Full opinion text
BIRCH, Circuit Judge: In this appeal, we hold as a matter of first impression that Article 25 of the Warsaw Convention, as clarified by Montreal Protocol No. 4, requires a passenger to prove that an air carrier subjectively knew its conduct likely would result in harm to its passengers in order to escape the Convention’s limitations on liability. On summary judgment, the district court held that Article 25 measures an air carrier’s conduct objectively and concluded as a matter of law that, under either an objective or subjective standard, the passengers’ claims for damages were not limited by Article 25’s liability cap. For the reasons that follow, we conclude that the district court incorrectly entered summary judgment against the air carrier on this issue and remand for a determination by the finder of fact whether the air carrier’s conduct precludes the application of the Convention’s liability cap to this case. In addition, we hold that the district court: (1) properly applied Florida compensatory damages law to this case; (2) properly refused to apply Florida’s apportionment of liability statute; and (3) did not abuse its discretion during the damages trial by excluding evidence relating to the facts of the underlying plane crash and prohibiting reference to the legal finding of willful misconduct. We therefore affirm the judgment of the district court with respect to these issues. I. BACKGROUND A. Facts On December 20, 1995, American Airlines Flight 965 (“Flight 965”) crashed as the plane attempted to navigate its arrival to the Alfonso Bonilla Aragon airport in Cali, Colombia. The crash killed 151 passengers, including Maria Constanza Piam-ba Cortes, a domiciliary of Colombia who was returning home after studying in the United States. Appellee-Cross Appellant Doris Cristina Piamba Cortes (“Piamba Cortes”), acting both individually and as the personal representative of her sister Maria Constanza Piamba Cortes, filed a tort action against Appellant-Cross Appel-lee American Airlines, Inc. (“American”). The facts leading up to the crash are largely undisputed and have been detailed comprehensively by the district court. See In re Air Crash Near Cali, Colombia on December 20, 1995, 985 F.Supp. 1106, 1109-22 (S.D.Fla.1997). We need not duplicate the district court’s detailed factual recitation; for purposes of our discussion, we set forth an abbreviated statement of facts that are relevant in resolving this appeal. Flight 965 left Miami International Airport on the afternoon of December 20, 1995, bound for Cali. Captain Nicholas Ta-furi and First Officer Donnie Ray Williams piloted the Boeing 757, which the parties agree was airworthy and in good mechanical and structural condition. At all material times during the flight, Williams flew the aircraft while Tafuri primarily handled radio communications. The Cali airport is located in a valley approximately forty-three miles long and twelve miles wide. The arrival and approach paths for aircraft landing at the airport are designed to keep planes in an “airway” in the center of the valley and away from the mountainous terrain that surrounds the valley. American provides special training to its pilots who fly into Central and South America in order to acquaint them with the unusual features of these regions. Among other things, pilots are instructed, in no uncertain terms, not to rely on local mation about their location or position in the sky. According to American’s training materials, Latin American ATCs will assume when providing clearance that the pilot is on course, the plane is located where the pilot says it is, the pilot knows where the mountains are, and the pilot will refuse a clearance that will take the plane into a mountain. Because these assumptions may be incorrect, the ATCs will clear pilots to descend below minimum safe altitudes in mountainous areas. American also instructs its pilots that they must continually verify their exact location by every means available; if they are unable to locate and cross-check their position or are otherwise unsure of where they are, they must suspend any descent of the airplane until their position is verified and the safe minimum altitude is determined. Furthermore, American teaches its pilots to insist on the complete published or assigned flight plan for the plane’s approach to the airport unless the pilot is sure of the plane’s location and the terrain below. If the plane is operating on an unpublished route, American’s training materials and FAA regulations provide that “the pilot, when an approach clearance is received, shall maintain the last altitude assigned until the aircraft is established on a segment of the published route.” Id. at 1129-30. The flight plan assigned to Flight 965 called for the plane to follow a specified route during its arrival and approach to Cali. The arrival phase typically is conducted in accordance with a specified route that consists of a series of waypoints that define the path to the landing strip. In this case, the waypoints were marked by radio beacons known as “navaids,” which emit radio waves that can be tuned in from the cockpit and allow the pilot to determine the compass direction to, and in some instances the distance to, the waypoint. The waypoints also may help a pilot establish the plane’s position in the sky, as well as its distance to a certain point. Based on data recovered from Flight 965’s digital flight data recorder and the statements of Tafuri and Williams on the plane’s cockpit voice recorder, the parties have reconstructed the following events that led to the crash. Flight 965 approached Cali at night. Originally, Flight 965 was assigned a published arrival path to Cali that called for the plane to fly over the “Tulua” waypoint, located approximately thirty-four miles northeast of the airport, proceed to the “D21 CLO” waypoint, and then fly over the “Rozo” waypoint, which is located approximately three miles north of the airport. From there, the arrival path called for the plane to continue south to the “Cali” waypoint, located nine miles south of the airport, and, after executing a 180-degree turn, return north to the airport and land. When Flight 965 was approximately fifty-four miles north of the airport, the ATC stationed in Cali