The Soo Line R.R. bought the railroad operations of the bankrupt Chicago, Milwaukee, St. Paul & Pacific R.R. in February 1985 for $150 million plus the assumption of liabilities estimated at $420 million. The district court approved the Soo’s bid even though the Chicago & North Western Transportation Co. (CNW) bid $360 million plus assumption of essentially the same liabilities, and even though the Interstate Commerce Commission had found both proposed acquisitions to be consistent with the public interest. The district court thought the Soo’s bid preferable because the Soo planned to abandon less of the Milwaukee’s track — abandonments that the district court thought might impair the country’s military security.
The investors in the Milwaukee, not understanding why they should be required to contribute $210 million to the national defense, contend that the district court’s approval of the sale “took” their property. They have asked the Claims Court, which under 28 U.S.C. § 1491 has exclusive jurisdiction of money claims against the United States arising out of takings, for relief. They ask us to set aside a paragraph in the district court’s order that approved a finding of the ICC that the Soo’s bid was constitutionally adequate, a paragraph that will impair if not preclude their suit in the Claims Court.
The Railway Labor Executives’ Association (RLEA) challenges the sale, too, but on different grounds: it says that the district court did not impose sufficient conditions for the preservation of employment of the Milwaukee’s employees. The RLEA also asks us to set aside an order of the ICC declining to review the Milwaukee’s final plan of reorganization. Finally, the Seaboard Line wants us to set aside the dis*319trict court’s decision (and an order of the ICC) to the extent they allow the Soo to operate trains over the Seaboard’s line between Bedford and New Albany, Indiana. These trackage rights enable the Soo to haul traffic from Chicago to Louisville, traffic the Seaboard otherwise might move.
Because the challenges to the sale are so disparate, we treat the facts pertinent to each challenge separately. Part I addresses the investors’ constitutional argument, Part II the RLEA’s arguments about protective conditions, and Part III the Seaboard’s request to reclaim control of its track.
I
A
The Trustee of the Milwaukee managed its business for several years, paring unprofitable operations and refurbishing deteriorating track. The ICC and the reorganization court decided that the Milwaukee would not be a successful line if operated alone, so the Trustee put its railroad lines and related rail assets up for bids. Three bidders remained at the end: the Soo Line, the Grand Trunk, and the CNW. The ICC, whose approval was required under § 5(b) of the Milwaukee Railroad Restructuring Act, 45 U.S.C. § 904(b), and the Interstate Commerce Act, 49 U.S.C. § 11344, rejected the Grand Trunk’s proposal. Chicago, Milwaukee, St. Paul & Pacific R.R.—Reorganization, Fin.Dkt. No. 28640,—I.C.C.2d—(Sept. 26, 1984). In the same decision the ICC expressed concern that the CNW’s proposal would eliminate some competition (the Milwaukee competed with the CNW in many markets), while acquisition by the Soo would have a smaller competitive effect. It therefore proposed conditions that would have required the CNW to grant trackage rights, allowing other railroads to operate over its lines to furnish competing service. Chairman Taylor and Vice Chairman Andre approved the Soo’s proposal but did not state their reasons (except to the extent they may be inferred from the ICC’s exceedingly lengthy analysis of all the proposals). Commissioner Gradison favored the Soo’s proposal only because it required fewer conditions; she thought the public benefits of the two proposals were otherwise equal. Commissioner Sterrett favored the CNW’s proposal. The ICC “approved” only the Soo’s proposal but noted that its failure to disapprove the CNW’s left the reorganization court with the discretion to choose either.
One of the issues before the ICC was the adequacy of the consideration offered by the bidders. Both Soo and CNW offered some $150 million in addition to the assumption of debts. Chicago Milwaukee Corporation (since renamed CMC Real Estate Corp.), the entity that was to emerge with the Milwaukee’s non-rail assets and the net payment from the winning bidder (and thus the representative of the Milwaukee’s equity investors), objected that $150 million was too little. The net value of the railroad assets was at least $360 million, according to CMC. The ICC rejected the CMC’s position for two reasons: first, the $360 million valuation was the result of discounting to present value an assumed flow of future earnings, a methodology the Commission thought unreliable; second, the value did not represent any actual bid that CMC was willing to accept. The ICC instead decided to value the Milwaukee’s rail assets by looking for a multiple of current earnings. A bid of $150 million represented a multiple of 16.9 times the Milwaukee’s current earnings and eight times its projected earnings for 1984. The multiple of 16.9 was almost twice the average of five large carriers reflected in stock market valuations in 1983. CMC’s asking price of $360 million, more than 40 times current earnings, struck the ICC as unrealistic.
CNW then met CMC’s “unrealistic” demand, offering $360 million (in addition to the assumption of more than $420 million in liabilities) for the Milwaukee’s rail operations. CMC and the Trustee declared they were satisfied. The ICC considered this revised bid, reopened the former decision (which had suspended judgment on the CNW’s proposal) and formally approved *320CNW’s bid.—I.C.C.2d—(Jan. 8, 1985). It concluded that CNW could realize operational savings of $123 million without abandoning any lines. But by a vote of 4-3 the ICC again favored the acquisition by the Soo. The majority (Chairman Taylor and Commissioners Simmons, Lamboley, and Strenio) concluded that the Soo’s bid, though $210 million less than the CNW’s, remained fair and constitutionally adequate because it exceeded the net liquidation value of the Milwaukee’s rail properties and the average price-earnings multiple of other railroads. The majority was divided, with Chairman Taylor and Commissioner Simmons voting to disapprove the CNW’s bid, largely on the ground that acquisition by the Soo would create new competition while acquisition by CNW would not and, without the use of conditions and trackage rights, would reduce competition. Commissioner Strenio, on the other hand, thought the matter “extremely close” and explained that he favored the acquisition by Soo because it “calls for more head-to-head competition between rail carriers”. Commissioner Gradison favored the CNW’s proposal because of its higher price, raising doubts that the Soo’s could be called “fair” to CMC. Vice Chairman Andre and Commissioner Sterrett, although favoring CNW’s bid, did not explain their positions.
The reorganization court spoke next, approving the Soo’s bid despite the request by the Trustee and CMC that it sell to the CNW. The judge did not write an opinion explaining why. His oral remarks state that he was concerned that CNW planned to file with the ICC an application to abandon as much as 1,100 miles of track if they should prove unprofitable. (CNW had not asked for permission to abandon any particular track, and any concrete proposals would have triggered their own proceedings.) The district court objected to the potential abandonment on several grounds.
Now, what is the matter with abandonment? Well, for one thing it costs jobs. People are going to be out of work that were working on the abandoned transportation system or routes. Another consideration that hasn’t been, really, considered very much but was mentioned by one of the Commissioners ... is the matter of national security.
... I would say once a railroad is abandoned once a piece of trackage is pulled up and the land sold it isn’t going to be replaced. It is almost irresponsible under the international situation that we are faced with today to tear up pieces of a railroad system, which is probably the best in the world, for no better reason than a few million dollars in the bidding price.
I must say that I think that national security is a very important consideration for keeping an extensive and more viable system of transportation____
A few hundred million here and a few hundred million there, and we may be talking real money. (Apologies to Everett Dirksen.) But the Department of Defense had not expressed the view that abandonment of track is contrary to interests of national security, and Congress recently made it easier to abandon unprofitable track. See Hayfield Northern R.R. v. Chicago & North Western Transportation Co., 467 U.S. 622, 628-31, 104 S.Ct. 2610,. 2614-16, 81 L.Ed.2d 527 (1984) (describing the evolution of statutes concerning abandonment); cf. Chicago & North Western Transportation Co. v. Kalo Brick & Tile Co., 450 U.S. 311, 319-23, 101 S.Ct. 1124, 1131-33, 67 L.Ed.2d 258 (1981).
The court gave two other reasons for preferring Soo’s bid. One is that the track-age rights that the ICC had recommended to counteract competitive problems in acquisition by the CNW would require administrative and judicial supervision; the court thought that the less supervision the better. The other is that the Soo’s parent corporation is the Canadian Pacific Railway. The court stated:
Let’s face it. The United States needs foreign trade and needs friends. We are no longer an unsinkable battleship, as they called the British Isles in World War II. If there is going to be another World War, the United States and our *321transportation system and our cities are going to be targets for the enemy to hit. If we have friends in Canada, we have access to Canada, we have one more way of defending ourselves. But aside from that because I’m not an alarmist about international affairs, it certainly does promote international trade and we certainly are in need of that.
