concurring in part and dissenting in part:
The majority follows the D.C. Circuit by holding that the immunity for “governmental functions” conferred by section 80 of the METRO Compact is at least as broad as that conferred by the discretionary function exception, 28 U.S.C. § 2680(a), to the Federal Tort Claims Act. As the D.C. Circuit explained in Burkhart v. WMATA, 112 F.3d 1207, 1216-17 (D.C.Cir.1997), this means that the METRO is immune from liability for any acts and omissions that are “susceptible to policy analysis” under the standard established by the Supreme Court in United States v. Gaubert, 499 U.S. 315, 325, 111 S.Ct. 1267, 113 L.Ed.2d 335 (1991). Applying the Gaubert standard, the majority concludes that the METRO is entitled to immunity on all but one of the theories of negligence advanced by Richard H. Smith’s family and estate. Most strikingly, the majority holds in part III B of its opinion that the METRO is immune from liability for its decision not to reassemble Escalator Three because that decision could have been based on a judgment that the costs of reassembling the escalator outweighed the safety benefits of reassembly. Because I cannot believe that the framers of the METRO Compact intended for the governmental function exception in section 80 to sweep so broadly, I must respectfully dissent from the majority’s analysis.
The majority suggests that under Gau-bert any decision that can be characterized as an effort to cut costs is “the kind of policy judgment that the discretionary function exception was designed to shield.” Id. at 332, 111 S.Ct. 1267. Remarkably, this extremely broad reading of the discretionary function exception appears to be consistent with our prior cases. Although there, is something commendable in- the majority’s refusal to flinch from Gaubert’s implications, the majority’s analysis makes it disturbingly clear that the discretionary function exception has in fact swallowed up the FTCA’s ostensibly broad waiver of sovereign immunity. This means that the majority’s initial decision to follow the D.C. Circuit comes at a heavy price: the price of maintaining consistency between our circuit and-the D.C. Circuit is nothing less than evisceration of the- waiver of sovereign immunity in section 80 of the METRO Compact. Because I think that price is too high, I dissent from the .majority’s decision in parts II B and C of its opinion to analyze the METRO’S immunity by using Gaubert’s susceptible-to-policy-analysis standard. I would adopt a narrower view of the immunity conferred by section 80 of the METRO Compact by reading the governmental function exception along the lines suggested by Justice Scalia in his concurring opinion in Gaubert. Under that standard the METRO (as the district court concluded) has not established its immunity on any of the theories of negligence that survived the district court’s summary, judgment ruling. It follows that I also dissent from parts III A, B, and C of the majority opinion. I concur in the remainder of the opinion.
*213I will first explain why the majority’s analysis allows the exception for governmental functions in section 80 of the METRO Compact to swallow up the Compact’s general waiver of sovereign immunity. I then explain why the better course would be to analyze the governmental function exception using the standard suggested by Justice Scalia in his Gaubert concurrence. Finally, I explain how I would apply that standard to the claims before us.
I.
Section 80 of the METRO Compact provides that the METRO “shall be liable for its contracts and for its torts and those of its directors, officers, employees and agents committed in the conduct of any proprietary function ... but shall not be liable for any torts occurring in the performance of a governmental function.” As the majority explains, ante at 207, the D.C. Circuit has read section 80 to confer immunity on the METRO for all its “quintessential governmental funetion[s],” such as police activity. Burkhart, 112 F.3d at 1216 (internal quotation marks and citation omitted). Where, as here, the allegedly tortious conduct did not occur in the METRO’S performance of a quintessentially governmental function, the D.C. Circuit looks to the case law interpreting the discretionary function exception in the Federal Tort Claims Act (FTCA) to decide whether the conduct occurred in the performance of a governmental function or a proprietary function. Id. Following the D.C. Circuit, the majority today holds that “all ‘discretionary’ activities of a governmental entity under the FTCA constitute ‘governmental’ activities within the meaning of the ‘governmental/proprietary’ test.” Ante at 206. To understand the consequences of this decision, it will be necessary to review briefly the Supreme Court’s recent decisions interpreting the discretionary function exception.
