KLEINFELD, Circuit Judge,
with whom Circuit Judges KOZINSKI, O’SCANNLAIN, TALLMAN, BYBEE and CALLAHAN join,
dissenting from denial of rehearing en banc:
I respectfully dissent from the order denying rehearing en banc.
“Doe” is a pseudonym, because Mr. and Mrs. Doe were apparently spies for the United States in the Soviet empire. Of course, all we know is what the Does now allege. The facts are unconfirmed. The case comes up on an interlocutory appeal of a district court order denying the CIA’s motion to dismiss.
Mr. Doe was a diplomat for one of the “republics.” He sought to defect to the United States. He and Mrs. Doe allegedly made a deal with the CIA. Mr. Doe would stay in his position for a while and spy for us, and the CIA would then arrange the Does’ defection and resettlement, and ensure their personal and financial security for the rest of their lives. All went fine for some time, with false identities and backgrounds. But then the American bank Mr. Doe worked for in Seattle merged with another, and Mr. Doe was laid off. The CIA left him without assistance, despite an earlier promise to resume financial aid if he became unemployed. The Does sued for an order directing the CIA to provide them with due process and to pay them the money they were promised.
The issue is whether they can get into a federal district court with their claims. The long-established answer, under the Supreme Court’s opinion in Totten v. United States, has been that they cannot. Under Totten, those who spy for us cannot bring lawsuits to enforce our intelligence agencies’ promises, because that would require exposure of matters that must be kept secret in the interest of effective foreign policy. The panel opinion effectively overrules Totten. It holds that Totten’s rationale is out of date and that subsequent Supreme Court decisions undermine it. Under Agostini v. Felton, we cannot do that. Also, in reaching its conclusion, the panel decision puts us is in conflict with the Federal Circuit, which complied with Totten in Guong v. United States.
Aside from the intelligence aspect of this case, the Tucker Act requires those who sue the government for broken promises to do so in the Court of Federal Claims, not in a district court. By allowing the Does’ suit to go forward despite the Tucker Act, the panel opinion departs from our precedent in an area where there already existed an intercircuit conflict.
I. Spies
The case at bar is factually indistinguishable from Totten. In 1861, Abraham *1143Lincoln, on behalf of the United States, personally hired William A. Lloyd to spy behind Confederate lines for the duration of the war, agreeing to pay him $200 a month for his services. Lloyd subsequently died intestate, and his administrator, Totten, sued the government in the Court of Claims for Lloyd’s unpaid compensation. The Supreme Court held, in the broadest terms, that contracts for clandestine service to the government can never be sued upon. It said that such agreements necessarily contain the term that the parties’ “lips ... were to be for ever sealed,” a term that is “implied in all secret employments of the government in time of war, or upon matters affecting our foreign relations.” Here is the heart of the Court’s holding:
Our objection is not to the contract, but to the action upon it in the Court of Claims. The service stipulated by the contract was a secret service; the information sought was to be obtained clandestinely, and was to be communicated privately; the employment and the service were to be equally concealed. Both employer and agent must have understood that the lips of the other were to be for ever sealed respecting the relation of either to the matter. This condition of the engagement was implied from the nature of the employment, and is implied in all secret employments of the government in time of war, or upon matters affecting our foreign relations, where a disclosure of the service might compromise or embarrass our government in its public duties, or endanger the person or injure the character of the agent. If upon contracts of such a nature an action against the government could be maintained in the Court of Claims, whenever an agent should deem himself entitled to greater or different compensation than that awarded to him, the whole service in any case, and the manner of its discharge, with the details of dealings with individuals and officers, might be exposed, to the serious detriment of the public. A secret service, with liability to publicity in this way, would be impossible; and, as such services are sometimes indispensable to the government, its agents in those services must look for their compensation to the contingent fund of the department employing them, and to such allowance from it as those who dispense that fund may award. The secrecy which such contracts impose precludes any action for their enforcement. The publicity produced by an action would itself be a breach of a contract of that kind, and thus defeat a recovery.
The Does’ case is factually indistinguishable from Totten. Like William Lloyd, the Does were engaged to provide secret services to the United States behind enemy lines. Like Lloyd, they served to the great benefit of the United States in circumstances that could have gotten them killed. And like Lloyd, they allegedly got stiffed by the government providing less compensation than required by the contracts when the time came for the United States to pay up.
The panel held, however, that Totten is no longer good authority. As for the contractual aspects of Totten, the Doe panel held that Totten was inapplicable for two reasons:
*1144First, ... unlike Totten, the Does do not seek only enforcement of a contract. Rather their principal concern at this point ... is to compel fair process and application of substantive law to their claims within the Central Intelligence Agency[ ].... Second, Totten assumed “publicity” inconsistent with the implicit promise of secrecy as inherent in any judicial proceeding and did not consider whether there are means to conduct judicial proceedings without unacceptable attendant “publicity.” Since Totten, courts, including the Supreme Court, have developed means of accommodating asserted national security interests in judicial proceedings....
The panel then reinterprets the remainder of Totten as “an early expression of the evidentiary state secrets privilege,” which only allows for dismissal “after complying with the formalities and court investigation requirements that have developed since Totten.”
As Judge Tallman’s dissent points out, every aspect of this rationale is mistaken. The most obvious way that the opinion is wrong is that, no matter how much the courts have developed routines for handling assertions of official secrets, a lower court cannot overrule the Supreme Court. Whether subsequent Supreme Court authority implicitly undermines a Supreme Court decision or not, lower courts must follow and apply a controlling Supreme Court decision. As the Court said in Agostini, “[w]e reaffirm that if a precedent of this Court has direct application in a ease, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions.” The Supreme Court can overrule Totten. We cannot.