Having established its foreign policy, the court considered whether the Soo’s bid was constitutionally adequate. The court relied on what it characterized as the ICC’s finding that each bid was “well above the fair market value” of the railroad properties of the Milwaukee. This created a problem, because the CNW had made the higher bid, and actual bids usually establish the “fair market value” of the property; the ICC’s initial decision had relied on the absence of a bid higher than the Soo’s to show that CMC’s valuation was unrealistic. The district judge dismissed the extra $210 million as a “windfall” that was “not the property of the CMC” and “not by any means a controlling factor in which railroads should prevail in this particular contest.” Although the ICC had held that an acquisition by the CNW would not violate the antitrust laws, a position that the Antitrust Division, representing the United States in this litigation, does not contest, a position the district court apparently accepted, the judge also offered the view that the $210 million difference reflected only the price at which the CNW could “buy out some of its competition”.
The court then entered an order directing that the Milwaukee’s rail assets be sold to the Soo. Paragraph 4 of this order states that the court has considered all of the objections to the ICC’s findings and recommendations and that:
the Commission’s findings and decisions should be and are approved by this Court, since the findings and decisions were in accordance with law and were not arbitrary, capricious, an abuse of discretion, contrary to constitutional right, power, privilege or immunity, in excess of statutory jurisdiction, authority or limitations, short of statutory right, or without observance of procedure required by law.
The court apparently meant this paragraph to adopt the ICC’s conclusion that the Soo’s price was constitutionally sufficient and thus dispose of CMC’s argument about the inadequacy of the price, because no other part of the order approving the sale deals with the subject of price.
B
After several parties asked for (and did not get) a stay from this court, Milwaukee and the Soo closed the sale on February 20, 1985. CNW has withdrawn its bid, and CMC does not want to upset the sale to the Soo. The United States and the ICC say that this makes the case moot. Paragraph 4 of the court’s order does not affect anything any more, and a court ought not to offer advisory opinions on what would have been. Paragraph 4 is not without moment, however. CMC believes that this paragraph, the equivalent of a declaratory judgment that the Soo’s bid satisfied all constitutional requirements, would block its effort to collect under the Tucker Act in the Claims Court. See United States v. Munsingwear, Inc., 340 U.S. 36, 37-38, 71 S.Ct. 104, 105-06, 95 L.Ed. 36 (1950) (preclusion applies even if review of the earlier judgment was prevented by mootness). CMC wants an obstacle out of the way.
Because the government took the view that paragraph 4 is meaningless, we asked at oral argument whether it would accept an order under Munsingwear, 340 U.S. at 39-41, 71 S.Ct. at 106-07, vacating paragraph 4 as moot. See also Department of the Treasury v. Galioto, — U.S.-, 106 S.Ct. 2683, 91 L.Ed.2d 459 (1986). The United States objected, for it apparently wants to use paragraph 4 to obtain issue preclusion (collateral estoppel) in the Tucker Act litigation. The continuing dispute means, however, that the case is not moot. The parties are embroiled in controversy about whether the Soo’s price was adequate. If they put the question before the Claims Court, that court would not find the matter moot. Because the underlying dis*322pute, the “case or controversy,” determines whether a case is moot, this case is not. Suppose Congress had authorized CMC to proceed in two stages: first, to obtain a declaratory judgment in Illinois that the price was inadequate; second, to ask the Claims Court to order the payment of money if the inadequacy is a “taking”. No one would say that an appeal from the judgment that the price is adequate is “moot” because the court in Illinois could not order any additional relief. We have that case. More than $210 million hangs in the balance. The accident that divides the issues in the Tucker Act litigation between the Northern District of Illinois and the Claims Court does not make this issue moot. Paragraph 4 resolved a point of contention in an ongoing dispute. It is no different from a declaratory judgment entered by the district court.
Nonetheless there is a substantial question whether we should review paragraph 4. It is one of comity between courts. The Claims Court has exclusive jurisdiction over Tucker Act cases, as the Northern District of Illinois has exclusive jurisdiction over this, the only case under the Milwaukee Railroad Restructuring Act, 45 U.S.C. §§ 901-22. One issue, whether the Soo’s price was adequate, is common to both cases. The district court had to address the question to decide who got the Milwaukee’s rail properties and at what price; the Claims Court may need to resolve it (if principles of preclusion do not bar the way) to decide whether to order the United States to compensate CMC for the difference between the Soo’s bid and the CNW’s. Each court is authorized to decide. But the reason the district court was authorized to do so has disappeared. The sale to the Soo may not be undone. Its price is fixed. The question of comity is whether, now that the district court’s objective in answering the question has been realized, further litigation that would affect the takings claim should be in the Claims Court and the Federal Circuit.
It is not clear, however, how we might bring that about. The statute authorizing transfers of appeals, 28 U.S.C. § 1631, is limited to appeals that do not invoke the jurisdiction of the court in which they were filed. An appeal from paragraph 4 is within our jurisdiction and may not be transferred to the Federal Circuit. We have explained why it is inappropriate to vacate paragraph 4 as moot. Still a third possibility is to dismiss CMC’s appeal because paragraph 4 is no longer relevant to the disposition of any issue concerning Soo’s right to own and operate the rail properties. But under Munsingwear an order that left paragraph 4 standing would preclude any collateral challenge in the Claims Court. See 340 U.S. at 37-38; see also Becker v. Contoure Laboratories, Inc., 279 U.S. 388, 391, 49 S.Ct. 356, 357, 73 L.Ed. 752 (1929) (a decision by a state court precludes relitigation of issues that affect a federal case even though the federal court has exclusive jurisdiction of patent issues); Murphy v. Gallagher, 761 F.2d 878, 885-86 (2d Cir.1985) (same in securities law); Restatement (Second) of Judgments § 28(3), comment d at 279, and reporter’s note at 287-88 (when the first court was afforded a full litigation of an issue, and the parties had the incentive to present their claims fully, the decision binds in litigation before a specialized court of exclusive jurisdiction). Cf. Marrese v. American Academy of Orthopaedic Surgeons, 726 F.2d 1150, 1152-54 (7th Cir.1984) (en banc) (as a matter of federal law of claim preclusion, a court of exclusive jurisdiction is bound by an earlier judgment of another court on the same claim for relief), reversed, 470 U.S. 373, 105 S.Ct. 1327, 84 L.Ed.2d 274 (1985) (holding that state law of preclusion applies when the first judgment is rendered by a state court). The Federal Circuit, as successor to the Court of Claims, takes this view of preclusion. E.g., General Dynamics Corp. v. United States, 558 F.2d 985, 993-94 (Ct.Cl.1977); Gary Aircraft Corp., 226 Ct.Cl. 568, 573 (1981) (dictum); cf. McDonnell Douglas Corp. v. United States, 754 F.2d 365, 368 (Fed.Cir.1985).
Still another possibility is to vacate paragraph 4 as a matter of comity between courts, leaving all issues to the Claims *323Court. There is some support for this approach in cases involving state-federal relations. E.g., Colorado River Water Conservation District v. United States, 424 U.S. 800, 96 S.Ct. 1236, 47 L.Ed.2d 483 (1976) (federal court should abstain when a state court has broad jurisdiction over water rights issues and should consider all claims to water in a single case); Samuels v. Mackell, 401 U.S. 66, 91 S.Ct. 764, 27 L.Ed.2d 688 (1971) (federal court should not issue a declaratory judgment that will conclude central issues in ongoing state criminal prosecution). See also David L. Shapiro, Jurisdiction and Discretion, 60 N.Y.U. L.Rev. 543 (1985) (discussing the extent to which courts have discretion not to resolve cases within their jurisdiction). But these abstention eases all involved federal litigation that would interfere with ongoing state litigation; the Supreme Court has never suggested that a federal court ought to vacate an existing judgment in order to permit relitigation of a settled claim elsewhere. It has held, instead, that a federal court may issue a declaratory judgment whose principal if not sole purpose is to preclude later state litigation. See Steffel v. Thompson, 415 U.S. 452, 94 S.Ct. 1209, 39 L.Ed.2d 505 (1974). (The case did not decide, however, whether state courts were bound to respect the federal judgment, and the Justices took divergent views in separate opinions.) More recently the Court held it appropriate to issue a declaratory judgment designed to preempt issues of law that will come before what would otherwise be a tribunal of exclusive jurisdiction. See United Auto Workers v. Brock, — U.S.-, 106 S.Ct. 2523, 2529-31, 91 L.Ed.2d 228 (1986). Unless the second tribunal has immunity from such interference, as in Green v. Mansour, — U.S. -, 106 S.Ct. 423, 88 L.Ed.2d 371 (1986), it is appropriate for the first court properly seized of jurisdiction to decide the complete question.
