Plaintiffs Carrie A. McMellon, Lori Dawn White, Kathy D. Templeton, and Cheri Call seek damages under the Suits in Admiralty Act, 46 U.S.C.A.App. §§ 741-52 (West Supp.2003) (the "SIAA" or the "Act") for injuries they suffered when they went over the gates of the Robert C. Byrd Locks and Dam while riding jet skis. The district court granted summary judgment in favor of the government, and the plaintiffs appealed. A divided panel of this court concluded that prior precedent from this circuit foreclosed the government's argument that the Suits in Admiralty Act contained an implied discretionary function exception that barred the plaintiffs' claims. The panel also concluded that the government had a duty to warn about the existence of the dam, and the panel reversed the district court's grant of summary judgment and remanded for further proceedings. See McMellon v. United States, 338 F.3d 287 (4th Cir.2003) ("McMellon I").

The government filed a petition for rehearing en banc with regard to the discretionary function exception issue. Sitting en banc, we now conclude that the government's waiver of sovereign immunity reflected in the Suits in Admiralty Act is subject to an implied exception similar to the discretionary function exception contained within the Federal Tort Claims Act, 28 U.S.C.A. §§ 2671-2680 (West 1994 & Supp.2003). Accordingly, we vacate the district court's summary judgment order and remand to allow the district court to determine whether the facts of this case fall within that exception to the waiver of sovereign immunity and to conduct any other proceedings that might become necessary.

The relevant facts are set out in detail in the panel's opinion, and we will only briefly recount them here. The plaintiffs were riding two jet skis on the Ohio River in the vicinity of the Robert C. Byrd Locks and Dam, a government-owned and operated facility on the Ohio River. The plaintiffs approached what they believed to be a bridge but which turned out to be the gates of the dam. Unable to stop or turn around, the plaintiffs were injured when they went over the gates and dropped approximately twenty-five feet to the water below. At the time of the accident, there were several warning signs on the upstream side of the dam, but the plaintiffs did not see them, and their evidence indicated that the signs were difficult to see from the river. The plaintiffs brought this action under the SIAA, alleging that the government had a duty to warn about the dangers of the dam and that the signs in place were inadequate to satisfy this duty.

The government moved to dismiss, arguing that it was protected by an implied discretionary function exception to the SIAA's waiver of sovereign immunity. The government also moved for summary judgment on the merits of the plaintiffs' claims on the grounds that it had no duty to warn about the dam and that the warnings it provided were, in any event, adequate.

Relying on prior authority from this court, the district court rejected the government's claim that it was protected by an implied discretionary function exception. The court, however, granted the government's motion for summary judgment, concluding that the government had no duty to warn about the dam. The panel in McMellon I agreed with the district judge on the first point, but reversed the grant of summary judgment, concluding that the government in fact had a duty to warn. As noted above, we granted rehearing en banc to consider whether the SIAA contains an implied discretionary function exception to its waiver of sovereign immunity.

At the heart of the question presented to this en banc court is the continuing viability of Lane v. United States, 529 F.2d 175 (4th Cir.1975). In Lane, this court flatly rejected the argument that a discretionary function exception should be read into the SIAA. After Lane, however, two cases from this circuit arguably applied some form of a discretionary function exception to cases arising under the SIAA. See Tiffany v. United States, 931 F.2d 271, 276-77 (4th Cir.1991); Faust v. South Carolina State Highway Dep't, 721 F.2d 934, 939 (4th Cir.1983).

Because we are sitting en banc, there is no doubt that we have the power to overrule Lane should we conclude it was wrongly decided. See, e.g., United States v. Lancaster, 96 F.3d 734, 742 n. 7 (4th Cir.1996) (en banc). The panel opinions in this case, however, raised the question of whether a panel of this court may likewise overrule a decision issued by another panel. The question of the binding effect of a panel opinion on subsequent panels is of utmost importance to the operation of this court and the development of the law in this circuit. Accordingly, before considering the merits of the discretionary function question, we first address this important procedural issue.1

A number of cases from this court have stated the basic principle that one panel cannot overrule a decision issued by another panel. This statement is typically made in the course of a party's request that a panel opinion be overruled by another panel. See, e.g., United States v. Prince-Oyibo, 320 F.3d 494, 497-98 (4th Cir.), cert. denied, ___ U.S. ___, 124 S.Ct. 957, 157 L.Ed.2d 796 (2003); Scotts Co. v. United Indus. Corp., 315 F.3d 264, 271 n. 2 (4th Cir.2002); United States v. Chong, 285 F.3d 343, 346 (4th Cir.2002); Potomac Elec. Power Co. v. Electric Motor & Supply, Inc., 262 F.3d 260, 264 n. 2 (4th Cir. 2001); Young v. City of Mount Ranier, 238 F.3d 567, 579 n. 9 (4th Cir.2001). Instances of conflicting panel opinions, however, are apparently fairly uncommon in this circuit, given the paucity of cases explaining how such conflicts should be resolved. Nonetheless, we have made it clear that, as to conflicts between panel opinions, application of the basic rule that one panel cannot overrule another requires a panel to follow the earlier of the conflicting opinions. See Booth v. Maryland, 327 F.3d 377, 383 (4th Cir.2003).

Most of the other circuits agree and follow the earlier of conflicting panel opinions. See, e.g., Hiller v. Oklahoma ex rel. Used Motor Vehicle & Parts Comm'n, 327 F.3d 1247, 1251 (10th Cir.2003) (explaining that when panel opinions are in conflict, "we are obligated to follow the earlier panel decision over the later one"); Morrison v. Amway Corp., 323 F.3d 920, 929 (11th Cir.2003) ("When faced with an intra-circuit split we must apply the earliest case rule, meaning when circuit authority is in conflict, a panel should look to the line of authority containing the earliest case, because a decision of a prior panel cannot be overturned by a later panel." (internal quotation marks omitted)); Southwestern Bell Tel. Co. v. City of El Paso, 243 F.3d 936, 940 (5th Cir.2001) ("When two holdings or lines of precedent conflict, the earlier holding or line of precedent controls."); Kovacevich v. Kent State Univ., 224 F.3d 806, 822 (6th Cir.2000) ("[W]e must defer to a prior case when two panel decisions conflict."); Ryan v. Johnson, 115 F.3d 193, 198 (3rd Cir.1997) ("Under Third Circuit Internal Operating Procedure 9.1, when two decisions of this court conflict, we are bound by the earlier decision."); Newell Cos. v. Kenney Mfg. Co., 864 F.2d 757, 765 (Fed.Cir.1988) ("This court has adopted the rule that prior decisions of a panel of the court are binding precedent on subsequent panels unless and until overturned in banc. Where there is direct conflict, the precedential decision is the first." (citations omitted)). The Eighth Circuit, however, follows a different approach — a panel "faced with conflicting precedents [is] free to choose which line of cases to follow." Graham v. Contract Transp., Inc., 220 F.3d 910, 914 (8th Cir.2000).

