (here, because the majority is convinced that the remand was mistaken and the judge wrong to allow the interim relief to continue in force), jurisdictional rules apply to future cases too. District judges rarely remand cases in error. After today’s decision, however, many cases must come to a halt pending appellate review, although the great majority will not be afflicted with error. Disputation over the identity of the right forum will stall litigation that ought to proceed to decision.

Even with § 1447(d) cleared away, we still don’t have appellate jurisdiction. As Thermtron holds, 423 U.S. at 352-53 , 96 S.Ct. at 593-94, it is mandamus or nothing. Mandamus depends on a demonstration of want of power, and the majority does not believe that the district judge lacked power to remand on account of waiver. My colleagues believe that the judge erred in the exercise of a power he possesses, which does not support an extraordinary writ. E.g., Mallard v. United States District Court, — U.S. -, 109 S.Ct. 1814, 1822 , 104 L.Ed.2d 318 (1989); Allied Chemical Corp. v. Daiflon, Inc., 449 U.S. 33 , 101 S.Ct. 188 , 66 L.Ed.2d 193 (1980); Kerr.

Unreviewable remands may deprive defendants of a federal forum to which they are entitled. Review of remands certainly delays the completion of litigation and increases its expense. How to accommodate these two kinds of costs? One way is to say “tough luck, but the decision of a single judge is all the subject calls for”. Four judges may be too many, or too slow, for particular decisions. Think of the “right” not to undergo expensive discovery — or for *1423 that matter the right in Lauro Lines to litigate in Italy rather than New York. An entitlement to litigate in one court rather than another may be valuable, but it is not so important that the case should come to a halt while an appellate court examines the claim. As Justice Scalia said, concurring in Lauro Lines, 109 S.Ct. at 1980 , “the right is not sufficiently important to overcome the policies militating against interlocutory appeals.” Even a grievous error does not authorize appellate review. Gravitt holds exactly this concerning § 1447(d).

Another approach is to allocate the risk of error according to the value attached to the claim, a legislative task on which Congress embarked. Section 1447(d) contains an exception. If the removal is based on 28 U.S.C. § 1443 , the civil rights removal statute, then remands are appealable immediately. In cases covered by § 1443, the plaintiff who wanted to be in state court bears the costs of delay; in all other cases, the defendant who wanted to be in federal court bears the risk of error while both sides obtain a faster final disposition. Chicago does not contend that § 1443 supplied authority for removal. Perhaps Congress ought to prefer the person claiming the right to the federal forum — much of the majority’s rationale, maj. op. at 1407-08, is based on the supposition that this is the optimal allocation — but the statute does not read that way.

Although my colleagues hint darkly that the district judge may have remanded the case because of hostility to removal by governmental defendants, rather than for the stated reason, Chicago disclaimed at oral argument any contention that the judge’s reasons were other than those he gave. I am unaware of a pattern of questionable remands by Judge Bua; we have no basis on which to doubt the candor of his description of reasons for remanding. The case presented by the parties and the record is one of remand because of waiver, nothing else. 3

Thermtron and Camegie-Mellon do not launch the federal courts on a trajectory producing ever-more appellate review. Gravity brings legal trends back to ground, just as it ends the flight of a knuckle ball. Even after a surprising and “liberating” construction of a law, decisions still must be founded on the statute rather than on projections from the first case. The problem is familiar:

[T]he Court, having already sanctioned a point of departure that is genuinely not to be found within the language of the statute, finds itself cut off from that authoritative source of the law, and ends up construing not the statute but its own construction. Applied to an erroneous point of departure, the logical reasoning that is ordinarily the mechanism of judicial adherence to the rule of law perversely carries the Court further and further from the meaning of the statute. Some distancé down that path, however, there comes a point at which a later incremental step, again rational in itself, leads to a result so far removed from the statute that obedience to the text must overcome fidelity to logic.

NLRB v. Electrical Workers, 481 U.S. 573, 597-98 , 107 S.Ct. 2002, 2016-17 , 95 L.Ed.2d 557 (1987) (Scalia, J., concurring). See also Hickey v. Duffy, 827 F.2d 234, 242-43 (7th Cir.1987). Neither precedent nor logic compels us to move farther and farther from the text of § 1447(d). Abhorrence of interlocutory review of procedural issues *1424 ought to lead us to carry out the rule laid down in the law. Two courts of appeals have done this. So should we.

. Ryan v. State Board of Elections, 661 F.2d 1130, 1133 (7th Cir.1981), is not so much inconsistent with this perspective as it is inconsistent with Camegie-Mellon. Ryan concluded that only the grounds specified in § 1447(c) permitted remand, a position that has now been rejected. The district judge in Ryan had remanded a case on the ground of abstention. It didn’t take much effort to jettison that, and the holding (as opposed to the rationale) that you can’t "abstain” by remanding a case doubtless survives Camegie-Mellon. The language in Ryan that the majority quotes was not only unnecessary to the disposition but also not inconsistent with the position that an order remanding on account of waiver sensibly could be characterized as a conclusion that the removing party waited too long in light of the progress of the litigation. Timeliness is one of the subjects covered by § 1447(c). Too, the only source for the language in Ryan is a decision of the Fifth Circuit —which has since held, in Weaver , that an order remanding on account of waiver is unreviewable under § 1447(d). If we go by patrimony, the unreasoned assertion in Ryan does not counsel against following Weaver .

. Two courts hold that waivers contained in pre-litigation documents are not encompassed by § 1447(d). See Clorox and Pelleport Investors, Inc. v. Budco Quality Theatres, Inc., 741 F.2d 273 (1984), both from the Ninth Circuit, and two cases from the Second Circuit, Karl Koch and Corcoran v. Ardra Insurance Co., 842 F.2d 31 (2d Cir.1988). These cases treat remands based on contractual provisions as preliminary dispositions of the merits and therefore outside the scope of § 1447(c). Schmitt distinguished Clorox and Pelleport on these grounds when holding that waiver inferred from conduct in state court is within § 1447(c). 845 F.2d at 1549-50 . We needn't arbitrate this contrariety but, if Step Three is justified, then the distinction may be supported on the ground that waiver during litigation is a question of timeliness under § 1446(b), while waiver in pre-litigation documents may not be so characterized.

. A belief that plaintiffs should be entitled to litigate constitutional claims against local governments in state court is not unreasonable, even if it is not a proper ground for remand. A battery of doctrines funnels constitutional cases to state courts, or requires federal judges to abstain, certify issues, or defer to state courts’ decisions; state and local governments also enjoy in federal courts immunity defenses and the shield of the eleventh amendment, which complicates litigation in federal court and may deny the plaintiff a remedy available in state court. Constitutional analysis often could be avoided by clarification of state and local law, a task that state courts can perform and federal courts cannot. If the plaintiff wants to pursue constitutional litigation in state court, it is hard to see why he should be frustrated. Yet 28 U.S.C. § 1445 , which sets limits on removals, does not preclude a state or local government from removing a case of this kind. Until Congress revises § 1445, Chicago is entitled to a federal forum unless it waived the privilege.