Although Steffel and UAWv. Brock show that one court may issue declaratory relief that binds a second tribunal of exclusive jurisdiction, they do not hold that it must. Each court must decide whether the potential interference with the business of another is prudent. Kerotest Manufacturing Co. v. C-O-Two Fire Equipment Co., 342 U.S. 180, 72 S.Ct. 219, 96 L.Ed. 200 (1952); Brillhart v. Excess Insurance Co., 316 U.S. 491, 62 S.Ct. 1173, 86 L.Ed. 1620 (1942). It is prudent here. The Claims Court does not possess unique expertise in valuing property or resolving takings claims. The ICC also values railroads (as it did here), and district courts regularly value property for takings purposes in condemnation suits by the government under 28 U.S.C. § 1358 and 40 U.S.C. §§ 257-258f. The exclusivity of the Claims Court is designed to put in one forum all suits seeking money when the United States maintains that it owes nothing; the purposes of that exclusivity would not be undercut by a decision that paragraph 4 of the district court’s judgment prevents recovery against the United States.
Several things counsel immediate resolution of the dispute.. One is that the valuation was fully litigated before the ICC, which issued mountainous valuation studies and two full opinions, and the district court. The parties invested their all; with $210 million on the table, no one held back. The parties are entitled to the benefit of the judgment for which they fought, not least because the transaction would not have occurred if the district court had found the Soo’s bid insufficient. The disputed finding was an essential ingredient of the transaction; had the district court made a different finding, the CNW would own the rail assets and there would be no claim of taking. The reliance interests of the parties — including the United States, which through the ICC could have blocked this deal if it thought the decision exposed the Treasury to a $210 million award — are substantial. Only the gravest reasons would lead us to vacate the district court’s judgment and send the parties onward to a fresh litigation. There may be reasons on the merits why the judgment must be vacated (though we hold otherwise below), but it is on the merits that the judgment should stand or fall.
*324The parties have briefed two issues: whether the district judge was right in declaring the Soo’s bid sufficient, and whether the accuracy of the decision matters. Either ground supplies a sufficient reason to leave paragraph 4 undisturbed. It is tempting to address the merits and leave the rest to the Claims Court. If the Soo’s bid is adequate, the case is over; if not, the Claims Court may decide whether compensation is due. Yet a resolution of this issue would prove to be advisory if the accuracy of the district court’s decision does not matter. Courts should address issues in a sequence that avoids the resolution of surplus issues. Cf. Adams v. Indiana, 795 F.2d 27, 30 (7th Cir.1986). The first issue as a logical matter is whether an error in paragraph 4 could be a taking. The parties have briefed this issue fully. We recognize that consideration of this issue may seem to trench even more directly on the Claims Court’s jurisdiction than would consideration of the accuracy of the district court’s decision. But the Claims Court has exclusive jurisdiction of cases, not of issues (as the district court’s jurisdiction of condemnation cases shows), and under the rationale of Steffel and UAW v. Brock we are free to dispose of this issue even though it will conclude the outcome of the Tucker Act litigation. It is best to bring this controversy to a conclusion, if that is within our power. It is.
C
We shall assume that the district judge erred in believing that $570 million would have been “just compensation”, had the United States condemned the Milwaukee’s rail properties. See United States v. 50 Acres of Land, 469 U.S. 24, 105 S.Ct. 451, 83 L.Ed.2d 376 (1984) (the price that a willing buyer would pay a willing seller is the constitutional measure of just compensation); Chicago & North Western Transportation Co. v. United States, 678 F.2d 665 (7th Cir.1982) (the fair market value of. a line of railroad is the line’s scrap value when there is no higher bid); In re Valuation Proceeding, 445 F.Supp. 994, 1008-29 (Spec.Rail Ct.1977) (Friendly, J.) (rejecting the argument that valuation should be at scrap value when a sale at a higher price for a rail use is possible); cf. In re Chicago, Rock Island & Pacific R.R., 794 F.2d 1182, 1185 (7th Cir.1986) (actual bids, not hypothetical calculations, are the best indi-cia of value). CMC thinks that error is the same as a taking.
There are two preliminary obstacles. The government contends that judges do not “take” property; a “taking” is a deliberate act by the political branches, and judges have no authority to act for the political branches. The government also insists that owners of railroads accept the risk that the value of their property will be reduced by regulation, which is all that happened here. Neither of these arguments is complete.
It is a matter of indifference to CMC whether a judge or the ICC determines that the Soo rather than the CNW prevails. The Takings Clause of the fifth amendment, which states “nor shall private property be taken for public use, without just compensation”, addresses the United States as an entity. If the United States adopts a policy that takes property, it should not matter that a judge is the instrument of the policy. The ICC in the first instance found $570 million sufficient, and had the ICC been authorized to sell the line this would have exposed the United States to a claim; the court’s approval of the ICC’s decision did not make the sale any the less a transfer at the behest of the United States at a price it found sufficient. The role of the agency is not indispensable, either. Suppose the Milwaukee Railroad Restructuring Act, 45 U.S.C. §§ 901-22, under which the court acted, had authorized the judge to donate the property of the Milwaukee to a charity of his choice. The United States could not use the judge’s role to escape paying compensation. Cf. Regional Rail Reorganization Act Cases, 419 U.S. 102, 122-25, 137-39, 154-56, 95 S.Ct. 335, 348-49, 364-65, 42 L.Ed.2d 320 (1974) (diminution of value of a railroad while a court reorganizes and values it could be a taking).
*325The Restructuring Act does not authorize a judge to give the railroad away, but it does grant the judge control over sales of the Milwaukee’s lines, 45 U.S.C. § 904; instead of holding an auction, as in an ordinary bankruptcy case, the judge may choose the buyer. See In re Chicago, Milwaukee, St. Paul & Pacific R.R., 756 F.2d 508 (7th Cir.1985). There is no big difference between giving away half of the Milwaukee’s property for nothing and selling all of it for half of its value. More, former Bankruptcy Rule 8-703(a)(6), which applies to this reorganization, provides that once a sale to a good faith purchaser has taken place it may not be undone on appeal. See In re Chicago, Milwaukee, St. Paul & Pacific R.R., 641 F.2d 482, 484-86 (7th Cir.1981). See also 11 U.S.C. § 363(m). This rule increases the chance of error in the process as a whole by making it harder for an appellate court to review the district court’s decisions; sometimes the appellate court will act quickly (without adequate information and contemplation), and sometimes, as here, it will not act at all. The political branches therefore share with the judicial branch responsibility for what may have been a mistake that diminished the value of the Milwaukee’s estate. If an incorrect valuation and final approval by the ICC would be a taking by the United States, the same mistake by a court exercising administrative powers is no less a taking.
The argument that because railroads are heavily regulated their owners must put up with diminutions in value fares no better. Regulation may diminish the value of property without “taking” it. Ruckelshaus v. Monsanto Co., 467 U.S. 986, 104 S.Ct. 2862, 81 L.Ed.2d 815 (1984); Penn Central Transportation Co. v. New York City, 438 U.S. 104, 98 S.Ct. 2646, 57 L.Ed.2d 631 (1978); cf. McDonald, Sommer & Frates v. Yolo County, — U.S.-, 106 S.Ct. 2561, 2566-67, 91 L.Ed.2d 285 (1986). Railroads have long been subject to regulation that diminishes their profitability. But it does not follow that diminution by all means is permitted. If Congress condemned a railroad station to serve as a post office, this would be a taking notwithstanding the web of regulation surrounding the railroad, notwithstanding that Congress could have diminished the value of the entire railroad in some other way by an amount equal to the value of the station. Cf. Loretto v. Teleprompter Manhattan CATV Corp., 458 U.S. 419, 102 S.Ct. 3164, 73 L.Ed.2d 868 (1982) (although rent control may diminish the value of rental property, a rule requiring lessors to dedicate part of their property to a particular use is still a taking).