The Supreme Court’s decision in Berkovitz v. United States, 486 U.S. 531, 108 S.Ct. 1954, 100 L.Ed.2d 531 (1988), created a two-pronged test for applying the discretionary function exception. First, courts ask whether the governmental action complained of “involves an element of judgment or choice.” Id. at 536, 108 S.Ct. 1954. When a statute, regulation, or policy prescribes a specific course of action, the negligent failure to follow that course is not protected by the discretionary function exception. Id. Second, if the challenged conduct is discretionary, courts then ask whether the judgment involved is “of the kind that the discretionary function exception was designed to shield.” Id. The exception protects only decisions “grounded in social, economic, [or] political policy.” Id. at 537, 108 S.Ct. 1954 (internal quotation marks and citation omitted).
Because most government decisions involve at least some degree of choice, courts applying the Berkovitz test have frequently had to decide whether a government decision was sufficiently policy based to deserve the protection of the discretionary function exception. Prior to the Court’s 1991 decision in Gaubert, the case law interpreting the discretionary function exception suggested several possible limits on the range of government conduct that could be seen as grounded in social, economic, or political policy. See, e.g., Gaubert v. United States, 885 F.2d 1284, 1289 (5th Cir.1989) (suggesting that decisions made at the operational rather than the planning level do not qualify as policy based under the second prong of Berkovitz), rev’d sub nom. United States v. Gaubert, 499 U.S. 315, 111 S.Ct. 1267, 113 L.Ed.2d 335 (1991); Dube v. Pittsburgh Coming, 870 F.2d 790, 797-800 (1st Cir.1989) (holding that the discretionary function exception does not apply when the *214government fails to make an actual policy judgment), abrogation recognized, by Shansky v. United States, 164 F.3d 688, 692 n. 4 (1st Cir.1999); Blessing v. United States, 447 F.Supp. 1160, 1184 (E.D.Pa.1978) (stating that the discretionary function exception should apply only if the official who made a challenged decision had the authority to make that decision). Gaubert, however, rejected these limitations on the scope of the discretionary function exception. First, Gaubert made clear that an operational decision by a relatively low-level employee can be grounded in social, economic, or political policy for purposes of the discretionary function exception. Gaubert, 499 U.S. at 325-26, 111 S.Ct. 1267. Second, Gaubert held that the government may enjoy the protection of the discretionary function exception without any showing that its employees actually considered policy goals in making the decisions alleged to be negligent: “The focus of the inquiry is not on the agent’s subjective intent in exercising the discretion conferred by statute or regulation, but on the nature of the actions taken and on whether they are susceptible to policy analysis.” Id. at 325, 111 S.Ct. 1267. See also Baum v. United States, 986 F.2d 716, 721 (4th Cir.1993). As the First Circuit has explained, the critical question after Gaubert is “whether some plausible policy justification could have un-dergirded the challenged conduct.” Shansky v. United States, 164 F.3d 688, 692 (1st Cir.1999) (emphasis added). Finally, the Gaubert majority apparently held that a governmental decision can be susceptible to policy analysis (and thus within the scope of the discretionary function exception) even if the decision was made by an employee who lacked the authority to consider matters of social, economic, or political policy.1
Gaubert’s susceptible-to-policy analysis standard has significantly broadened the scope of the discretionary function exception. See Bruce A. Peterson and Mark E. Van Der Weide, Susceptible to Faulty Analysis: United States v. Gaubert and the Resurrection of Federal Sovereign Immunity, 72 Notre Dame L.Rev. 447, 465-73 (1997) (surveying the post-Gaubert case law and concluding that the government wins more cases, and wins them at earlier stages of litigation, after Gaubert than before). Because the Gaubert standard asks only whether allegedly negligent government conduct could have been based on policy judgments, courts have had difficulty in placing any principled limits on the range of conduct that counts as grounded in social, economic, or political policy.2 *215See Rosebush v. United States, 119 F.3d 438, 444-45 (6th Cir.1997) (Merritt, J., dissenting) (stating that the second prong of the Berkovitz-Gaubert test “presents an ambiguous standard that is difficult to apply and that has produced a large number of inconsistent holdings in the circuit and district courts”). One fertile ground of dispute has been the question of whether decisions that can be seen as responsive to agency budget constraints should count as “grounded in economic policy” for purposes of the discretionary function exception. See Domme v. United States, 61 F.3d 787, 793-94 (10th Cir.1995) (Henry, J., concurring) (recognizing disagreement in the case law about whether government decisions should be treated as policy based simply because they implicate economic considerations). Several judges have voiced concern that if government choices are immune simply because those choices could have been motivated by a desire to cut costs, the discretionary function exception will gut the FTCA’s waiver of sovereign immunity. See, e.g., Cope v. Scott, 45 F.3d 445, 449 (D.C.Cir.1995) (observing that if any decision responsive to budgetary constraints counts as grounded in economic policy, the discretionary function exception would “swallow the FTCA’s sweeping waiver of sovereign immunity”); Domme, 61 F.3d at 795 (Henry, J., concurring) (observing that “unless courts routinely inquire as to whether government employees have the authority to make the sorts of decisions that result in injuring people, the fact that it is reasonable to view most government decisions as having economic implications could eviscerate the Federal Tort Claims Act”).