We believe the better practice is the one articulated by the panel majority and followed by most other circuits. When published panel opinions are in direct conflict on a given issue, the earliest opinion controls, unless the prior opinion has been overruled by an intervening opinion from this court sitting en banc or the Supreme Court. We recognize, of course, that application of this rule does require a panel to effectively ignore certain opinions duly decided by a properly constituted panel of the court. And as pointed out by Judge Niemeyer in McMellon I, to ignore a panel opinion is, at least on one level, inconsistent with our rule prohibiting one panel from overruling another panel opinion. See Walker v. Mortham, 158 F.3d 1177, 1189 (11th Cir.1998) ("Of course, by adopting the `earliest case' rule to resolve intra-circuit splits, we are still in a sense ignoring the prior panel precedent rule — by choosing one line of cases, we are implicitly overruling the other line of cases. This is, however, a necessary consequence of an intra-circuit split. . . ."). The other alternative (followed by the Eighth Circuit), however, suffers from precisely the same problem, because it allows a panel to ignore one panel opinion in favor of another panel opinion that it finds more persuasive. Thus, neither approach is a perfect one. In our view, however, the alternative approach utilized by the Eighth Circuit has certain negative attributes that are not shared by the earliest-case-governs rule.

The alternative approach would have the effect of extending the life of intra-circuit conflicts indefinitely, as each subsequent panel considering an issue over which there was once a conflict would apparently be free to decide for itself which prior decision it wished to follow. This iteration of the rule could almost encourage the creation of intra-circuit conflicts. A panel displeased with the way an issue previously had been decided could simply ignore the case with which it disagreed, thus creating a conflict and at least the possibility that the analysis of the second panel opinion would ultimately prevail.

By contrast, requiring subsequent panels to follow the earliest of the conflicting cases does not allow the possibility of manufactured intra-circuit conflicts. In addition, the earliest-case-governs rule brings intra-circuit conflicts to an end as soon as they are recognized. See Walker, 158 F.3d at 1189 n. 25 ("The `earliest case' rule also has the virtue (if consistently applied) of bringing intra-circuit splits to a screeching halt; the `common sense and reason' rule, in contrast, can drag such splits out indefinitely as different panels reach different conclusions about what is common-sensical and reasonable."). The rule that requires subsequent panels to follow the earliest case in the event of a conflict thus minimizes the instability and unpredictability that intra-circuit conflicts inevitably create.

While we recognize that a three-judge panel has the statutory and constitutional power to overrule the decision of another three-judge panel, we believe that, as a matter of prudence, a three-judge panel of this court should not exercise that power. Accordingly, we conclude that when there is an irreconcilable conflict between opinions issued by three-judge panels of this court, the first case to decide the issue is the one that must be followed, unless and until it is overruled by this court sitting en banc or by the Supreme Court.2

We now proceed to the question that we granted rehearing en banc to consider: Whether we should overrule Lane and read into the SIAA a discretionary function exception. To put this issue in the proper context, some historical background is necessary.

During the first part of this century, the United States government had yet to waive its sovereign immunity in admiralty actions. Thus, if a vessel owned or operated by the government caused damage to a private vessel, the private owner was without recourse, even though the government, of course, could seek damages from a private vessel owner who negligently damaged a government vessel. See Canadian Aviator, Ltd. v. United States, 324 U.S. 215, 219, 65 S.Ct. 639, 89 L.Ed. 901 (1945). Congress frequently passed private bills authorizing relief for particular vessel owners damaged by the government's actions, a method that proved to be rather cumbersome and inefficient, see id., particularly once the government, through the Shipping Board, became the owner of many merchant vessels. See Marine Coatings of Alabama v. United States, 71 F.3d 1558, 1560 (11th Cir.1996). In 1916, Congress passed the Shipping Act, 46 U.S.C.A.App. § 801, which provided that "Shipping Board vessels while employed as merchant vessels were subject to all laws, regulations, and liabilities governing merchant vessels regardless of the fact that the United States owned or had an interest in them." Canadian Aviator, 324 U.S. at 219, 65 S.Ct. 639 (internal quotation marks omitted).

The Supreme Court interpreted the Shipping Act to authorize in rem actions and attendant arrests and seizure of government vessels. See The Lake Monroe, 250 U.S. 246, 39 S.Ct. 460, 63 L.Ed. 962 (1919). Congress responded to that decision in 1920 by passing the Suits in Admiralty Act, which prohibited in rem actions but instead authorized in personam admiralty actions in cases involving government-owned or operated vessels that were "employed as a merchant vessel." SIAA § 2, 41 Stat. 525-26 (1920). The SIAA expressly granted the government the right to take advantage of all of the statutory limitations of liability available to private parties,3 see SIAA § 6, 41 Stat. 527 (1920), but the Act did not include any exceptions to its waiver of sovereign immunity for the cases that fell within its scope.

For purposes of the original SIAA, a public vessel (such as a naval vessel) was distinct from a government-owned or operated merchant vessel. Accordingly, "the Government's sovereign immunity still prevented a claimant from bringing an in rem or any other proceeding in admiralty against the United States for injury caused by a public vessel." Marine Coatings, 71 F.3d at 1560. Congress rectified that anomaly in 1925 by passing the Public Vessels Act (the "PVA"), which authorized in personam admiralty actions seeking recovery for "damages caused by a public vessel of the United States." 46 U.S.C.A.App. § 781 (West 1975). Like the SIAA, the PVA contained no exceptions to its waiver of sovereign immunity for any particular claims otherwise falling within its scope.

When PVA and SIAA were enacted, Congress had yet to implement a waiver of the government's sovereign immunity as to non-maritime torts. While Congress had long believed that "the Government should assume the obligation to pay damages for the misfeasance of employees in carrying out its work," Dalehite v. United States, 346 U.S. 15, 24, 73 S.Ct. 956, 97 L.Ed. 1427 (1953), it took Congress nearly thirty years to reach agreement on the form that the waiver of sovereign immunity should take, see id. Finally, in 1946, Congress passed the Federal Tort Claims Act ("FTCA"), 28 U.S.C.A. §§ 2671-2680 (West 1994 & Supp.2003), which waives sovereign immunity for most torts committed by government employees, subject to several statutory exceptions. See 28 U.S.C.A. § 2680.

The most important of these exceptions to the waiver of sovereign immunity is the discretionary function exception. In 1942, while Congress was working on what would ultimately become the FTCA, an assistant attorney general testified before the House Judiciary Committee and stated his view that courts probably would not impose liability on the government for discretionary actions, even if the act waiving sovereign immunity did not include a specific exception for such actions. See Dalehite, 346 U.S. at 27, 73 S.Ct. 956; see also United States v. S.A. Empresa de Viacao Aerea Rio Grandense (Varig Airlines), 467 U.S. 797, 810, 104 S.Ct. 2755, 81 L.Ed.2d 660 (1984). Apparently not content to rely on the courts to protect the government from unintended liability, Congress included an express discretionary function exception in the FTCA. The exception states that the FTCA's waiver of sovereign immunity does not apply to claims "based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a federal agency or an employee of the Government, whether or not the discretion involved be abused." 28 U.S.C.A. § 2680(a).

The FTCA excludes from its reach claims for which a remedy is provided by the SIAA or the PVA. See 28 U.S.C.A. § 2680(d). In 1946, when the FTCA was enacted, the SIAA and PVA waived sovereign immunity only in cases involving public or merchant vessels. Thus, admiralty tort actions not involving public or merchant vessels could be pursued against the government under the FTCA. See United States v. United Cont'l Tuna Corp., 425 U.S. 164, 172, 96 S.Ct. 1319, 47 L.Ed.2d 653 (1976) (explaining that "[m]aritime tort claims deemed beyond the reach of [the SIAA and PVA] could be brought only on the law side of the district courts under the Federal Tort Claims Act"); see also Somerset Seafood Co. v. United States, 193 F.2d 631, 634 (4th Cir.1951) (concluding that certain maritime claims fell within the jurisdictional scope of the FTCA rather than the SIAA). Of course, any such actions under the FTCA would be subject to the limitations of that act, including the discretionary function exception.