Whether a diminution in value is a taking depends at least in part on whether the government’s regulatory objectives entail that diminution. We assume that if the ICC had found CNW’s bid unacceptable because CNW’s ownership of the lines would disrupt transportation or create excessive market power, this decision, although very costly to CMC, would not have “taken” its property. But the ICC did not disqualify CNW. The court’s decision therefore was not compelled by the regulatory system. Consider the point from a different angle. The system of regulation governing the railroads may explain why Soo bid only $570 million (and CNW only $780 million) rather than some higher figure such as $1 billion. The difference between $780 million and $1 billion — the value the Milwaukee’s rail assets might have if free from regulation — is not a taking. See In re Valuation Proceeding, 445 F.Supp. at 1022 & n. 35. The bids that were before the district court already reflected the cost to CMC of federal and state regulation. The diminution in the bids did not immunize a further reduction in the name of national defense, trade with Canada, or other objectives of the court’s choosing.
So when does error “take” property? Suppose agents of the FBI, while chasing a kidnapper, demolish someone’s car, or suppose a postal van runs over a child’s tricycle. Do these accidents “take” the car and tricycle? Certainly they are casualties of the operation of government. Yet despite the contention that all torts by the government are takings, see Richard A. Epstein, *326Takings: Private Property and the Power of Eminent Domain 35-56 (1985), the Supreme Court has distinguished the two. E.g., Keokuk & Hamilton Bridge Co. v. United States, 260 U.S. 125, 43 S.Ct. 37, 67 L.Ed. 165 (1922) (damage to a bridge caused by the government’s blasting is not a taking); YMCA v. United States, 395 U.S. 85, 89 S.Ct. 1511, 23 L.Ed.2d 117 (1969) (damage to a building caused by using it as a military command post during an insurrection in Panama is not a taking). Accidental, unintended injuries inflicted by governmental actors are treated as torts, not takings. And torts are compensable only to the extent the Federal Tort Claims Act permits. The Court has never treated limitations on liability in tort as mere pleading obstacles, to be surmounted by shifting ground to the Tucker Act. See United States v. James, — U.S. -, 106 S.Ct. 3116, 92 L.Ed.2d 483 (1986); Kosak v. United States, 465 U.S. 848, 104 S.Ct. 1519, 79 L.Ed.2d 860 (1984); Laird v. Nelms, 406 U.S. 797, 92 S.Ct. 1899, 32 L.Ed.2d 499 (1972).
The question is whether the program, properly carried out, would be permissible. For example, if planes owned by the government regularly overfly a farm, causing the chickens to kill themselves in fright, this planned operation may “take” an easement across the farm. United States v. Causby, 328 U.S. 256, 66 S.Ct. 1062, 90 L.Ed. 1206 (1946). But if a stray military plane crashes into a chicken coop, killing an equal number of fowl, this is a tort rather than a taking, and compensation depends on the extent to which the Tort Claims Act provides relief. Similarly, if in order to go into the dairy business the government condemns a herd of cattle, that is a compensable taking; if the government condemns the herd to stamp out an infectious disease, that is not a taking (both because the disease is the principal cause of the loss and because there are reciprocal benefits in the program of disease control), see Miller v. Schoene, 276 U.S. 272, 48 S.Ct. 246, 72 L.Ed. 568 (1928); and if the government kills a given animal mistakenly thinking it diseased, that is a tort rather than a taking. Had the government carried out the program accurately and distinguished the healthy from the diseased, there would have been no taking; a foul-up in the implementation of the program does not require compensation outside the scope of the Federal Tort Claims Act. See Langford v. United States, 101 U.S. 341, 344-45, 11 Otto 341, 25 L.Ed. 1010 (1879), holding that when agents of the United States wrongly believe that the government owns some land, and occupy it under a claim of right, the occupation is a noncompensable tort rather than a taking. Authorized acts of the government may be takings, the Court concluded, but unauthorized or mistaken ones are torts for which the officer alone is answerable (unless immune). Compare United States v. Great Falls Manufacturing Co., 112 U.S. 645, 656-58, 5 S.Ct. 306, 310-12, 28 L.Ed. 846 (1884).
So too with unnecessary searches and seizures (those supported by probable cause but that turn up no evidence), YMCA, 395 U.S. at 92, 89 S.Ct. at 1515 (dictum), cf. Baker v. McCollan, 443 U.S. 137, 99 S.Ct. 2689, 61 L.Ed.2d 433 (1979) (arresting the wrong person, though with probable cause, does not violate the Constitution), with unfortunate criminal prosecutions (those in which the charge was warranted but the jury acquits or a conviction is reversed), and with the long list of other governmental acts that can cause deep injury if committed in error. None of these costly errors is a taking, and by parallel reasoning a judicial error in the implementation of a program is not a taking, if a correct implementation of the program would not be a taking.
A parallel line of cases dealing with the Due Process Clause comes to the same conclusion. The Takings Clause does not apply directly to the states, see Barron v. Baltimore, 32 U.S. (7 Pet.) 243, 8 L.Ed. 672 (1833), but the Supreme Court has used the Due Process Clause of the fourteenth amendment to supply a similar set of rules, see Chicago, Burlington & Quincy R.R. v. Chicago, 166 U.S. 226, 233-41,17 S.Ct. 581, 583-86, 41 L.Ed. 979 (1897); Pennsylvania *327Coal Co. v. Mahon, 260 U.S. 393, 43 S.Ct. 158, 67 L.Ed. 322 (1922). The Due Process Clause forbids the state to “deprive” a person of “property” save with certain process. Process sometimes misfires. The state may err in the administration of its programs, misunderstand the facts or misconstrue the law. Do these errors violate principles of due process? “[A]n erroneous decision of a state court does not deprive the unsuccessful party of his property without due process of law”. Central Land Co. v. Laidley, 159 U.S. 103, 112, 16 S.Ct. 80, 82, 40 L.Ed. 91 (1895) (collecting cases). See also, e.g., Bishop v. Wood, 426 U.S. 341, 349-50, 96 S.Ct. 2074, 2079-80, 48 L.Ed.2d 684 (1976); Davidson v. New Orleans, 96 U.S. 97, 104-05, 6 Otto 97, 24 L.Ed. 616 (1877); Lynk v. LaPorte Superior Court No. 2, 789 F.2d 554, 563-64 (7th Cir.1986); Toney-El v. Franzen, 777 F.2d 1224, 1227-28 (7th Cir.1985), cert. denied, — U.S. —, 106 S.Ct. 2909, 90 L.Ed.2d 994 (1986). Mistaken applications of the law are inevitable, but no principle of constitutional law requires compensation for every mistake. Daniels v. Williams, — U.S. -, 106 S.Ct. 662, 665, 88 L.Ed.2d 662 (1986); see also Carson v. Block, 790 F.2d 562, 565 (7th Cir.1986) (collecting cases holding that the Constitution does not require states to follow their own laws).
There are other illustrations, but the point has been made. The government is not absolutely liable for mistaken implementation of a law that does not otherwise take property. Were it otherwise, the Treasury would guarantee the accuracy of all administrative and judicial decisions. That was certainly not the understanding in 1791, when the Takings Clause was added to the Constitution; to the contrary, there was consensus that the sovereign could not be sued for errors save with its consent. We do not think that the Takings Clause, unbeknownst to all, established a principle of governmental liability for mistakes in the implementation of federal programs. The principle would have substantial effects, too large to escape notice in 1791 (and since). Many decisions are effectively unreviewable or cause damage before they may be reviewed. When the FTC announces its opposition to a merger, this may lead the merging parties to abandon their plans; they may not turn around and collect damages on the ground that the FTC holds mistaken views about the antitrust laws. If the Antitrust Division should get a preliminary injunction that blocks the merger, the interim order may doom the plans even if the court of appeals later reverses; here the judge’s error is established, but the error does not expose the United States to a suit under the Tucker Act, because the antitrust laws, when properly carried out, do not take property. (No one has a legitimate claim of entitlement to monopoly profits.) Pretrial detention, the grant of a motion to compel extensive discovery in a securities case, these and many more may inflict great injury, and erroneous decisions by district judges may escape effective review.
A rule requiring the Nation to bear the costs would not reduce the number of errors or the incidence of their costs. It would, however, produce much litigation as people tried to tap the Treasury by showing judicial error. A rule of cost shifting via the Takings Clause might even increase the number of judicial errors. People would fear judicial error less and so would do less to prevent it. Knowing that they could turn to the Treasury for relief, people would be tempted to skimp on developing the record and briefing the case to get it right the first time. One reason for res judicata and other rules of preclusion is to induce parties to concentrate their energies on a single full and fair contest in which error will be held to a minimum. So, too, a rule foreclosing resort to takings litigation may improve the quality of the initial litigation. A principle that the adequacy of the rule of decision, and not its implementation in a given case, controls the constitutional question, will do the trick.