Today, the majority squarely aligns this circuit with those courts that have been willing to treat decisions that could have been motivated by a desire to reduce costs as grounded in economic policy under Gaubert and Berkovitz. In part III B of its opinion, the majority holds that the discretionary function • exception protects the METRO’S decision not to reassemble Escalator Three for use as a walker. According to the majority, this decision involved the kind of discretion the discretionary function exception was meant to shield because it “implicated the economic policy of the METRO, i.e., whether it was more cost-effective to reassemble Escalator Three pending repair, or whether to wait until replacement parts arrived.” Ante at 209. This, to me, is a remarkable result. As any first-year law student knows, the basic approach to negligence law outlined by Judge Learned Hand in United States v. Carroll Towing Co., 159 F.2d 169 (2d Cir.1947), essentially defines negligence as the unreasonable balancing of the cost of safety measures against the risk of accidents. See id. at 173 (explaining that “if the probability [of an accident] be called P; the injury, L; and the burden [of adequate precautions], B; liability depends upon whether B is less than L multiplied by P: i.e., whether B<PL”). Even on the most charitable interpretation of the facts, this is exactly the sort of balancing that was going on when the METRO decided not to reassemble the steps on Escalator Three. The decision that the risks of not reassembling Escalator Three were outweighed by the cost of reassembly involved “the same kind of balancing which citizens do at their peril,” Dalehite v. United States, 346 U.S. 15, 60, 73 S.Ct. 956, 97 L.Ed. 1427 (1953) (Jackson, J., dissenting), and which courts review every day using ordinary tort law standards, cf. Blessing, 447 F.Supp. at 1183 n. 30 (ex*216plaining that the discretionary function exception should apply to the sorts of policy judgments that cannot be reviewed under ordinary tort law standards of due care and reasonableness). ' If this sort of balancing is immune from judicial review, the discretionary function exception has eclipsed the FTCA’s basic goal of making-the government liable for tort claims “under circumstances where the United States, if a private person, would be liable to the claimant in accordance with the law of the place where the [tortious] act or omission occurred.” 28 U.S.C. § 1346(b)(1).
The majority’s holding in part III B is all the more striking because, consistent with Gaubert, the majority pays no attention to the actual processes that led to the decision not to reassemble Escalator Three. The summary judgment record contains no evidence that anyone consciously weighed the costs in time and labor of reassembling Escalator Three versus the. benefits in increased passenger safety and convenience. The record does not indicate who made the decision. Indeed, the record does not even indicate who had the authority to make the decision. Viewing the evidence in the light most favorable to Smith’s family and estate, a reasonable jury could conclude that the decision not , to reassemble Escalator Three was made by the escalator repair crew without any significant consultation with supervisory officials and that the repair crew neither sought to balance competing policy considerations nor had the authority to do so. Cf. Domme, 61 F.3d at 795 (Henry, J., concurring) (“[A] nonman-agerial employee who, away from the quiet and measured reflection of a budget meeting, decides to compromise safety in order to save government resources may simply be trying to make his or her job easier rather than trying to serve the common good.”). If the decision not to reassemble Escalator Three counts as a judgment grounded in the economic policy of the METRO, the discretionary function exception truly has swallowed up the FTCA’s waiver of immunity because there are precious few governmental decisions that cannot be seen as an effort to save costs. See Cope, 45 F.3d at 449 (noting that after government counsel had argued that any decision implicating budgetary concerns should fall within the discretionary function exception, counsel failed during oral argument to provide “even one example of a discretionary decision that would not be exempt for failure to implicate policy concerns” aside from the standard example that the discretionary function exception does not cover the negligent operation of motor vehicles by government employees).3 Consequently, the upshot of the majority’s approach to the discretionary function exception is that the government will nearly always be immune for its actions so long as it has not enacted regulations that completely eliminate the discretion of its employees.