During this time, the practice of maritime law proved to be exceedingly complex. The distinction between public vessels (subject to suit under the PVA) and merchant vessels (subject to suit under the SIAA) was elusive, and, beyond noting that the categories were mutually exclusive, courts had difficulty precisely articulating the difference between the types of vessels. See Continental Tuna, 425 U.S. at 172 n. 1, 96 S.Ct. 1319; see also Blanco v. United States, 775 F.2d 53, 57 n. 4 (2nd Cir.1985). To further complicate matters, the Tucker Act provided that general contract actions against the government fell either within the exclusive jurisdiction of the Court of Claims or the concurrent jurisdiction of the Court of Claims and the district court, depending on the amount in controversy. See Blanco, 775 F.2d at 57 n. 4. But once the SIAA was passed, most (but not all) maritime contract actions involving the government fell within the admiralty jurisdiction of the district courts and were not subject to the Tucker Act. See Matson Navigation Co. v. United States, 284 U.S. 352, 359-60, 52 S.Ct. 162, 76 L.Ed. 336 (1932) (holding that the Court of Claims lacked jurisdiction over a contract whose subject matter was covered by the Suits in Admiralty Act); see also Continental Tuna, 425 U.S. at 172, 96 S.Ct. 1319 (explaining that "contract claims not encompassed by [the PVA or the SIAA] fell within the Tucker Act, which lodged exclusive jurisdiction in the Court of Claims for claims exceeding $10,000"). If an admiralty practitioner guessed wrong and filed in the wrong court or under the wrong act, the consequences could be dire, because the statutes of limitations under the SIAA and the PVA were substantially shorter than that of the Tucker Act, and there was no procedure for transferring cases between the Court of Claims and the district courts. See Continental Tuna, 425 U.S. at 172-73, 96 S.Ct. 1319; Blanco, 775 F.2d at 57 n. 4.

Congress put an end to these problems in 1960. First, Congress authorized transfers between the Court of Claims and district courts for cases filed in the wrong court. See Pub.L. No. 86-770, §§ 1-2, 74 Stat. 912 (1960). In addition, Congress amended the SIAA by eliminating the reference to "merchant vessel." See id., § 3, 74 Stat. 912 (1960). Thus, after the 1960 amendments, the SIAA authorized in personam admiralty actions against the United States "[i]n cases where if such vessel were privately owned or operated, or if such cargo were privately owned or possessed, or if a private person or property were involved, a proceeding in admiralty could be maintained." 46 U.S.C.A.App. § 742 (West Supp.2003). As is relevant to this case, the 1960 amendments worked no other substantive change to the SIAA. Specifically, no limitations on the waiver of sovereign immunity similar to those contained in the FTCA were added to the Act.

There is no indication in the legislative history that Congress intended the 1960 amendments to do anything other than correct the jurisdictional problems mentioned above. Nonetheless, courts have consistently concluded that the 1960 amendments greatly expanded the reach of the SIAA to include essentially all admiralty tort actions that could be asserted against the government. See, e.g., Trautman v. Buck Steber, Inc., 693 F.2d 440, 444 (5th Cir.1982); Bearce v. United States, 614 F.2d 556, 558 (7th Cir.1980); Kelly v. United States, 531 F.2d 1144, 1148-49 (2nd Cir.1976); Lane, 529 F.2d at 179. Thus, after the 1960 amendments, admiralty actions that previously would have been brought under the FTCA instead had to be brought under the SIAA — both the SIAA and the FTCA make it clear that the SIAA provides the exclusive remedy for cases falling within its scope. See 46 U.S.C.A. § 745 ("[W]here a remedy is provided by this chapter it shall hereafter be exclusive of any other action by reason of the same subject matter. . . ."); 28 U.S.C.A. § 2680(d) (excluding from the FTCA "[a]ny claim for which a remedy is provided by sections 741-752, 781-790 of Title 46, relating to claims or suits in admiralty against the United States").

After the 1960 amendments to the SIAA, questions began to arise as to whether the SIAA should be read to include an implied discretionary function exception. After all, the admiralty claims that had previously fallen under the FTCA had been subject to that act's discretionary function exception. If the SIAA did not include a discretionary function exception at least as to those claims that previously would have been brought under the FTCA, then the government would suddenly be exposed to liability in areas where it had been protected.

The argument that the SIAA included an implied discretionary function exception initially was not very well received. In De Bardeleben Marine Corp. v. United States, 451 F.2d 140 (5th Cir.1971), the Fifth Circuit concluded that a review of the language of the amended SIAA and its legislative history foreclosed the argument:

It is true that the legislative history says nothing concerning a purpose to surrender immunity [as to claims previously within the scope of the FTCA]. It is equally true, though, that § 742 by its own terms disavows governmental immunity in admiralty actions against the United States. Had the sole purpose of the legislation been to clarify the confusing language of the old SIA this would have been better done by modifying the old § 742 to contain a clear definition of merchant, public, vessels and cargoes plus a delineation of contract claims growing out of Governmental shipping operations.

More positively, the legislative history shows that almost on the eve of a probable enactment of a narrowly constructed solution to conflicts in jurisdiction between the Court of Claims and the District Courts, Congress . . . set out to solve the underlying problems by eliminating the historic restriction of SIA-PVA liability to non-contractual claims relating to ships or cargo. It was to assimilate the Government to the private person in relation to any or all transactions giving rise to liability in the Admiralty. It would be incongruous to impute to Congress a purpose to perpetrate confusion, not by reason of choosing the wrong forum, but by importing substantive standards of liability and governmental defenses by a retrospective analysis of what would have been the case prior to 1960. Reimportation of FTCA provisions or exceptions produces obviously unintended and irrational distinctions.

Id. at 145-46 (footnotes omitted). This court in Lane likewise rejected the argument:

[The SIAA] contains no discretionary function exception, and the Tort Claims Act contains a specific exception of claims for which the Suits in Admiralty Act provides a remedy. Thus it is clear that this action could not have been brought under the Tort Claims Act, and it is properly maintainable under the Suits in Admiralty Act, which is an effective waiver of sovereign immunity.

There is no basis upon which we can import the many exceptions in the Tort Claims Act into the Suits in Admiralty Act, where the United States is to be held accountable in admiralty whenever a private person, in similar circumstances, would be.

Lane, 529 F.2d at 179 (footnotes omitted).

Shortly after Lane was decided, however, the tide turned, and courts began accepting the argument that the post-1960 SIAA included an implied discretionary function exception. Coincidentally, the First Circuit was the first circuit to explicitly so conclude. See Gercey v. United States, 540 F.2d 536, 539 (1st Cir.1976). After Gercey, every circuit to consider the question likewise concluded that an implied discretionary function exception should be read in the SIAA. See Tew v. United States, 86 F.3d 1003, 1005 (10th Cir.1996); Earles v. United States, 935 F.2d 1028, 1032 (9th Cir.1991); Sea-Land Serv., Inc. v. United States, 919 F.2d 888, 893 (3d Cir.1990); Robinson v. United States (In re Joint E. & S. Dists. Asbestos Litig.), 891 F.2d 31, 34-35 (2d Cir.1989); Williams v. United States, 747 F.2d 700 (11th Cir.1984) (per curiam), aff'g Williams ex rel. Sharpley v. United States, 581 F.Supp. 847 (S.D.Ga.1983); Gemp v. United States, 684 F.2d 404, 408 (6th Cir.1982); Canadian Transp. Co. v. United States, 663 F.2d 1081, 1085 (D.C.Cir.1980); Bearce v. United States, 614 F.2d 556, 560 (7th Cir.1980). Even the Fifth Circuit, describing the discussion of the issue in De Bardeleben as non-binding dictum, has concluded that the SIAA includes a discretionary function exception. See Wiggins v. United States, 799 F.2d 962, 964-66 (5th Cir.1986).