The Milwaukee Railroad Restructuring Act neither requires nor authorizes a judge to transfer property for compensation that is less than the constitutional min*328imum. The Restructuring Act is thus unlike the statute at issue in the Regional Rail Reorganization Act Cases, which not only required a transfer without a prior finding of adequate value, see 419 U.S. at 155, 95 S.Ct. at 364, but also could have caused a taking by the erosion of value during the judicial proceedings, id. at 122-25, 95 S.Ct. at 348-49. The Restructuring Act makes adequate compensation a condition of any sale. The ICC believed that it should not approve any bid less than the fair market value of the Milwaukee’s rail assets. It found that a bid of $570 million was fair market value. The district court, too, proceeded as if the rule of decision were: “Choose from among the bidders that have offered at least the fair market value of the Milwaukee’s' rail assets.” The court determined that the Soo’s bid matched the fair market value. If the court was right, then the sale to the Soo did not take CMC’s property. For the reasons we have given, a judicial error in answering a constitutionally appropriate question does not subject the United States to a claim of taking. We therefore have no reason to vacate paragraph 4 of the district court’s order.
It is unsettling to contemplate the prospect of $210 million gone without redress— without even full appellate review of the decision to sell to the Soo. But we cannot undo the sale now, and to treat judicial error as a “taking” has no support in the history of the Constitution or the first 200 years of its interpretation. The size of an error (if there was one) does not justify constitutional innovation; modern exigency does not authorize courts to change the rules established at the beginning. The $210 million in one case is not different in principle from 210 cases, each of which entails $1 million worth of useless discovery under a court’s directive, or a million bicycles crushed by postal trucks.
We can imagine cases that test the limits of our approach. Suppose the disparity between market value and the price approved by a judge were so great that the purchaser would not even count as a “good faith purchaser for value” under the Uniform Commercial Code (for example, the person who buys a gold watch for $20 from a seller displaying an ample supply on the inside of a trench coat). Then a court might say that the sale could be undone under § 363(m), or that the disparity of value is so great that the court must have been dissembling in saying that it was trying to apply the constitutional standard. None of these hard cases is before us. The ICC found the value sufficient, and $570 million — a price earnings multiple exceeding that of other operating railroads — is not so shockingly deficient to call a limiting rule into play. This was a real sale, not a purchase from a fence or disguised judicial charity.
II
A
The purchaser of a line of railroad usually finds it encumbered with conditions designed to protect the jobs (or at least the income) of the line’s employees. The ICC recommended that the sale of the Milwaukee’s rail properties be accompanied by rules referred to as the New York Dock conditions, after New York Dock Ry.—Control, 360 I.C.C. 60 (1979), aff’d under the name New York Dock Ry. v. United States, 609 F.2d 83 (2d Cir.1979). It is not important to recount just what these are. See Simmons v. ICC, 760 F.2d 126, 128-29 (7th Cir.1985), cert. denied, — U.S. -, 106 S.Ct. 791, 88 L.Ed.2d 769 (1986), for a compact description. The Soo, which projected eliminating 670 jobs entirely by attrition and early retirement, agreed to be bound by them, and the ICC said that they were enough, rejecting the argument of the Railway Labor Executives’ Association (RLEA). The RLEA maintained that not only the Soo’s new employees but also employees of other railroads affected by new competition from the Soo should be protected; the ICC disagreed. The RLEA also wanted protection for employees on furlough at the time of the transfer; the ICC decided that only employees active at the *329time of the sale or recalled to work thereafter should be protected.
The district court was authorized by the Restructuring Act, 45 U.S.C. § 904(b)(1), to decide for itself what labor protective conditions should be imposed. The statute provides: “In authorizing any sale or transfer, the court shall provide a fair arrangement at least as protective of the interests of employees as that required under section 11347 of title 49 of the United States Code.” The CNW proposed that the district court use its authority to implement a different set of conditions, called the “Appendix B” conditions. The Trustee supported these conditions for CNW’s bid. Although the district court did not approve CNW’s bid, it chose the “Appendix B” conditions, while lawyers for the Soo looked on in silence. The principal difference between the “Appendix B” conditions and the New York Dock conditions is that although New York Dock requires the railroad to notify employees of impending changes 90 days in advance, and to negotiate with the unions concerning their consequences until agreement is reached, “Appendix B” shortens the time to ten days and allows the railroad to make the changes whether or not the unions consent. “Appendix B” also requires that any employee injured by this accelerated schedule be made whole. See In re Chicago, Milwaukee, St. Paul & Pacific R.R., 658 F.2d 1149, 1151-52 (7th Cir.1981) (Protective Conditions), cert. denied, 455 U.S. 1000, 102 S.Ct. 1632, 71 L.Ed.2d 867 (1982), for a description of the difference. (The Soo represents that this is the only difference between “Appendix B” and New York Dock conditions, a position to which we expect the Soo to adhere.) The order of February 1985 transferring the rail operations to the Soo contained a freeze of at least 90 days on all operational changes, so the full set of protective conditions is “Appendix B” plus the initial freeze.
The RLEA protests that the district court moved in the wrong direction; it should have made the conditions more rather than less favorable to labor. The “Appendix B” conditions, the RLEA insists, are illegal because less “protective of the interests of employees” than the minimum under 49 U.S.C. § 11347. The RLEA also challenges two related decisions. After the Soo bought the line, the RLEA asked the ICC to disapprove the Milwaukee’s final plan of reorganization, on the ground that the Milwaukee had not adequately provided for workers who agreed to defer increased wages during the reorganization; The RLEA wants the Milwaukee to repay the workers 130% of these deferrals. The ICC dismissed the petition, concluding that once the Milwaukee had transferred its rail operations to the Soo, the Milwaukee was no longer a railroad subject to the ICC’s jurisdiction. The RLEA also asked the district court to disapprove the plan of reorganization, but the court confirmed the plan. The “Appendix B” conditions and the ICC’s dismissal of the petition are before us for decision; the objections to the confirmation of the plan are not.
B
The RLEA’s request that we order the Soo to assume the New York Dock conditions is moot. Former Bankruptcy Rule 8-703(a)(6), which applies to the last stages of this reorganization under § 77 of the Bankruptcy Act of 1898, 11 U.S.C. § 205 (1976), provided:
Unless an order approving a sale of property ... is stayed pending appeal, the sale to a good faith purchaser ... shall not be affected by the reversal or modification of such order on appeal, whether or not the purchaser ... knows of the pendency of the appeal.
The Bankruptcy Code of 1978 codifies this rule in 11 U.S.C. § 363(m), which uses similar language. Several parties, including the RLEA, asked us to stay the sale to the Soo; the request was refused, and the sale was closed on February 20, 1985. The imposition of the New York Dock conditions would “affect” the sale just as surely as a sudden increase in the price would affect it, and former Rule 8-703(a)(6) therefore forbids us to grant the relief the *330RLEA proposes. See also In re Chicago, Milwaukee, St. Paul & Pacific R.R., 641 F.2d 482, 484-86 (7th Cir.1981) (rejecting on the basis of Rule 8-703(a)(6) a challenge to the sale of some of the Milwaukee’s property).
RLEA tries to escape the operation of the Rule by contending that the Soo is not a “good faith purchaser”, but this will not wash. Only fraud, collusion, or knowledge of the illegality of the sale defeat “good faith”. In re Magwood, 785 F.2d 1077, 1081 n. 6 (D.C.Cir.1986); In re Rock Indus. Machinery Corp., 572 F.2d 1195, 1198 (7th Cir.1978). See also In re Sax, 796 F.2d 994, 997-998 (7th Cir.1986) (sale need not be “proper” to be sheltered; it need only be “authorized”). The Rule and § 363(m) enable the bankruptcy court to promise the purchaser that it will get a good title and thus reduce the risk the purchaser must assume. Once the bidders know what they will get, they can bid its full market value; if the sale were subject to subsequent amendments, the bidders would reduce their valuations by the anticipated cost of accommodating these changes. Sax, slip op. 6. A reduction in the risk leads to higher bids, to the benefit of the bankrupt estate.