All that said, I cannot say that the majority’s analysis in part III B is an unreasonable, or even an incorrect, application of Gaubert’s susceptible-to-policy-analysis standard. Although some courts have ex*217pressed misgivings about an overly broad interpretation of -this standard, no court has held that a judgment based on the desire to cut costs can never qualify as “grounded in economic policy” for purposes of the discretionary function exception analysis. Indeed, such a rule would produce its own set of problems because in many instances government decisions based on cost considerations ought to enjoy the protection of the discretionary function exception. See, e.g., United States v. Varig Airlines, 467 U.S. 797, 820, 104 S.Ct. 2755, 81 L.Ed.2d 660 (1984) (explaining that decisions requiring an agency “to establish priorities for the accomplishment of its policy objectives by balancing the objectives sought to be obtained against such practical considerations as staffing and funding” are protected by the discretionary function exception); Cope, 45 F.3d at 451 (explaining that the government’s decisions about whether and when to improve the degree of skid resistance on a section of park road were protected by the discretionary function exception because they involved considerations including “the allocation of funds among significant project demands, the safety of drivers and other park visitors, and the inconvenience of repairs as compared to the risk of safety hazards”). Further, our circuit’s case law on the discretionary function exception has signaled our willingness to consider decisions that implicate budgetary concerns as sufficiently policy based to receive immunity. See Williams v. United States, 50 F.3d 299, 310 (4th Cir.1995) (“Contracting out the responsibility to maintain [office space leased by the United States] while balancing fiscal considerations entails exercising judgment based on policy.”); Bowman v. United States, 820 F.2d 1393, 1395 (4th Cir.1987) (characterizing a decision based on “a lack of financial- resources” as a “policy judgment”). For these reasons, I must conclude that, even in part III B, the majority has correctly applied our circuit’s discretionary .function ■ exception precedents to the facts of this case. It follows that if this case was about the meaning of the discretionary function exception, I would be bound to concur in the result reached by the majority. Here, however, we are asked to interpret the METRO Compact, and we are not required' to follow the Gaubert standard wherever it might lead. I suggest that if a correct application of the Gaubert standard leads to the result in part III B, this only highlights the shortcomings of that standard and provides a significant reason to consider using a different -standard to interpret the METRO Compact.
As the majority’s analysis illustrates, the Gaubert approach tends to push courts toward holding that judgments involving tradeoffs between cost and safety are grounded in economic policy for purposes of the discretionary function exception. Although such holdings threaten to eviscerate the FTCA, the alternative of claiming that decisions based on budget constraints and the like are never policy based is equally unpalatable. The more reasonable view is that sometimes government decisions requiring tradeoffs between cost and safety deserve the protection of the discretionary function exception, and sometimes they do not. For example, if an agency makes judgments about the cost-effectiveness of various safety measures and expresses those judgments in the form of a regulation, the discretionary function exception does (and should) protect the regulation (and any conduct it requires) from judicial second-guessing. As Justice Scalia points out, the FTCA’s exclusion of liability for all acts of government employees performed with due care “in the execution of a ... regulation, whether or not such ... regulation be *218valid,” 28 U.S.C. § 2680(a), “represents an absolute statutory presumption ... that all regulations involve policy judgments that must not be interfered with,” Gaubert, 499 U.S. at 336-37, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment). But as my discussion of the METRO’S decision not to reassemble Escalator Three illustrates, the thoughtless failure of low-level employees to take certain safety precautions ought not to be protected by the discretionary function .exception simply because the failure could be rationalized as an effort to save government dollars. These examples suggest that whether the discretionary function exception’s protection is warranted depends at least in part on the process by which government decisions are made. See id. at 335, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment in part) (suggesting that “the level at which the decision is made is often relevant to the discretionary function inquiry,' since the answer to that inquiry turns on both the subject matter [of the decision] and the office of the decisionmaker”); see also Peterson and Van Der Weide, supra, at 487-502 (suggesting that “[discretionary function immunity ought to be reserved for (1) actual decisions (2) made by government officials possessing authority to direct policy (3) in consideration of legitimate policy factors”); Harold J. Krent, Preserving Discretion Without Sacrificing Deterrence: Federal Government Liability in Tort, 38 UCLA L.Rev. 871, 915 (1991) (suggesting that courts should approach the discretionary function exception by “focusing] on the decision making process underlying the governmental action challenged”). By eliminating any consideration of the decisionmaker’s authority or the manner in which decisions were actually made, Gaubert leaves courts without the analytical tools needed to place sensible limits on the scope of the discretionary function exception. This means that we ought to think long and hard before deciding to follow the D.C. Circuit by exporting the shortcomings of the Gaubert standard into our law governing the scope of the METRO’S sovereign immunity.