The question, then, is whether this court should overrule Lane, and as a consequence join the other circuits in concluding that the SIAA contains an implied discretionary function exception. For reasons that we will explain below, we conclude that the SIAA's waiver of sovereign immunity should be read to include an implied discretionary function exception to that waiver. Accordingly, we hereby overrule Lane to the extent that it concludes the SIAA does not include an implied discretionary function exception.4

In support of its view that the SIAA includes an implied discretionary function exception, the government makes two primary arguments. First, the government contends that Congress did not intend to waive immunity under the SIAA for discretionary actions, and that this court, therefore, should not interpret the SIAA in a manner inconsistent with Congressional intent. Second, the government contends that principles of separation-of-powers require us to exclude discretionary actions from the SIAA's waiver of sovereign immunity. We consider these arguments in turn.

First, we consider the government's argument that we must give effect to what the government contends is a clear Congressional intent to exclude discretionary acts from the SIAA's waiver of sovereign immunity.

The Supreme Court has repeatedly explained that the plain language of a statute is the best evidence of Congressional intent. See, e.g., Holloway v. United States, 526 U.S. 1, 6, 119 S.Ct. 966, 143 L.Ed.2d 1 (1999). As noted above, the SIAA includes no list of exceptions to its waiver of sovereign immunity, but instead provides only that the government is entitled to the limitations of liability that are available in admiralty to private defendants. Thus, the plain language of the SIAA seems to reflect a Congressional intent that discretionary acts should not be excluded from the waiver of sovereign immunity. See Barnhart v. Sigmon Coal Co., 534 U.S. 438, 450, 122 S.Ct. 941, 151 L.Ed.2d 908 (2002) (explaining that when construing a statute, "[t]he first step is to determine whether the language at issue has a plain and unambiguous meaning with regard to the particular dispute in the case. The inquiry ceases if the statutory language is unambiguous and the statutory scheme is coherent and consistent." (citations and internal quotation marks omitted)).

The government, however, makes the rather remarkable argument that the legislative history of the FTCA supports its view of what Congress intended when it amended the SIAA nearly twenty years later. According to the government, the legislative history of the FTCA shows that Congress believed that the courts would not hold the government liable for discretionary acts whether or not the FTCA included an express exception for such actions. See Varig, 467 U.S. at 810, 104 S.Ct. 2755; Dalehite, 346 U.S. at 27, 73 S.Ct. 956. Accordingly, the government suggests that there was no reason for Congress to include a discretionary function exception when it amended the SIAA in 1960, because Congress assumed that courts would imply such an exception.

This is a difficult argument to accept. First of all, we fail to see how a 1942 legal opinion of an assistant attorney general as to the probability that courts would carve out discretionary acts from a waiver of sovereign immunity is indicative of what Congress did or did not intend some twenty years later. But more importantly, Congress in fact included a discretionary function exception in the FTCA even in the face of a legal opinion that the exception was not necessary. Thus, the discretionary function exception was important enough to Congress in 1946 that Congress included an express exception in the FTCA, to resolve any doubt about whether courts would create such an exception on their own. If the exception remained as important to Congress in 1960 when it amended the SIAA as it was when the FTCA was enacted, then it stands to reason that Congress would have written the exception into the SIAA then, particularly since the 1960 SIAA amendments transferred jurisdiction over a number of claims from the FTCA to the SIAA. Cf., e.g., Binder v. Long Island Lighting Co., 933 F.2d 187, 193 (2nd Cir.1991) ("Congress enacted the ADEA in the wake of Title VII, and we believe that any omission in the text of the ADEA of a provision found in Title VII is likely to reflect a deliberate decision on Congress's part.").

Likewise, resort to familiar canons of statutory construction fails to support the interpretation urged by the government. For example, it is well established that waivers of sovereign immunity "must be construed strictly in favor of the sovereign and not enlarged beyond what the language requires." United States v. Nordic Village, Inc., 503 U.S. 30, 34, 112 S.Ct. 1011, 117 L.Ed.2d 181 (1992) (citation, internal quotation marks, and alterations omitted); see also United States v. White Mountain Apache Tribe, 537 U.S. 465, 472, 123 S.Ct. 1126, 155 L.Ed.2d 40 (2003) ("Jurisdiction over any suit against the Government requires a clear statement from the United States waiving sovereign immunity.. . . The terms of consent to be sued may not be inferred, but must be unequivocally expressed." (internal quotation marks omitted)). The waiver of sovereign immunity contained within the SIAA, however, is clear and unequivocal, providing that an in personam admiralty action may be brought against the government if such an action could be maintained against a private person. Contrary to the government's suggestion, we simply cannot create an ambiguity in the SIAA by looking to the language and structure of the FTCA. Cf. Lamie v. United States Trustee, 540 U.S. 526, 124 S.Ct. 1023, 1030, 157 L.Ed.2d 1024 (2004) (rejecting argument that statute was ambiguous based on assumption that Congress intended an amended statute to reflect the parallelism of a prior version of the statute: "One determines ambiguity, under this contention, by relying on the grammatical soundness of the prior statute. That contention is wrong. The starting point in discerning congressional intent is the existing statutory text, and not the predecessor statutes." (citation omitted)).

Moreover, even if the maxim requiring narrow construction of waivers of sovereign immunity were applicable, the result urged by the government runs contrary to another maxim of statutory construction which cautions that courts cannot "assume the authority to narrow the waiver that Congress intended." Smith v. United States, 507 U.S. 197, 203, 113 S.Ct. 1178, 122 L.Ed.2d 548 (1993); accord Rayonier Inc. v. United States, 352 U.S. 315, 320, 77 S.Ct. 374, 1 L.Ed.2d 354 (1957) ("There is no justification for this Court to read exemptions into the [FTCA] beyond those provided by Congress. If the Act is to be altered that is a function for the same body that adopted it.").

Accordingly, we cannot conclude that Congress clearly intended for the SIAA's waiver of sovereign immunity to be subject to an exception for discretionary functions, nor can we reach that conclusion by resort to traditional tools of statutory construction. But as we will explain below, we reach that very result by consideration and application of separation-of-powers principles.

"Even before the birth of this country, separation of powers was known to be a defense against tyranny." Loving v. United States, 517 U.S. 748, 756, 116 S.Ct. 1737, 135 L.Ed.2d 36 (1996) (citing Montesquieu, The Spirit of the Laws 151-152 (T. Nugent transl. 1949); 1 W. Blackstone, Commentaries *146-*147, *269-*270). Thus,

[t]he Constitution sought to divide the delegated powers of the new Federal Government into three defined categories, Legislative, Executive and Judicial, to assure, as nearly as possible, that each branch of government would confine itself to its assigned responsibility. The hydraulic pressure inherent within each of the separate Branches to exceed the outer limits of its power, even to accomplish desirable objectives, must be resisted.

INS v. Chadha, 462 U.S. 919, 951, 103 S.Ct. 2764, 77 L.Ed.2d 317 (1983).