That there may be a fair debate about the order of sale does not undercut good faith. The statute and the rule are concerned only with incorrect orders, those reversed in fact. (Correct orders will be affirmed, so there will be no change in the conditions of the sale with or without the Rule and § 363(m).) Orders that are reversed surely were subject to fair debate. Yet both Rule and statute specify that knowledge of the pending appeal does not prevent the sale from being final. This means that the buyer’s knowledge that the court has resolved a debatable issue — even resolved it incorrectly — does not prevent the buyer from taking the property in good faith. Our decision in In re EDC Holding Co., 676 F.2d 945 (7th Cir.1982), one of the few cases ever to find “bad faith” under a statute similar to Rule 8-703(a)(6), relied not only on the party’s knowledge of illegality but also on an appearance that the party seeking shelter from the rule had used it to improve its position compared with other creditors. Nothing of the sort took place here. A party does not “know” to be illegal whatever turns out to be so. The knowledge component of this test is more like the sort of knowledge that leads to criminal liability (people “know” that wages are taxable even if they have deceived themselves into thinking they are not) or the sort of specious position that leads to the imposition of sanctions in litigation.
The Soo is a third party that made a bona fide bid, and the Soo did not know that the “Appendix B” conditions were illegal. To the contrary, Protective Conditions held that the district court possessed the discretion to use the “Appendix B” conditions in connection with the sale of one line of the Milwaukee’s track, and In re Chicago, Milwaukee, St. Paul & Pacific R.R., 756 F.2d 508, 515-16 (7th Cir.1985) (Soo Intervention), held that the portions of the Restructuring Act allowing the sale of any line of track also allow the sale of the whole rail system of the Milwaukee. Protective Conditions and Soo Intervention together give the Soo a strong argument that the “Appendix B” conditions may be applied to the sale of the whole railroad. Whether this argument is strong enough we need not say; all we decide is that it precludes a conclusion that the Soo is not a “good faith purchaser” under the Rule.
Although it might be said that the Soo and the Trustee closed the sale in great haste, just hours after this court denied the RLEA’s application for a stay, that does not subtract from the Soo’s good faith. That term of art has nothing to do with the speed at which a transaction proceeds — at least not if the parties wait long enough to allow the court of appeals to act on the application for a stay. Once this court had refused to stay the sale, it made no difference whether the deal closed the next hour or the next month. And although RLEA also argues that the district court’s order was so erroneous that the court lost its “jurisdiction,” Sax disposes of this position. *331The district court had jurisdiction to determine its authority to do what it did, which shelters the purchaser at the sale.
There has been some debate among the parties about whether labor conditions imposed on the Milwaukee in 1936 survived the sale to the Soo. This question was not decided by the district court and is not properly before us. The motion to supplement the record is denied.
C
In March 1985 the Trustee petitioned the ICC under § 77(d) of the old Bankruptcy Act to clarify its role in reviewing the Milwaukee’s final plan of reorganization into a real estate corporation (CMC). The Trustee argued that approval was unnecessary because the Commission’s earlier decisions had approved every significant feature of the plan of reorganization. The RLEA opposed the petition. 'In April 1985 the ICC dismissed the Trustee’s petition on the ground that because the Milwaukee had sold its rail operations to the Soo, it was no longer a railroad and was not within the Commission’s jurisdiction. Fin.Dkt. No. 28640 (Sub-No. 9), — I.C.C.2d — (1985).
The RLEA does not maintain that by April 1985 the Milwaukee was still a railroad, and § 77, with its approval provision, is limited to railroads. RLEA insists, however, that the Milwaukee had to be treated as if it were a railroad because the sale of the rail assets in February was illegal. This is so, RLEA says, because rail assets of a bankrupt railroad may be sold only as part of the plan of reorganization under § 77 of the 1898 Bankruptcy Act. The Soo took under the Milwaukee Restructuring Act, not the Bankruptcy Act, so according to the RLEA the ICC still needed to approve a reorganization under the 1898 Act. This requirement of approval becomes the hook that the RLEA proposes to use to require CMC to pay additional sums to its former workers to make up for what they deferred or lost during the reorganization and sale of the railroad, perhaps even to get the New York Dock conditions.
To the extent the RLEA wants to use the process of approving and confirming the plan to change the conditions under which the Soo took the rail assets, the claim is moot for reasons we discussed above. To the extent the RLEA wants the ICC to impose conditions on CMC as a condition of approval of the plan, it loses because we held in Soo Intervention, 756 F.2d at 515-16, that the provisions of the Restructuring Act could be used to sell the whole railroad. We are not disposed to reexamine Soo Intervention. See also In re Chicago, Milwaukee, St. Paul & Pacific R.R., 701 F.2d 604, 607 (7th Cir.) (emphasizing that a purpose of the Restructuring Act was to allow the court to sell the Milwaukee’s rail assets quickly), cert. dismissed, 463 U.S. 1233, 103 S.Ct. 3574, 77 L.Ed.2d 1416 (1983). The Restructuring Act required the district court to get the ICC’s views, just as § 77(d) of the Bankruptcy Act called for the ICC’s views on a plan of reorganization. The Restructuring Act changed the rules, however, by allowing the district court to make its own decision in the end. Here the district court took the ICC’s advice and sold the lines to the Soo. The district court had the authority to turn the Milwaukee into a real estate company on its own. The RLEA treats the Restructuring Act as a second hoop through which everyone had to jump, but Congress meant it to streamline the § 77 procedures. Compliance with the Restructuring Act is sufficient to transfer the rail properties; the parties need not obtain a redundant clearance under § 77.
There is no doubt that by April 1985 the Milwaukee was not operating a railroad. Section 77(d) refers to “the debtor”, and this might be the basis of a contention that the Milwaukee remained subject to § 77(d) even though it was no longer a railroad, needing the ICC’s permission to get out of bankruptcy. The ICC reads the statute otherwise, however, and its interpretation is entitled to respect. Chemical Manufacturers Ass’n v. Natural Resources Defense Council, Inc., 470 U.S. 116, 105 S.Ct. 1102, 1108, 84 L.Ed.2d 90 (1985); Watkins *332v. Blinzinger, 789 F.2d 474, 478 (7th Cir.1986). The ICC has said its piece on this reorganization. Section 77(d) of the Bankruptcy Act was designed to ensure that the ICC could control railroad reorganizations. The ICC dominated this one, prevailing on every issue except the New York Dock conditions. It had nothing more to say, and an interpretation that would apply § 77(d) to firms that are not railroads any longer, despite the ICC’s protestation that it does not want to review anything, would not have any plausible function.
D
The district court confirmed the Milwaukee’s plan of reorganization on July 12, 1985. The RLEA and several other parties have filed appeals from the order confirming the plan. Although many of the RLEA’s arguments about the order of sale and the ICC’s refusal to review the plan go to the merits of that plan, the RLEA’s appeal from the confirmation is not before us. It and all other appeals have been consolidated for separate disposition. We have disposed of RLEA’s arguments on jurisdictional grounds. Nothing we say here bears on the merits of any appeal from the order confirming the plan, and we mean to leave no clues about what our views might be.
Ill
A
The rail assets of the Milwaukee included rights to run trains between Bedford and New Albany, Indiana, over the track of the Seaboard System. These trackage rights enabled the Milwaukee to serve Louisville, Kentucky, and in the hands of the Soo (and the Canadian Pacific) apparently will offer formidable competition to the Seaboard, which otherwise might pick up the traffic in Chicago. The district court transferred these trackage rights to the Soo over the Seaboard’s opposition, and the Seaboard asks us to annul the transfer on the ground that the trackage rights agreement contains a clause allowing assignment only if the Seaboard concurs — which it certainly does not.
The trackage rights arose out of the acquisition of the Monon R.R. by the Louisville & Nashville R.R., one of the Seaboard’s predecessors. The acquisition would have given the L & N, which operated a single-line route between Chicago and Louisville, another. The ICC feared that this would reduce competition, and it ordered a protesting L & N to grant the Milwaukee trackage rights from Bedford to New Albany (which is across the Ohio River from Louisville). Louisville & Nashville R.R.—Merger—Monon R.R., 338 I.C.C. 134, 145-56, 168-69, 185 (1970). The Milwaukee had a line from Chicago to Bed-ford; with the trackage rights, it could offer single-line service from Chicago to Louisville in competition with the L & N. The ICC stated that it would not approve the acquisition unless the L & N granted trackage rights; the L & N consented and completed the acquisition.