II.
As I explained in part I, the majority’s disposition of this case is correct if Gau-bert provides the proper standard for interpreting section 80 of the METRO Compact. Before proceeding any further, however, it is worth pointing out that the majority’s disposition is correct only if Gaubert is the proper standard; any other approach to reading section 80 would require a different result. If, for example, we read section 80 literally and applied the distinction between governmental and proprietary activities, operating escalators as part of a public transportation system would surely be classified as proprietary and would therefore be unprotected by the METRO’S immunity. See Wainwright v. WMATA 958 F.Supp. 6, 9 (D.D.C.1997) (stating that the METRO’S operation of an escalator is a proprietary function); Warren v. WMATA, 880 F.Supp. 14, 16 (D.D.C.1995) (stating that the METRO’S “provision of mass transit is generally considered a proprietary function that would not be protected by sovereign immunity”). If we read section 80 as drawing a distinction between planning and operational activities — in other words, if we read section 80 in roughly the same way that many courts read the discretionary function exception prior to Gaubert—surely the actions of the METRO in this case fall on the operational side of the line. Cf. Dant v. District of Columbia, 829 F.2d 69, 74-75 (D.C.Cir.1987) (granting immunity to the METRO on claim that it negligently designed its fare collection system, but denying immunity on claim that the METRO *219negligently operated and maintained the fare collection system because negligent operation and maintenance are ministerial activities). Likewise, if we read section 80 along the lines suggested by Justice Sca-lia’s concurrence in Gaubert, we would again hold that the METRO was not entitled to summary judgment on its immunity claim because a reasonable jury could conclude that the allegedly negligent decisions in this case were not made by any person who had the authority to make judgments based on social, economic, or political policy. See Gaubert, 499 U.S. at 385, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment). In sum, only the Gaubert approach to governmental immunity is expansive enough to justify the result the majority reaches today. This point underscores the importance of asking whether there are good reasons for following the D.C. Circuit’s decision to use Gaubert to analyze the governmental function exception in section 80 of the METRO Compact. I submit that the D.C. Circuit’s approach to the governmental function exception is inconsistent with the intent of the Compact’s framers and that any pragmatic considerations supporting that circuit’s approach are not strong enough to justify it.
The D.C. Circuit first held in Sanders v. WMATA, 819 F.2d 1151 (D.C.Cir.1987), that the governmental function exception in the METRO Compact should be analogized to the discretionary function exception in the FTCA. In Sanders the court began its analysis by observing that the meaning of the governmental/proprietary distinction in section 80 is a question of federal law. See id. at 1154. The court then explained that because Congress had played an active role in the creation and approval of the METRO Compact, the decision to grant immunity for torts committed by METRO employees in the performance of governmental functions should be understood as a decision to adopt Congress’s understanding of “governmental function.” That understanding, the court reasoned, was embodied in the Supreme Court’s decision in Dalehite v. United States, 346 U.S. 15, 73 S.Ct. 956, 97 L.Ed. 1427 (1953), where the Supreme Court construed the discretionary function exception to the FTCA as an effort to free the government from “liability arising from acts of a governmental nature or function.” Id. at 28, 73 S.Ct. 956. Accordingly, the Sanders court concluded that the framers of the METRO Compact (Maryland, Virginia, and the District of Columbia) had “accepted the Dalehite conception” when they chose to grant the METRO immunity from tort liability for its performance of governmental functions. Sanders, 819 F.2d at 1155. The D.C. Circuit later extended this reasoning in Burkhart, holding that the Supreme Court’s analysis of the discretionary function exception in Gau-bert should also apply to the interpretation of “governmental function” in section 80 of the METRO Compact. Burkhart, 112 F.3d at 1216. The majority adopts the D.C. Circuit’s analysis in Sanders and Burkhart in parts II B and C of its opinion.