The "concept of separation of powers," then, is exemplified by "the very structure of the Constitution." Miller v. French, 530 U.S. 327, 341, 120 S.Ct. 2246, 147 L.Ed.2d 326 (2000) (internal quotation marks omitted). "The Framers regarded the checks and balances that they had built into the tripartite Federal Government as a self-executing safeguard against the encroachment or aggrandizement of one branch at the expense of the other." Buckley v. Valeo, 424 U.S. 1, 122, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976). "While the boundaries between the three branches are not `hermetically' sealed, the Constitution prohibits one branch from encroaching on the central prerogatives of another." Miller, 530 U.S. at 341, 120 S.Ct. 2246 (citation and internal quotation marks omitted).

Accordingly, the Supreme Court has "not hesitated to strike down provisions of law that either accrete to a single Branch powers more appropriately diffused among separate Branches or that undermine the authority and independence of one or another coordinate Branch." Mistretta v. United States, 488 U.S. 361, 382, 109 S.Ct. 647, 102 L.Ed.2d 714 (1989). With regard to the Executive Branch, separation-of-powers concerns are focused "on the extent to which [a statute] prevents the Executive Branch from accomplishing its constitutionally assigned functions." Nixon v. Administrator of Gen. Servs., 433 U.S. 425, 443, 97 S.Ct. 2777, 53 L.Ed.2d 867 (1977). In cases involving the Judicial Branch, the Court has traditionally acted to ensure "that the Judicial Branch neither be assigned nor allowed tasks that are more properly accomplished by other branches," and "that no provision of law impermissibly threatens the institutional integrity of the Judicial Branch." Mistretta, 488 U.S. at 383, 109 S.Ct. 647 (citation, internal quotation marks and alteration omitted). "Even when a branch does not arrogate power to itself, . . . the separation-of-powers doctrine requires that a branch not impair another in the performance of its constitutional duties." Loving, 517 U.S. at 757, 116 S.Ct. 1737.

The Supreme Court has made clear that the discretionary function exception contained in the FTCA is grounded in separation-of-powers concerns. As the Court has explained, the exception "marks the boundary between Congress' willingness to impose tort liability upon the United States and its desire to protect certain governmental activities from exposure to suit by private individuals." Varig Airlines, 467 U.S. at 808, 104 S.Ct. 2755. Although Varig does not use the phrase "separation of powers," the Court's explanation of the purpose behind the exception makes it clear that the exception is a statutory embodiment of separation-of-powers concerns:

Congress wished to prevent judicial `second-guessing' of legislative and administrative decisions grounded in social, economic, and political policy through the medium of an action in tort. By fashioning an exception for discretionary governmental functions, including regulatory activities, Congress took steps to protect the Government from liability that would seriously handicap efficient government operations.

Id. at 814, 104 S.Ct. 2755 (internal quotation marks omitted); see also Payton v. United States, 636 F.2d 132, 143 (5th Cir. 1981) ("The crux of the concept embodied in the discretionary function exception is that of the separation of powers."); Allen v. United States, 527 F.Supp. 476, 485 (D.Utah 1981) ("[T]he words `discretionary function' as used in the Tort Claims Act are really the correlative, the other side of the coin, of the exercise of executive power.").

When the purpose of the discretionary function exception in the FTCA is considered, it becomes apparent that the absence of such an exception in the SIAA is problematic, to say the least. For example, without a discretionary function exception, the government could be held liable for an initial decision to build a dam across a particular navigable waterway or to otherwise change the course of a navigable waterway. See Coates v. United States, 181 F.2d 816, 817 (8th Cir.1950) (concluding that plaintiffs' claim for damage to property caused by the government's decision to change the course of the Missouri River was barred by the discretionary function exception). The government could be held liable for the Coast Guard's drug-interdiction activities. See Mid-South Holding Co. v. United States, 225 F.3d 1201, 1206-07 (11th Cir.2000) (concluding that discretionary function exception precluded claim against government for damages to a private vessel that occurred during Coast Guard's search for drugs). The government could perhaps even be held liable for an inaccurate weather forecast. See Brown v. United States, 790 F.2d 199, 203-04 (1st Cir.1986) (concluding that discretionary function exception barred negligence claims brought by relatives of fishermen who drowned during a storm that the National Oceanic and Atmospheric Administration failed to predict).

As these examples illustrate, if the SIAA does not include a discretionary function exception, the executive branch's ability to "faithfully execute[]" the law, U.S. Const., art. II § 3, would be substantially impaired. As the Second Circuit has explained,

The wellspring of the discretionary function exception is the doctrine of separation of powers. Simply stated, principles of separation of powers mandate that the judiciary refrain from deciding questions consigned to the concurrent branches of the government. . . .

The doctrine of separation of powers is a doctrine to which the courts must adhere even in the absence of an explicit statutory command. Were we to find the discretionary function exception not to be applicable to the SAA, we would subject all administrative and legislative decisions concerning the public interest in maritime matters to independent judicial review in the not unlikely event that the implementation of those policy judgments were to cause private injuries. Such an outcome is intolerable under our constitutional system of separation of powers.

In re Asbestos Litig., 891 F.2d at 35 (citations and internal quotation marks omitted); see also Tiffany, 931 F.2d at 276 ("It is plain that the discretionary function exception to tort liability serves separation of powers principles by preventing judicial second-guessing of legislative and administrative decisions grounded in social, economic, and political policy through the medium of an action in tort." (alteration and internal quotation marks omitted)).

Moreover, if all executive-branch actions taking place in the maritime arena were subject to judicial review, the judiciary would be called upon to decide issues it is not equipped to resolve. We do not mean to suggest, of course, that judicial review is not the core responsibility of the judiciary, or that judicial review of all executive actions would impair the executive's obligation to faithfully execute the laws. But where the executive's discretionary functions are at issue, interference from the judicial branch is inappropriate. The discretionary function exception

articulate[s] a policy of preventing tort actions from becoming a vehicle for judicial interference with decisionmaking that is properly exercised by other branches of government. . . . Statutes, regulations, and discretionary functions, the subject matter of § 2680(a), are, as a rule, manifestations of policy judgments made by the political branches. In our tripartite governmental structure, the courts generally have no substantive part to play in such decisions. Rather the judiciary confines itself . . . to adjudication of facts based on discernible objective standards of law. In the context of tort actions, . . . these objective standards are notably lacking when the question is not negligence but social wisdom, not due care but political practicability, not reasonableness but economic expediency. Tort law simply furnishes an inadequate crucible for testing the merits of social, political, or economic decisions.

Blessing v. United States, 447 F.Supp. 1160, 1170 (E.D.Pa.1978) (Becker, J.) (footnote omitted).

Because our structural separation of powers is "a self-executing safeguard against the encroachment or aggrandizement of one branch at the expense of another," Buckley, 424 U.S. at 122, 96 S.Ct. 612, we must read the SIAA in a way that is consistent with those principles. See Limar Shipping, Ltd. v. United States, 324 F.3d 1, 7 (1st Cir.2003) (applying discretionary function exception to SIAA and explaining that the "[a]bsence of an express Congressional directive to the contrary will not be read as a green light for federal courts to assume power to review all administrative and legislative decisions concerning the public interest in maritime matters") (internal quotation marks omitted); Sea-Land Serv., 919 F.2d at 891 ("We understand Varig to teach that, as a matter of judicial construction, we should not read a general waiver of sovereign immunity to include a waiver of immunity with respect to damage occasioned by policy decisions. Accordingly, we hold that the SAA, which explicitly contains only a general waiver, also implicitly contains a discretionary function exception to its waiver of sovereign immunity."); In re Joint Asbestos Litig., 891 F.2d at 35 ("[W]e find the SAA to be subject to the discretionary function exception. This result is compelled by our steadfast refusal to assume powers that are vested in the concurrent branches."); Canadian Transport Co. v. United States, 663 F.2d 1081, 1086 (D.C.Cir.1980) (explaining that the discretionary function exception is "derived from the doctrine of separation of powers, a doctrine to which the courts must adhere even in the absence of an explicit statutory command.. . . Our recognition of a discretionary function exception in the Suits in Admiralty Act, therefore, is not an attempt to rewrite the statute, but merely an acknowledgment of the limits of judicial power."). Accordingly, like the other circuits to have considered the question, we now conclude that separation-of-powers principles require us to read into the SIAA's waiver of sovereign immunity a discretionary function exception.5