The L & N and the Milwaukee had trouble agreeing on the terms of the trackage rights, so in 1973 the ICC dictated them. Chicago, Milwaukee, St. Paul & Pacific R.R.—Trackage Rights, 342 I.C.C. 578, aff’d under the name Louisville & Nashville R.R. v. United States, 369 F.Supp. 621 (W.D.Ky.) (three-judge court), aff’d mem., 414 U.S. 1105, 94 S.Ct. 832, 38 L.Ed.2d 733 (1973). The L & N wanted a time limit on the trackage rights, a prohibition against their sale or assignment by the Milwaukee, and automatic termination in the event of the Milwaukee’s merger, consolidation, or bankruptcy. The ICC declined to accept these proposals. It explained in part, 342 I.C.C. at 595, 600-01:
[I]t is difficult to perceive how any change may come about in Milwaukee’s status which would have such adverse effect on L & N and its interest to require summary termination of trackage rights operations____ [T]he Commission’s authorization must first be obtained before Milwaukee may merge or consolidate, or fall under common control, with any other carrier, and L & N *333would have ample opportunity in appropriate proceedings before the Commission to oppose acquisition of control by competing railroads over the Milwaukee trackage rights.
Milwaukee is here the instrument of the Commission’s policy, in the public interest, to foster competition among railroads in a particular situation, and the trackage rights operations authorized should be permitted, and indeed required, to continue just so long as that policy is valid and serves the public interest. On petition of the parties, or either party to the agreement, ... the public interest may be reconsidered or reexamined at any time, and the appropriate action taken to modify the terms of the agreement or to require termination thereof.
L & N and the Milwaukee then signed a trackage rights agreement. Paragraph 14 reserves the ICC’s jurisdiction to alter or terminate the agreement. Paragraph 15, which was added to the agreement after the ICC’s opinion, provides:
This agreement and all rights granted hereunder to Milwaukee Company may not be transferred or assigned, in whole or in part, unless Louisville Company, its successors or assigns, shall consent in writing thereto. Any such attempted transfer or assignment, in whole or in part, shall be void and of no effect.
Paragraph 13 contains a clause providing for arbitration of “any difference or dispute” concerning the terms of the agreement or operations using the trackage rights.
Before the ICC considered the three proposals to acquire all of the Milwaukee’s rail operations, it published a notice in the Federal Register, 49 Fed.Reg. 12333 (Mar. 29, 1984), that invited interested persons to comment. The Seaboard did not, although it does not deny that it had actual notice of the proceedings before the ICC. The ICC’s orders of September 1984 (approving a sale to the Soo) and January 1985 (approving both the Soo and the CNW) are crammed with descriptions of trackage rights and conditions on transfers. Although they mention the Milwaukee’s trackage rights to New Albany, the orders do not discuss their transfer to the Soo or the CNW. This is no surprise, because no one asked the ICC to consider the propriety of the transfer.
The Seaboard, which was not a party to the reorganization, then asked the court to forbid a transfer of the trackage rights to the Soo or the CNW. There followed some confusing exchanges. At the oral argument held on February 7 to discuss the terms of the transfer, the court recognized that the ICC had not addressed the Bed-ford-New Albany trackage rights and asked the Trustee whether the Asset Purchase Agreement (the complex document establishing the terms of the sale) would extinguish the Seaboard’s claims. The judge asked: “Wouldn’t the Trustee makes [sic] his conveyance without warranties, simply convey whatever rights he has and let the purchaser worry about that?” Counsel for the Trustee answered: “[T]hat is exactly the type of conveyance that is being considered. But the Seaboard is asking this Court to decide the issue.”
The matter was dropped until the next day, when the court read findings and conclusions from the bench. The court stated that the “Trustee is, of course, only going to convey what rights in interest he has”, that the dispute with the Seaboard “will have to be resolved by the acquiring carrier”, and that “[i]f the Seaboard believes they are not transferable, they will have to litigate that or make some kind of arrangement with the Soo.” The full statement, including some exchanges with counsel, appears in Seaboard System R.R. v. Soo Line R.R., No. C-85-1569-A (N.D.Ga. May 28, 1986), slip op. 20-24 n. 8 [Available on WESTLAW, DCTU database]. After hearing this, Seaboard doubtless thought it had won, and on February 14 it intervened to secure the spoils.
It was sadly disappointed, because paragraph 9 of the court’s formal order approving the sale (Order No. 809), entered on February 19, stated that “[t]he assignment and assumption of the trackage agree*334ments ... pursuant to the [Asset Purchase Agreement] will not effect a termination of the Trustee’s rights and interests ..., and those rights and interests are assignable to Soo and [its subsidiary] in accordance with the terms of the [Asset Purchase Agreement], and notwithstanding any provisions in any such contracts, leases or agreements to the contrary.” On seeing this clause counsel for the Seaboard protested that the court was snatching back the permission it had granted the Seaboard on February 8 to continue in another forum its dispute with the Soo about the effect of the transfer. The court eliminated from the draft of paragraph 9 the word “fully” before “assignable” but otherwise rebuffed the Seaboard’s arguments. On the one hand the judge said that Order No. 809 “leave[s] the matter open to subsequent litigation” and on the other remarked that the Order “means they are assignable. I don’t see what prejudice the Seaboard sustains, particularly having been silent when they had an opportunity to raise the question before the ICC.” Then the judge suggested a middle ground, that the Order tells the Seaboard “[t]hey can go to the Commission or they can go to the Soo” to seek relief. Understandably frustrated by all this back- and-forth, the Seaboard requested relief from Order No. 809. On March 25, without explanation, the court entered a minute order denying the Seaboard’s request.
The Seaboard meanwhile filed suit in the Northern District of Georgia, asking for an order compelling the Soo to arbitrate the transfer and for a declaratory judgment that the transfer violates the clause requiring the Seaboard’s consent. The court held, Seaboard v. Soo, supra, that both requests are barred by claim preclusion. Either the reorganization court resolved them adversely to the Seaboard in Order No. 809 or the Seaboard, having intervened, should have sought from the reorganization court its declaratory judgment and order to compel arbitration. Either way, the Georgia court concluded, the Seaboard was not entitled to move its dispute to a new forum and try again.
B
What is before us for review? No. 85-1573 is Seaboard’s petition to review the January 1985 decision of the ICC to the extent it approved the transfer of the trackage rights. The Commission, which did not mention the transfer, contends that the Seaboard has not preserved a challenge yet insists that it indeed approved the transfer. The ICC approved the transfer in the sense that it approved the Soo’s proposal. But whether the Milwaukee had a right to transfer the Bedford-New Albany rights without the Seaboard’s approval was not raised, litigated, or decided. The Commission’s silence does not construe or modify its orders of 1970 or 1973, or the terms of the contract between Milwaukee and the L & N. Yet the Seaboard is not entitled to contest this non-decision. If it had been a party, its challenge to the ICC’s order would be precluded by the rule that a party seeking to raise an issue in court must first raise it before the agency. E.g., United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S.Ct. 67, 69, 97 L.Ed. 54 (1952); Illinois Commerce Commission v. ICC, 776 F.2d 355, 362-63 (D.C.Cir.1985) (Scalia, J.). The Seaboard was not a party, however, and its challenge is barred by the even more fundamental principle that only a party before the Commission may seek judicial review of its orders. See 28 U.S.C. §§ 2342(5) and 2344 (limiting review to a “party aggrieved” by the order); see also Radiofone, Inc. v. FCC, 759 F.2d 936 (D.C.Cir.1985) (Scalia, J.); Simmons v. ICC, 716 F.2d 40, 42-43 (D.C.Cir. 1983). No. 85-1573 must be dismissed for want of jurisdiction.
One court has suggested that despite § 2344 a non-party may obtain judicial review of “agency action [that] is ‘attacked as exceeding the power of the [ICC].’ ” American Trucking Associations, Inc. v. ICC, 673 F.2d 82, 85 n. 4 (5th Cir.1982), cert. denied, 460 U.S. 1022, 103 S.Ct. 1272, 75 L.Ed.2d 493 (1983). This decision cited only two cases before the Administrative Procedure Act that had allowed a non-party to bring an independent suit in district
*335court seeking an injunction. Edward Hines Yellow Pipe Trustees v. United States, 263 U.S. 143, 147-48, 44 S.Ct. 72, 73-74, 68 L.Ed. 216 (1923); Skinner & Eddy Corp. v. United States, 249 U.S. 557, 562-63, 39 S.Ct. 375, 377, 63 L.Ed. 772 (1919). We need not know the health of these cases, see FCC v. ITT World Communications, Inc., 466 U.S. 463, 468-69, 104 S.Ct. 1936, 80 L.Ed.2d 480 (1984), to be confident that non-parties may not seek judicial review in the courts of appeals under the Hobbs Act. A suggestion that non-parties may obtain review of orders that “exceed the power” of the agency is dubious for several reasons. One is that “exceeding the power” of the agency may be a synonym for “wrong,” so that the statute then precludes review only when there is no reason for review anyway. We doubt that this is what Congress meant. Another is that § 2344 is the source of the court’s jurisdiction, and jurisdiction means power to decide. A court may not decide a case just because that would be a good idea; power must be granted, not assumed. In re Chicago, Rock Island & Pacific R.R., 794 F.2d 1182, 1187-88 (7th Cir.1986). If a non-party tried to appeal from a judgment of a district court, we would dismiss the appeal no matter how much in “excess of power” the decision might be; if a party tried to appeal after the time for appeal had expired, the clarity of the district court’s error would not supply the missing power to adjudicate the appeal. Browder v. Director, Department of Corrections, 434 U.S. 257, 98 S.Ct. 556, 54 L.Ed.2d 521 (1978). So it is with review of administrative action. The statute limits review to petitions filed by parties, and that is that. (The Administrative Procedure Act’s provision for judicial review, 5 U.S.C. § 702, on which Seaboard relies, does not supply jurisdiction in any court, let alone authorize direct review in the court of appeals. See Califano v. Sanders, 430 U.S. 99, 104-07, 97 S.Ct. 980, 983-85, 51 L.Ed.2d 192 (1977). Only § 2344 and the associated provisions of the Hobbs Act authorize the type of review Seaboard requests.)