With-all respect to the majority and our sister circuit, J do not believe that the framers of the METRO Compact intended the governmental/proprietary distinction in section 80 to track the distinction between discretionary and ■ ministerial acts in § 2680(a) of the FTCA. To begin with the most obvious point, surely the Compact’s framers would have used the terms “discretionary” and “ministerial” in -section 80 if they had meant to signal their allegiance to Dalehite. Instead, the framers used the traditional language of municipal immunity. See 18 Eugene MeQuillin, The Law of Municipal Corporations § 53.02.05 (3d ed.1993) (explaining that the distinction be*220tween governmental and proprietary functions is the oldest approach to the law of municipal immunity); The Supreme Court’s decision in Indian Towing Co. v. United States, 350 U.S. 61, 76 S.Ct. 122, 100 L.Ed. 48 (1955), casts further doubt on the claim that the- Compact’s framers intended to adopt the understanding ,of governmental functions expressed in Dalehite. In Indian Towing the Supreme Court explicitly rejected the governmental/proprietary distinction as a proper framework for interpreting the FTCA. The Court characterized the law of municipal liability as a “quagmire” and the rule of law distinguishing “governmental” from “non-governmental” (or proprietary) functions as “inherently unsound.” Id. at 65, 76 S.Ct. 122. It explained that the FTCA is a repudiation of the “casuistries of municipal liability for torts.” Id. If the framers of the Compact can be presumed to have been aware of Dalehite, they must also be presumed to have been aware of Indian Towing. In light of Indian Towing’s insistence that the FTCA does not carry forward the doctrines of municipal liability, it is hard to believe that the framers of the Compact would have employed the governmental/proprietary distinction in section 80 in order to adopt the reading of the FTCA set forth in Dalehite. It makes far more sense to believe that section 80 was drafted as a compromise between the differing rules of governmental immunity then followed by the Compact’s signatories. Cf. Martin v. WMATA, 667 F.2d 435, 436 (4th Cir.1981).
Moreover, even if it was reasonable to believe that the framers of the METRO Compact meant to accept the Dalehite conception of governmental tort immunity, it would not follow that they meant for the governmental function exception in the METRO Compact to expand or contract with the Supreme Court’s rulings on the scope of the discretionary function exception. Dalehite was widely understood to employ a distinction between planning and operational activities, see Krent, supra, at 880, a distinction the Sanders court seemed to assume when it held that the framers of the METRO Compact had accepted the Dalehite conception. See Sanders, 819 F.2d at 1156 (distinguishing between “general attacks on the testing plan itself’ and attacks on the “manner of testing in a particular case” in holding that the METRO was immune from tort liability for the adoption of policy requiring drug and alcohol testing immediately after on-the-job accidents or unusual operating incidents). No one would have imagined in 1966 that the discretionary function would come to be read so broadly as the Supreme Court read it in Gaubert, and there is no reason to believe that the framers of the Compact would have intended that the governmental function exception in section 80 be given such a broad reading. Finally, the fact that the meaning of the governmental function exception is a matter of federal law does not indicate that the meaning of “governmental function” in the Compact should track the meaning of “discretionary function” in the FTCA. The meaning of each term may present a question of federal law, but it does not follow that each presents the same question of federal law. Consequently, I am unpersuaded by the D.C. Circuit’s arguments that its approach to the METRO Compact is consistent with the intent of the Compact’s framers.