We pause to note, however, that this conclusion is not as obvious as the cursory analysis of some opinions from other circuits might suggest. We have just explained our belief that the discretionary function exception embodies separation-of-powers principles that are important enough to require courts to apply a discretionary function exception to statutes that are silent on the issue. Under that analysis, then, one would expect to find cases from the early days of the SIAA and PVA where the courts refused to impose liability on the government for its conduct of discretionary functions. After all, judicial recognition of the inherent constraints of our constitutional structure is hardly new. The early case law, however, is more equivocal on this score than might be expected.

Certainly there are some early cases where courts using separation-of-powers-like language have questioned the wisdom of holding the government liable for the actions at issue in those cases. For example, in Mandel v. United States, 191 F.2d 164 (3rd Cir.1951), aff'd sub nom. Johansen v. United States, 343 U.S. 427, 72 S.Ct. 849, 96 L.Ed. 1051 (1952), the court considered a claim filed under the PVA by the estate of a civilian employee killed when the vessel to which he was assigned hit a mine in an Italian port during World War II. After concluding that the estate's sole remedy was under the Federal Employees Compensation Act, see id. at 166, the court stated that

it would not be in the public interest to have judicial review of the question of negligence in the conduct of military, or semi-military, operations. The operation of ships or land forces in the presence of the enemy is a matter where judgments frequently have to be made quickly and where judgments so made by commanding officers must have prompt and immediate response. It will not, we think, aid in the operation of the armed forces if the propriety of a commander's judgment is to be tested months or years afterwards by a court or a court and jury. What, in the light of subsequent events, may appear to be a lack of caution may have been the very thing necessary, or apparently necessary, at the time the action was taken.

. . .

. . . . No judge has it as part of his task to act as an intelligence officer for the armed forces. He cannot tell how the facts developed out of one incident, seemingly isolated and unimportant, may fit into a larger picture worked upon by an active and skilled hostile espionage system. We do not think he should be called upon to pass upon such a question.

Id. at 167, 168.

The Third Circuit articulated a similar analysis in P. Dougherty Co. v. United States, 207 F.2d 626 (3rd Cir.1953) (en banc). In Dougherty, a private barge was damaged in a collision, and a Coast Guard cutter was sent to tow the barge to a harbor. Problems arose as the cutter approached the harbor with the barge in tow, and the cutter was eventually forced to cut the towing hawser. Adrift, the barge pounded against a breakwater for almost an hour and suffered substantial damage. The Third Circuit determined, for several reasons, that while the Coast Guard was negligent, that negligence did not give rise to liability on the part of the government. The court went on to say that public policy prevents the government from being held liable under the PVA "for fault of the Coast Guard in the conduct of a rescue operation at sea." Id. at 634. The court explained:

There are two arrows in the quiver of this public policy. The first may be directed to the inevitable consequence on the morale and effectiveness of the Coast Guard if the conduct of its officers and personnel in the field of rescue operations under the indescribable strains, hazards and crises which attend them, is to be scrutinized, weighed in delicate balance and adjudicated by Monday-morning judicial quarterbacks functioning in an atmosphere of Serenity and deliberation far from the madding crowd of tensions, immediacy and compulsions which confront the doers and not the reviewers.

. . . . A judicial determination that officers or men of the Coast Guard have been negligent in rescue operations would inevitably have a concomitant effect upon their service records. Aware of that fact, the instinct of self-preservation would inevitably function even under the pressures of life or death crises which so often arise in rescue operations when members of the Coast Guard are called upon to make decisions. If men are to be brought to an abrupt halt in the midst of crisis — to think first that if they err in their performance they may expose their Government to financial loss and themselves to disciplinary measures or loss of existing status, and then to pause and deliberate and weigh the chances of success or failure in alternate rescue procedures, the delay may often prove fatal to the distressed who urgently require their immediate aid. Thus would the point of the second arrow in the quiver of public policy be blunted the arrow which is directed to preserve in the public interest our merchant marine and that of other nations with which we trade.

History establishes that tragic losses in men and ships all too frequently attend disasters at sea, and too often is it impossible to give successful succor despite the most gallant and efficient of efforts. To expose the men in the Coast Guard to the double jeopardy of possible loss of their own lives, and loss of status in their chosen careers, because they failed, in coping with the intrinsic perils of navigation, to select the most desirable of available procedures, or their skill was not equal to the occasion, is unthinkable and against the public interest.

Id. at 634-35 (footnotes omitted).

Thus, both Mandel and Dougherty seem to apply what amounts to a discretionary function exception.6 However, there is little indication in these opinions that the courts were considering the broader question of whether it is ever appropriate for a court to hold the government liable for its discretionary functions.

On the other hand, there are cases decided under the pre-1960 SIAA or PVA where the government's potential liability was determined by a straightforward application of common law principles, without mention of a discretionary function exception, even though the circumstances of the cases would seem to at least warrant a discussion of the possibility of such an exception. For example, in McAllister v. United States, 348 U.S. 19, 75 S.Ct. 6, 99 L.Ed. 20 (1954), the Supreme Court concluded that the government was liable under the SIAA to a seaman who contracted polio while serving on a ship located in Chinese waters in the fall of 1945, shortly after the World-War-II surrender of Japan. The government knew that polio was then prevalent in Shanghai, and the Court concluded that the government's decision to allow Chinese soldiers and stevedores from Shanghai "to have the run of the ship and use of its facilities" supported the district court's determination that the government was negligent. Id. at 21, 75 S.Ct. 6. There was no consideration of the possibility that the government's actions might fall within a discretionary function exception to the SIAA's waiver of sovereign immunity.

Likewise, in Canadian Aviator, Ltd. v. United States, 324 U.S. 215, 65 S.Ct. 639, 89 L.Ed. 901 (1945), the crew of a Navy patrol boat during World War II instructed a private vessel that the patrol boat would escort it into the Delaware Bay. As the private vessel followed the patrol boat directly astern, as ordered by the Navy crew, the private vessel struck a submerged wreck and was damaged. The owner of the private vessel sued the government under the PVA, arguing that the collision was caused by the negligence of the Navy crew. The issue before the Supreme Court was whether the PVA extended to claims where the damage was caused by the crew of a public vessel rather than by the public vessel itself. The Supreme Court concluded that the PVA did extend to such claims, and the Court therefore vacated the decision of the appellate court and remanded for further proceedings. See id. at 224-25, 65 S.Ct. 639. The factual setting of Canadian Aviator — the actions of a Navy patrol boat during war — would seem to make the case a good candidate for consideration of a discretionary function exception. But there is no indication in the opinions of the district court, the court of appeals, or the Supreme Court that the possibility of such an exception was ever suggested by the government or considered by the courts.