Second, the gossamer distinctions the panel tries to draw between the Does’ and Totten’s claims for the purpose of defeating Totten’s language about implied secrecy terms do not amount to anything. Whether it is called a plea for fairer process or a simple contract claim for damages, the Does, like Totten’s decedent, sue the government to obtain a remedy for its breach of an agreement to compensate them for intelligence services. In both cases, the engagements were secret, obviously and necessarily so, since doubtless both engagements involved the commission of serious crimes in the locations where the intelligence agents were to perform them. Both also would amount to serious provocations by the United States in the eyes of the governing forces in those locations.
The panel’s contention that the purposes of Totten can be served by the CIA asserting the state-secrets privilege, which would then permit in camera inspection of papers by the district court and so forth, leaves out the most important purpose of all: to keep the whole engagement utterly and entirely secret. I see no reason that progress in judicial techniques should make any difference. Intelligence is such an ancient part of conflict that the Bible suggested nothing novel when Joshua sent spies to reconnoiter Jericho. Circumstances have not changed materially. The very existence of the engagement has to be secret and has to remain secret. If a lawsuit is filed but some papers remain secret, that is not enough. An intelligent observer, knowing something of the events, can figure out from the barest *1145indications in a lawsuit what it is all about. As Totten says, “such services are sometimes indispensable to the government,” and “[a] secret service, with liability to publicity in this way, would be impossible.” Sometimes, to avoid provocations to war or other diplomatic imbroglios, the government has to avoid disclosing what it did, even long after it has done it. Spying is among the “matters affecting our foreign relations, where a disclosure of the service might compromise or embarrass our government in its public duties.”
The panel’s assurances that there is nothing to fear are not enough in light of Totten and its purposes. Requiring the CIA to abide by the formalities of making a privilege claim will involve the CIA having to deny or disclose the very existence of the secret relationship. Even asserting that there is a secret to protect, as the state-secrets privilege used in other contexts requires, amounts to letting the cat out of the bag. It is such disclosure of the relationship’s very existence that Totten sought to avoid. And where there is no espionage relationship to protect, the CIA will have to say as much. That will make all non-denials effectively confirmations.
The panel’s attempt to interpret Totten so narrowly that it does not even apply to a case materially indistinguishable from it puts us in conflict with the Federal Circuit. In Guong v. United States, the plaintiff tried to distinguish Totten on the ground that the CIA had hired him as a saboteur in North Vietnam, and not as a spy. The Federal Circuit rejected the distinction and any other narrowing of Totten: “Tot-ten stands for the proposition that no action can be brought to enforce an alleged contract with the government when, at the time of its creation, the contract was secret or covert.”
Guong also rejected the plaintiffs attempt, similar to the Doe panel’s, to assimilate Totten into cases about the state secrets evidentiary privilege. Privilege cases, including Webster v. Doe and United States v. Reynolds, do not involve engagements of spies in foreign countries, but rather disclosures during litigation about other kinds of secrets. As Guong held, “[a] close reading of Reynolds reveals that it does not limit or modify the authority of Totten or its rationale.” Guong also makes the point that Totten is not limited to currently active spies. Even in Totten, there was no direct risk of harm from ongoing conflicts, because the Civil War had been over for ten years when the Court decided the case. Thus, in the Federal Circuit, unlike in the Ninth Circuit after Doe, Totten remains good law.
II. Tucker Act
The Tucker Act grants jurisdiction to the Court of Federal Claims to hear contract actions against the United States. Totten itself was brought under a predecessor to the present Tucker Act. The Tucker Act’s conferral of jurisdiction on the Court of Federal Claims has been read to be exclusive of district court jurisdiction *1146when the amount at stake is greater than $10,000.
The panel opinion, however, reads the Tucker Act narrowly, so that even though the Does’ claims are in the first instance dependent on the contract, the Does can proceed in district court. The panel so holds because the Does purport to base their claims on a denial of due process and phrase the relief they seek as declaratory and injunctive. Those distinctions have no force, however, as we held in Tucson Airport. There is apparently a circuit conflict between Tucson Airport and the District of Columbia Circuit’s decision in Transohio Savings Bank v. Director, Office of Thrift Supervision. But this issue should not have been reached here, because the case should have been dismissed at the outset because it was based on a covert engagement.
III. Conclusion
I sincerely hope that Totten’s decedent, William Lloyd, was the only spy the United States has ever failed to pay what it promised (assuming Lloyd was telling the truth), and that the failure occurred only because Lincoln was assassinated before he could see to payment. I hope that the Does’ account is fictional (though I do not intimate that it is, having no knowledge). Little could be worse for our ability to engage spies than insecurity about whether they will get what was promised to them. If what the Does allege is true, a serious injustice has been done to them, and the injustice to them is seriously harmful to the long-term security interests of the United States.
Nevertheless, the judicial branch cannot right such a wrong without disclosure of the engagement’s existence, which, as Tot-ten said, must remain forever secret. It will not do to have word circulating in whatever former Iron Curtain country the Does come from that the collapse of its totalitarian regime was brought about partly by CIA spies and not wholly by its own people’s thirst for freedom. Joshua needed spies, Lincoln needed spies, we needed spies to deal with the Soviet empire, and spies will be needed as long as there are men on earth. Judicial determination of what must remain secret, even after in camera inspection of documents, is no substitute for dismissal at the outset, because the CIA cannot come into court and assert the existence of a secret without revealing that there is a secret. The use of spies is far more humane than some of the alternatives for dealing with serious international conflicts. And their use must remain secret.