Perhaps, though, the D.C. Circuit’s decision to use the Gaubert standard in interpreting the METRO Compact’s governmental function exception rests more on pragmatic considerations than on any judgment about the intent of the Compact’s framers. Our sister circuit may believe that we should adopt the body of law governing the discretionary function ex*221ception when interpreting the METRO Compact simply because, when contrasted with the traditional distinctions between governmental and proprietary functions or between planning and operational level activities, Gaubert’s susceptible-to-policy-analysis test at least provides a workable standard that can be consistently applied. In other words, the D.C. Circuit’s argument may simply be that the Gaubert standard ought to be adopted because, for all its flaws, it is really the only game in town. Although I agree that the traditional distinction between governmental and proprietary functions is untenable and that we need a workable standard for determining the scope of the METRO’S immunity, this argument both overstates the workability of the Gaubert standard and understates the viability of possible alternatives to that standard. First, I have already noted that courts have had considerable difficulty in deciding whether government actions are grounded in economic, social, or political policy. See infra at 214. This has led to significant inconsistency in the case law. Compare, e.g., Tippett v. United States, 108 F.3d 1194, 1197-99 (10th Cir.1997) (holding that government’s failure to warn about the dangers of a charging moose was grounded in policy and was therefore protected by the discretionary function exception), with George v. United States, 735 F.Supp. 1524, 1533-34 (M.D.Ala.1990) (stating that government’s failure to warn about the presence of large alligators in a designated swimming area was not grounded in policy and was therefore outside the scope of the discretionary function exception). Indeed, I suggested above that by preventing courts from considering the decisionmaker’s authority to make policy judgments, Gaubert deprives courts of the analytical tools they need to draw sensible distinctions in the area of governmental immunity. Second, any increase in the predictability and consistency of judicial decisions that might have come about in the wake of Gaubert has come at the significant cost of allowing the discretionary function exception to swallow up the FTCA’s general waiver of sovereign immunity. Finally, there is at least one approach that would provide a viable alternative to the Gaubert standard. This is the approach suggested by Justice Scalia in his Gaubert concurrence. According to Justice Scalia, “a choice is shielded from liability by the discretionary function exception if the choice is, under the particular circumstances, one that -ought to be informed by considerations of social, economic, or political policy and is made by an officer whose official responsibilities include assessment of those considerations.” Gaubert, 499 U.S. at 335, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment). This standard is at least as easy to apply as the suseeptible-to-policy-analysis standard adopted by the Gaubert majority, and it would provide a more reasonable approach to interpreting the governmental function exception in section 80 of the METRO Compact. Further, even if it could be said that the framers of the METRO Compact “accepted the Dalehite conception” of governmental functions, Sanders, 819 F.2d at 1155, this acceptance would be better reflected in Justice Scalia’s approach than in that of the Gaubert majority because Justice Sca-lia sought to preserve the insights of the planning/operational distinction suggested by Dalehite. See Gaubert, 499 U.S. at 335, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment) (explaining that his approach “recognizes that there is something to the planning vs. operational dichotomy”).4 These reasons suggest that there is at least one viable *222alternative to the Gaubert standard and that the “only game in town” argument for adopting that standard is unconvincing.
I conclude, then, that the. D.C. Circuit erred in deciding that Gaubert provides the proper standard for defining the governmental function exception in section 80 of the METRO Compact. Yet because our sister circuit’s approach to section 80 appears-well-established, the cost of refusing to follow that approach is to create a circuit split between the only two circuits that are likely to hear tort claims against the METRO. As the majority points out, we have recognized that maintaining consistency between our circuit and the D.C. Circuit is an important consideration when interpreting the METRO Compact. Ante at 206 n. 9 (citing Lizzi v. Alexander, 255 F.3d 128, 134 (4th Cir.2001)). This raises the difficult question of whether adopting what I regard as a better reading of the METRO Compact is worth the price of creating a circuit split. While I appreciate the practical considerations behind the majority’s decision to follow the D.C. Circuit’s reading of section 80, the price of intercir-cuit consistency on this issue is the gutting of the Compact’s waiver of sovereign immunity. I believe that price is too high. I would therefore hold that Justice Scalia’s Gaubert concurrence provides a better standard for deciding whether tortious conduct that does not involve quintessentially governmental activities is shielded by the METRO’S immunity. Consequently, I must respectfully dissent from Parts II B and C of the majority opinion. This conclusion leaves the question of whether, under the standard proposed by Justice Sca-lia, the district court erred in refusing to recognize the METRO’S immunity for (1) the decision to use Escalator One as a stationary walker, (2) the decision not to reassemble Escalator Three, and (3) the failure to warn its passengers that there was no ascending escalator in operation at the Bethesda station.
III.