Likewise, there are several older circuit-court cases where the factual setting would seem to warrant consideration of a discretionary function exception, yet the opinions are silent in that regard. See United States v. The S.S. Washington, 241 F.2d 819, 821 (4th Cir.1957) (concluding that government and private vessel were both at fault for collision between private vessel and Navy destroyer that was returning to its position in a flotilla of Navy vessels after completing a mission to pick up soldiers; possibility of a discretionary function exception was not discussed); Pacific-Atlantic S.S. Co. v. United States, 175 F.2d 632 (4th Cir.1949) (concluding that government was not at fault in collision between private vessel and Navy battleship which, "[d]ue to war conditions," was zig-zagging and operating without lights; court did not consider the possibility of a discretionary function exception); United States v. The Australia Star, 172 F.2d 472, 476 (2nd Cir.1949) (concluding that, with regard to 1944 collision between the S.S. Hindoo and the S.S. Australian Star, the United States naval vessel that was escorting the Hindoo was partially responsible for the collision; no discussion of a possible discretionary function exception even though duty of commander of Navy escort was "to do what . . . would safeguard the Hindoo" and the commander "had authority to give an emergency war time order to any Allied merchant vessel").

The ambiguity of the pre-1960 cases with regard to a broad discretionary function exception to the SIAA or PVA could indicate that, at least in the view of early courts, the exception was not warranted in maritime cases. But since some admiralty actions were cognizable under the FTCA before the 1960 SIAA amendments, it is clear that a discretionary function exception is not per se inappropriate or unwarranted in the maritime arena. Alternatively, the relative silence of the pre-1960 cases with regard to a discretionary function exception under the SIAA or PVA could indicate that there is something about those particular claims that warrants different treatment. That is, it could be that the nature of claims that were never cognizable under the FTCA and have always been cognizable only under the SIAA or the PVA is such that it would rarely be appropriate to apply a discretionary function exception as to those cases. If that were the case, then perhaps the discretionary function exception that we believe must be read into the SIAA should be limited to those claims that, before the 1960 amendments, were cognizable only under the FTCA. Claims that could have been brought under the SIAA before it was amended would not be subject to the exception.

We cannot, however, conceive of a difference in the nature of the claims that were cognizable under the FTCA and those that were cognizable only under the SIAA or the PVA that is substantial enough to warrant such a result. Broadly speaking, the PVA and pre-1960 SIAA applied to claims involving government vessels, public or merchant. Thus, the cases that were cognizable under the FTCA were those cases where the injury was not caused by a government vessel or its crew yet still fell within the admiralty jurisdiction of the federal courts. If anything, it seems that cases involving public or merchant vessels are more likely to involve the executive's discretionary functions into which the judiciary should not intrude.

More importantly, however, any such limitation on the application of the discretionary function exception would give too much effect to the ambiguity of pre-1960 case law. While it may be that the preamendments courts did not consider the possibility of some form of a discretionary function exception because such an exception was not viewed as appropriate, the failure to address the issue could just as easily be attributed to some other factor — for example, the government's failure to press the issue. In any event, we do not believe that we can subordinate compelling separation-of-powers concerns to the ambiguous silence of the early SIAA and PVA cases. Accordingly, we conclude that a discretionary function exception applies to all cases brought under the SIAA, without regard to whether the claims, prior to the 1960 amendments, would have been brought under the FTCA or the SIAA.7

We now turn to one final point with regard to the scope of the SIAA's discretionary function exception. At oral argument, counsel for the plaintiffs suggested that the exception as it has developed under the FTCA has strayed beyond that which is required by separation-of-powers principles. That is, counsel contends that courts have applied the exception to exonerate the government from cases where an imposition of liability would not have been inconsistent with separation-of-powers principles. Thus, counsel argues that this court should not read into the SIAA an exception that is co-extensive with the exception as applied under the FTCA, but that we should instead hold that the SIAA is subject to a discretionary function exception only and to the precise extent necessary to serve the principles of separation of powers. If imposition of liability in any given case would not be offensive to separation-of-powers principles, then the government's conduct should not be excused, even if case law developed under the FTCA would characterize the government's action as falling within the discretionary function exception.8

At bottom, counsel's argument reflects a concern that the discretionary function exception has been applied too broadly under the FTCA and that it will likewise be applied too broadly under the SIAA. While it may be that courts have in some instances applied the FTCA's discretionary function exception more broadly than Congress intended or than might be strictly required under separation-of-powers principles, these occasional judicial errors do not warrant a wholesale retreat from a body of law that has been developed and refined over the course of almost fifty years. As we have explained, separation-of-powers principles require us to read a discretionary function exception into the SIAA, and it was those same separation-of-powers concerns that drove Congress to create the discretionary function exception to the FTCA. See Varig Airlines, 467 U.S. at 808, 104 S.Ct. 2755 (explaining that the discretionary function exception "marks the boundary between Congress' willingness to impose tort liability upon the United States and its desire to protect certain governmental activities from exposure to suit by private individuals"). In our view, the discretionary function exception as it has been developed and applied under the FTCA is the best embodiment of those separation-of-powers concerns, and we believe that it is therefore appropriate for FTCA cases to guide the application of the exception under the SIAA.

To summarize, we adhere to the rule previously applied in this circuit that requires a panel of this court faced with conflicting panel opinions to follow the earlier case. On the merits of the issue raised by the government, we conclude that, although the statute is silent on the issue, the SIAA must be read to include a discretionary function exception to its waiver of sovereign immunity. We hereby overrule Lane v. United States, 529 F.2d 175 (4th Cir.1975), to the extent that it concluded that a discretionary function exception does not apply to cases brought under the SIAA. Because the discretionary function exception under the FTCA and the exception that we apply today to the SIAA are both grounded in concerns of separation of powers, the scope of the discretionary function exception under the SIAA should mirror that of the FTCA, and discretionary function cases decided under the FTCA should guide decisions under the SIAA. The district court in this case concluded as a legal matter that a discretionary function exception was not available to the government in this case, and the court therefore did not consider whether the facts of the case warranted application of such an exception. Accordingly, we vacate the order of the district court and remand to give the district court the opportunity to decide in the first instance whether the discretionary function exception we recognize today precludes the plaintiffs' claims and to conduct any other proceedings that might become necessary.9

I am happy to concur in Judge Traxler's fine opinion in this case. The court holds today that the Suits in Admiralty Act (SIAA), 46 U.S.C.A. §§ 741-52 (West 1975 & Supp.2003), is subject to an exception similar to the discretionary function exception embodied in the Federal Tort Claims Act (FTCA), 28 U.S.C.A. §§ 2671-80 (West 1994 & Supp.2003). I write simply to state my view that any different result would not be supportable.

Tort liability will certainly lie against the United States under the SIAA. After all, that is the point of any waiver of immunity in the first place. It may well be that the discretionary function exception is inapplicable on the facts of this case. But appellants ask us to go much, much farther — to indulge in effect the broadest possible waiver of sovereign immunity for the performance of every discretionary governmental function and to disregard the principle that such waivers of immunity must be narrowly construed. See Lane v. Pena, 518 U.S. 187, 192, 116 S.Ct. 2092, 135 L.Ed.2d 486 (1996); United States Dep't of Energy v. Ohio, 503 U.S. 607, 615, 112 S.Ct. 1627, 118 L.Ed.2d 255 (1992); McMahon v. United States, 342 U.S. 25, 27, 72 S.Ct. 17, 96 L.Ed. 26 (1951). When we address the scope of a sovereign immunity waiver, the Supreme Court requires us to take a cautious approach, not a sweeping and momentous one.