This does not mean that a non-party adversely affected by an agency’s decision has no recourse. If the ICC should purport to invalidate the Seaboard’s certificates of public interest, convenience, and necessity in a footnote to a case involving only the Burlington Northern — without notice to the Seaboard that its interests could be affected — the Seaboard could take either of two approaches. One is to intervene before the ICC as soon as it learns of the Commission’s action, file a petition for rehearing in the agency, and then seek judicial review under § 2344. The other is simply to ignore the order. Then it could not petition for review in the court of appeals, but if the ICC should try to enforce its order, the Seaboard could resist during the enforcement proceeding. This is the same principle that allows a party to ignore a judicial proceeding when the party believes the court lacks jurisdiction, and then to resist enforcement of the court’s order. See Restatement (Second) of Judgments § 81 (1980) (an order entered without notice or jurisdiction does not bind a subsequent tribunal). But this is not a case in which the Commission proceeded without notice to Seaboard; it had actual notice. It is not a case in which the Commission lacked “jurisdiction” to decide; it was authorized to act on the Soo’s request.
C
We have jurisdiction of Seaboard’s appeal from Order No. 809, but the Soo insists that the appeal is moot because we may not alter the terms of the sale. If the district court had decided that the Soo has a clean title to the trackage rights, free of any claims under the contract between the Milwaukee and the L & N, the case would be moot for the reasons given in Sax and Part II.B. of this opinion. But we do not read Order No. 809, in conjunction with the district court’s varying explanations of it, as precluding any claim arising under the contract. The terms of the sale are fixed by the Asset Purchase Agreement as well as Order No. 809, and in another case dealing with the transfer of trackage rights to the Soo we held that the Asset Purchase *336Agreement requires the Soo to arbitrate differences arising out of clauses in track-age rights agreements that restrict the Milwaukee’s ability to transfer its rights. See In re Chicago, Milwaukee, St. Paul & Pacific R.R., 784 F.2d 831 (7th Cir.1986) (Escanaba I); In re Chicago, Milwaukee, St. Paul & Pacific R.R., 789 F.2d 1281 (7th Cir.1986) (Escanaba II).
Many of the Milwaukee’s assets had encumbrances. Property came with liens; some property had been leased to others; the Milwaukee’s leaseholds came with time limits; and trackage rights came with arbitration clauses and restrictions on assignment. See Escanaba I and II. CMC and the ICC agree that the Asset Purchase Agreement and Order No. 809 do not obliterate the encumbrances on the Milwaukee’s property. The Soo got no more than the Milwaukee had.
Order No. 809 does not eliminate the Milwaukee’s liabilities arising out of its rail operations; the Asset Purchase Agreement transfers them to the Soo. A claim arising out of breach of the trackage rights agreement between the Milwaukee and the L & N is a liability transferred to the Soo. The principal effect of paragraph 9 of Order No. 809 is to transfer the operating rights to the Soo in the interim, not to foreclose any other claim.
D
Order No. 816 qualifies Order No. 809 with respect to some trackage rights between Crivitz and Marinette, Wisconsin. The district court ordered the Soo to arbitrate with another railroad concerning the transfer of these trackage rights. See Es-canaba I and II. The Seaboard wanted the trackage rights between Bedford and New Albany carved out as well. We do not think the district court was required to treat the Indiana trackage rights the way it did the rights in Escanaba I and II.
The Seaboard, unlike the Escanaba, did not ask the reorganization court for an order compelling the Soo to arbitrate the disputes growing out of the transfer. More, the contract between the Milwaukee and the Escanaba was a private undertaking. The ICC had expressed no concern about whether the Soo or the Escanaba ran trains over the Crivitz-Marinette line, and it approved the transfer of rights to the Esca-naba. The trackage rights between Bed-ford and New Albany were crammed down the throat of the L & N as a condition of its acquisition of the Monon, and the ICC then fixed the terms of the rights. The Commission expressed the view that the Seaboard should face competition between Chicago and Louisville and that the trackage rights should continue as long as the need for competition continues.
Because the Bedford-New Albany rights exist at the insistence of the ICC, that agency may have primary jurisdiction to construe the contract between the Milwaukee and the L & N. We do not decide whether primary jurisdiction applies because the Commission has asked us not to do so. Even though the district court may have been authorized to construe the contract as an original matter, it was required to do so with care, lest it undermine the purposes the ICC sought to achieve in 1970 and 1973. Any construction of the contract that would allow the Seaboard to block a transfer as part of a reorganization would cut at cross purposes with the ICC’s decision not to allow the Milwaukee’s bankruptcy or reorganization to end the track-age rights. By transferring operating rights to the Soo without foreclosing all later disputes, the district court allowed the competitive rail service to continue pending an application by the Seaboard to the ICC to terminate the rights. This was the appropriate course of action, and it is how we interpret Order No. 809.
The Seaboard insists that the district court “exceeded its jurisdiction” in approving the transfer, but there was jurisdiction aplenty under the Restructuring Act and the Bankruptcy Code. The Seaboard means only that the district court erred. See also In re Sax, supra. And we do not think the court erred. It would have been unreasonable to construe the contract the way the Seaboard does, as giving it the *337right to veto any assignment of the rights in reorganization. The ICC refused it this right, and the reasons for the competitive operations survive. Clauses requiring approval for assignments do not apply to mergers, see United States Shoe Corp. v. Hackett, 793 F.2d 161, 165 (7th Cir.1986). The sale of the whole rail operation in reorganization is closer to a merger than to the sale of trackage rights standing alone — the paradigm for application of paragraph 15 of the contract. We need not consider whether these arguments are dis-positive; they are enough to suggest the wisdom of passing operating rights to the Soo while leaving further problems of interpretation to the ICC. Order No. 809 does not bar an application to the ICC or any changes that agency may wish to make. It does, however, bar any other form of collateral attack on the transfer of the trackage rights. The parties have disputed with some vigor the powers, if any, the ICC possesses to construe or alter the contract. We do not pass on these arguments, which will be open on review of the ICC’s decision if it is called on to make one.
IV
It all comes down to this: (1) The motion to supplement the record in No. 85-2121 is denied. (2) The Seaboard’s petition, No. 85-1573, for review of the ICC’s orders is dismissed for want of jurisdiction because the Seaboard was not a party to the ICC’s proceedings. (3) The RLEA’s petition, No. 85-1953, for review of the ICC’s order dismissing the request to approve the Milwaukee’s plan of reorganization is denied because the ICC correctly determined that it did not need to approve the plan. (4) The RLEA’s appeal, No. 85-1554, from Order No. 809 is dismissed as moot because the RLEA seeks an alteration of the terms of the sale. (5) CMC’s and the Milwaukee’s appeal, No. 85-1683, from paragraph 4 of Order No. 809 does not establish a reason to vacate the paragraph. Although we do not adopt or reject the district court’s conclusion that the Soo paid a constitutionally adequate price, it is inappropriate to facilitate a collateral attack on that judgment in the Claims Court. Appeal No. 85-1683 is dismissed. (6) Paragraph 9 of Order No. 809, to the extent it is attacked by Seaboard’s appeals Nos. 85-2122 and 85-2123, is affirmed. (7) The United States, the ICC, and the Soo shall recover their costs.