To reiterate, Justice Scalia’s standard regards a discretionary choice as policy based, and therefore immune, only “if the choice is, under the particular circumstances, one that ought to be . informed by considerations of social, economic, or political policy and is made by an officer whose official responsibilities include assessment of those considerations.” Gaubert, 499 U.S. at 335, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment). As my earlier discussion of the METRO’S decision not to reassemble Escalator Three indicates, the summary judgment record here would allow a reasonable jury to conclude that the person or persons who made the decision not to reassemble Escalator Three did not have the authority to make policy judgments about whether the increased repair costs of reassembling the escalator outweighed the increased risk to passengers created by leaving the. escalator disassembled. Accordingly, I would affirm the district court’s refusal to grant the METRO immunity for its decision not to reassemble Escalator Three. I must therefore dissent from part III B of the majority’s opinion. Because the record is likewise unclear about who made the decision to brake *223Escalator One and the decision not to warn passengers about the unavailability of an ascending escalator at the Bethesda station, I would also conclude that the METRO has not proven that it is entitled to immunity for these decisions. I am unclear, however, on whether these two theories of negligence survived the district court’s summary judgment ruling. The district court clearly ruled that Smith’s family and estate .had forecast sufficient evidence for a jury to conclude that the METRO had breached its duty of care to Smith by failing to provide a safe means of exiting the Bethesda METRO station and that this breach was the proximate cause of Smith’s death. In so ruling, the district court seemed to focus mainly on the METRO’S alleged failure to properly maintain and repair the escalators at the Bethesda station, but the court did not break down the theories of negligence advanced by Smith’s family and estate as explicitly as the majority does today. As a result, I am not sure the district court has decided whether the METRO’S decision to use Escalator One as a stationary walker (given the unavailability of any other functioning escalator) and its failure to warn passengers about the unavailability of an escalator at the Bethesda station raise jury questions on the issues of negligence and proximate cause. I would therefore remand for clarification on these points, much as the majority does in Part III E of its opinion.
IV.
In sum, .1 acknowledge that if Gaubert is the proper standard for applying the METRO Compact’s governmental function exception, the majority has correctly disposed of this case. This point underscores the need for close attention to the question of whether we should follow the D.C. Circuit’s reading of the Compact. Because I believe that Gaubert’s susceptible-to-policy-analysis standard protects a far greater range of conduct than the Compact’s framers would have envisioned, I dissent from the majority’s decision to adopt the Gaubert standard. Instead, I would adopt the standard proposed by Justice Scalia in his Gaubert concurrence because it is more consistent with the intent of the framers of the METRO Compact and does a far better job than the Gaubert standard of identifying those government decisions that are sufficiently policy based to warrant the protection of the governmental function exception. , Applying that standard, I would allow the case to proceed to trial on the question of whether the METRO’S decision not to reassemble Escalator Three was negligence and was the proximate cause of Smith’s death. I would instruct the district court to clarify on remand whether the other theories of negligence advanced by Smith’s family and estate survived the summary judgment ruling.
. The Gaubert majority did not explicitly address this point, but Justice Scalia’s concurrence clearly states that the discretionary function exception should apply only when “the decisionmaker [is] an official who possesses the relevant policy responsibility." Gaubert, 499 U.S. at 336, 111 S.Ct. 1267 (Scalia, J., concurring in part and concurring in the judgment). The Gaubert majority's silence in the face of Justice Scalia’s insistence that the actual decisionmaker be authorized to make policy judgments is naturally read to signal indifference to that requirement. The majority’s opinion in this case appears to agree with my reading of Gaubert, for it fails to address the questions of who made the decisions challenged by Smith’s family and estate and whether those persons had the authority to make policy judgments.
. There are exceptions, of course. The First Circuit has identified a line of cases suggesting that the discretionary function exception covers judgments that balance "incommensurable values in order to establish [policy] priorities” but does not protect "judgments] that embod[y] a professional assessment undertaken pursuant to a policy of settled priorities." Shansky, 164 F.3d at 694. See also Cope v. Scott, 45 F.3d 445, 451-52 (D.C.Cir.1995) (holding that the discretionary function exception did not protect the government's allegedly negligent failure to post proper warnings at certain points along a park road *215because the government had already chosen to place numerous road signs along the same stretch of road and “the discretion regarding where and what type of signs to post is -not the kind of discretion protected by the discretionary function exception”).
. Like the government’s counsel in Cope, the majority acknowledges that negligent operation of motor vehicles does not fall within the discretionary function exception. See ante at 207 n. 11. Thus, the majority would presumably conclude that the METRO would not be immune from liability if-one of its bus drivers caused an accident by driving at an unsafe rate of speed. But it is not obvious why a bus driver’s decision to drive at an excessive rate of speed in an effort to be on time at the next bus stop is distinguishable from the METRO’S decision that reassembling Escalator Three would be too costly. Each decision can be said with equal plausibility to implicate the METRO’S policy objectives.
. Some commentators have suggested that the best approach to government immunity *222would go farther than Justice Scalia's Gaubert concurrence by requiring not only that deci-sionmakers have the authority to make policy judgments, but also that the decisionmakers actually consider policy variables in making those judgments. See Peterson and Van Der Weide, supra, at 487-90. This proposal has considerable appeal, but I would not decide its merits today because Justice Scalia’s approach adequately addresses the facts of this case.