The majority's action is not one of impermissible judicial implication. There is nothing implicit about the separation-of-powers concerns that underlie the discretionary function exception, concerns that we are obliged to honor even in the absence of a statutory directive. See, e.g., Robinson v. United States (In re Joint E. & S. Dists. Asbestos Litig.), 891 F.2d 31, 35 (2d Cir.1989); Canadian Transport Co. v. United States, 663 F.2d 1081, 1086 (D.C.Cir.1980). The executive has an explicit, not an implicit, duty to see that the laws are faithfully executed. See U.S. Const. art. II, § 3. This duty cannot be discharged without the exercise of some discretion. For the court to subject every such discretionary act to the prospect of tort liability would not only be to undercut an explicit constitutional command; it would wrongly assign to Congress the desire to debilitate the executive branch. The discretionary function exception is thus implied only in the same sense that the doctrine of qualified immunity is implied in the interpretation of 42 U.S.C. § 1983 — on the premise that without the ability to exercise some element of judgment in the execution of law, neither federal, state, nor local government could function.1

The language that this court applies today is not language that the judiciary has somehow made up on its own. Rather, the Court adopts Congress's own explicit expression of separation-of-powers principles in 28 U.S.C. § 2680(a). See United States v. S.A. Empresa de Viacao Aerea Rio Grandense (Varig Airlines), 467 U.S. 797, 814, 104 S.Ct. 2755, 81 L.Ed.2d 660 (1984) (recognizing that FTCA's discretionary function exception was Congress's attempt "to prevent judicial `second-guessing' of legislative and administrative decisions grounded in social, economic, and political policy"). The fact that we have at hand such a carefully crafted expression of separation-of-powers principles from a coordinate branch of government rebuts any suggestion that the court is somehow on a statutory frolic of its own.

The dissent attempts to say that Varig Airlines, and by extension the discretionary function exception itself, is not an expression of separation-of-powers principles. See dissenting op. at 370. Even in our modern age, however, some things are indeed what they seem. The discretionary function exception expresses Congress' view of that degree of "separation" required by the executive branch to carry out its duties. It further underscores the need for judicial forbearance in the face of policy-laden decisions made by the coordinate branches of our government. These are classic separation-of-powers concerns. Not surprisingly, then, our court and our sister circuits have interpreted Varig Airlines to embody separation-of-powers principles. See, e.g., Tiffany v. United States, 931 F.2d 271, 276 (4th Cir.1991); Irving v. United States, 162 F.3d 154, 160 n. 4 (1st Cir.1998); In re Joint E. & S. Dists. Asbestos Litig., 891 F.2d at 35; Wiggins v. United States, 799 F.2d 962, 965-66 (5th Cir.1986).

The majority's action thus heeds not only the executive's constitutional prerogatives, but Congress's respect for those prerogatives as well. It is difficult to imagine that Congress was seeking to eliminate executive branch discretion in the execution of what are, after all, Congress's own policy mandates and directives. See Berkovitz v. United States, 486 U.S. 531, 537, 108 S.Ct. 1954, 100 L.Ed.2d 531 (1988) (finding that discretionary function exception protects "governmental actions and decisions based on considerations of public policy"). A failure to recognize any discretionary function would allow the deterrent effect of tort liability in those very areas where Congress has mandated an active executive role. Shorn of a discretionary function exception, the executive branch would be profoundly impaired in carrying out the very functions that Congress has assigned to it.

Nor is it easy to imagine that the SIAA somehow sought to disable the executive branch from invoking separation-of-powers principles via its discretionary functions as a defense to unlimited tort liability. Without the defense, the United States would be subject not only to constitutional constraints, but under the SIAA to tort duties and negligence actions, for attempts to enforce immigration law; to intercept narcotics-smuggling; to protect its airspace from hostile, incoming aircraft; and to safeguard its harbors from biological agents in container cargo. "Were there no such immunity for basic policy making decisions, all administrative and legislative decisions concerning the public interest in maritime matters would be subject to independent judicial review in the not unlikely event that the implementation of those policy judgments were to cause private injuries." Gercey v. United States, 540 F.2d 536, 539 (1st Cir.1976).

The dissent speaks so derisively of the discretionary function exception that perhaps it would find no impediment to judicial consideration of the countless policy judgments attendant to armed warfare at sea. After all, ascertainable standards for the resolution of such questions may well exist. See dissenting op. at 372-73. In In re Joint E. & S. Dists. Asbestos Litig., the Second Circuit in 1989 faced a challenge under the SIAA to President Roosevelt's use of asbestos in the construction of ships for the merchant marine in World War II. 891 F.2d at 33-34. Dismissing the suit under the discretionary function exception, the court noted: The fact that the challenged actions were matters of choice cannot be overcome by clothing the discretionary acts in the maritime uniform of a breach of a duty to provide a seaworthy vessel. We are unwilling to declare that during a world war, when ships were being sunk by the enemy as fast as they could be constructed, it was impermissible for the government to choose to deploy ships in less than seaworthy condition. We need spend little time discussing whether the contested choices involved considerations of public policy. It is difficult to imagine a clearer example of a decision grounded in social, economic, and political policy than the choice of how to prosecute a world war.

Id. at 37 (internal quotations and citation omitted).

Toward the end of his opinion, my dissenting colleague attempts to crawl back from the far limb. "I agree," he says, "that the SIAA may well authorize some suits that call upon the courts to make political judgments that they are neither prepared to make, nor capable of competently making." Dissenting op. at 373. But while the dissent professes to recognize the same concerns as the majority, its answer is to jettison congressional language tailored to this very context—governmental tort liability—in favor of the all-purposive political question doctrine. The dissent thus commits, in even more serious fashion, the same sin that it seeks to ascribe to the majority. It concedes that a blanket waiver of immunity would run headlong into a constitutional problem, but it refuses to respect Congress's solution to that problem. The dissent's substitution of a judicially-derived doctrine for congressionally-crafted language makes for a truly anomalous statutory scheme: executive officials could be liable for discretionary functions, but only in admiralty.

The dissent thus advocates a free-floating separation-of-powers approach, which might be necessary in a context where courts had no other alternative. Here, however, there is most certainly an alternative: Congress's adoption of a discretionary function test in the FTCA, combined with Congress's refusal to disturb many decades of unanimous judicial interpretation relying on that same congressionally-grounded test in the SIAA. See majority op. at 348 n.7. To upend this settled scheme serves no purpose whatsoever, particularly when no practical reason for differentiating between the FTCA and the SIAA has ever been advanced. While my dissenting colleague may make all sorts of assumptions about whether I would or would not dismiss this case under the discretionary function exception, the point is that the answer lies in the scope of the exception to governmental tort liability, not in a generalized application of the political question doctrine.

In the end, it makes sense to reflect on the sheer enormity of what appellants and our dissenting colleagues ask the court to do. They would have us strip the government of a discretionary function defense in the face of the considered wisdom of ten other circuit courts of appeals, each of which has held the discretionary function exception applicable to the SIAA. See majority op. at 338. They would do so in a context where sovereign prerogatives can be salient and where uniformity of interpretation would seem an imperative. They would have us adopt discordant approaches to two companion acts. And they would fasten upon the government the broadest conceivable waiver of its own separation-of-powers defenses, in derogation of the principle that sweeping waivers of immunity should not be casually assumed. See Dep't of the Army v. Blue Fox, Inc., 525 U.S. 255, 261, 119 S.Ct. 687, 142 L.Ed.2d 718 (1999). Whether the discretionary function exception applies on these facts is rightly left to the district court upon remand. But our failure to recognize any exception whatsoever would set us on an aggressive course far afield of judicial competence and replete with matters of policy entrusted elsewhere. The court's opinion rightly rejects that option.2