cleared the plane to the Cali waypoint and instructed the pilots to descend and maintain 15,000 feet and to “report uh, Tulua.” Id. at 1117. Moments later, however, the Cali ATC offered the pilots the option of landing straight onto the runway without having to turn the plane around at the Cali waypoint. The pilots accepted the offer, and thus accepted a published flight route that began at the Tulua waypoint, proceeded to the Rozo waypoint, and ended at the runway. After accepting the offer, the cockpit voice recorder suggests that Williams erroneously believed the flight route began at the Rozo waypoint instead of the Tulua waypoint. Tafuri told Williams that the flight route began at the Tulua waypoint, but then asked the ATC for permission to go “direct to Rozo and then do the Rozo arrival,” a request that set in motion a chain of events that culminated in the crash. Id. at 1118. The ATC responded by saying, “Affirmative,” but added instructions to “take the Rozo One” approach and to “report Tulua at twenty-one miles and five thousand feet.” Id. at 1119. After this exchange, one of the pilots sought to program the flight management computer (“FMC”) to fly automatically to the Rozo waypoint by typing the letter “R” into the FMC’s keypad. A total of twelve waypoints appeared on the FMC screen, the first of which was for the “Romeo” waypoint, located approximately 132 miles to the northeast of the plane. Although the pilots were required to verify that the chosen waypoint was actually Rozo, the pilot did not verify the Rozo waypoint and instead selected the Romeo waypoint. The FMC immediately began to fly the plane in the direction of the Romeo way-point, sending the plane on a prolonged, and pronounced, turn to the left, toward the east and toward the mountains. At the time the plane began turning, descending past an altitude of 16,880 and was flying adjacent to, or slightly the southwest of, the Tulua waypoint. During the turn east, Tafuri told Williams he wished the plane to fly to the Tulua waypoint, but instead of dialing the proper frequency for the Tulua waypoint (117.7) into his electronic horizon situation indicator (“EHSI”), Tafuri unwittingly di116.7, the frequency for a different waypoint located 160 miles to the east of valley. Consequently, the course deviation indicator (“D-bar”) function of his EHSI indicated that the Tulua waypoint located to the left of the plane. Tafu-however, instructed Williams to turn back toward the right, which sent the plane in a westerly direction and back toward the valley. At that time, the plane was south of the Tulua waypoint, well to the east of the valley, and east of the radials that define the flight route to the Rozo waypoint. The plane also had continued its descent, dropping more than 5,000 feet since the “R” had been entered into the FMC. Less than one minute later, Tafuri dialed 117.7, the correct frequency for the Tulua waypoint, into his EHSI. Because the plane already had passed the Tulua waypoint, this caused the D-bar indicator to shift on the EHSI screen. Tafuri instructed Williams to fly to the Cali way-point, although he also confirmed with the ATC that the flight plan called for the plane to fly first to the Tulua waypoint and then to the Rozo waypoint. Tafuri commented to Williams that he was having difficulty locating the Tulua waypoint, so Williams suggested that they intersect with the flight route and fly directly to the Rozo waypoint. At that time, the plane had descended to 10,000 feet and was still heading west. When the plane dropped to 8,480 feet, the plane’s ground proximity warning system sounded, directing the pilot to pull up. Williams attempted to plane’s ability to climb rapidly was hampered by the fact that Tafuri and Williams failed to pull back the speed brakes, which had been deployed several minutes earlier. Approximately thirty seconds later the plane crashed near the summit of El Delu-vio, a peak located approximately twenty-four miles northeast of the airport and approximately ten miles east of the airway. B. Procedural History After Piamba Cortes filed suit in Florida state court, American removed the case to federal court where it was consolidated for multidistrict pretrial proceedings with almost 160 other passenger lawsuits. Piam-ba Cortes, through the Plaintiffs’ Steering Committee, filed a motion for partial summary judgment on the issue of American’s liability. After a four-day hearing on the motion, the district court granted the motion. In a 118-page order granting the plaintiffs’ motions for summary judgment, the district court concluded that all the passengers’ suits against American fell under the terms of the Warsaw Convention. According to the language in effect at the time the district court entered its order, the Convention limited an air carrier’s liability except in cases of “willful misconduct.” The district court concluded that Eleventh Circuit law allows a passenger to establish willful misconduct in three ways, one of which is defined as “reckless disregard of the consequences.” In re Crash Near Cali, 985 F.Supp. at 1127. Noting that all the passengers’ representatives proceeded under a reckless disregard theory, the district court further concluded that reckless disregard contemplates a “rigorous objective inquiry” that is satisfied “by showing that the defendant’s conduct amounted to an extreme deviation from the standard of care under circumstances where the danger of likely harm was plain and obvious,” even if the defendant did not subjectively realize that its conduct placed its passengers at significant risk of harm. Id. at 1128, 29. Upon reviewing the evidence, the district court held that no reasonable jury could find that Tafuri and Williams’ conduct—in particular, the decision to continue descending at night in mountainous terrain when the circumstances made clear that the plane had strayed dramatically from the published arrival route—amounted to anything less than willful misconduct. Id. at 1138. The district court reached this conclusion by applying its objective standard for reckless disregard, although the court held in the alternative that, even if reckless disregard contemplates a subjective test, the evidence compelled a conclusion that Tafu-ri and Williams engaged in willful misconduct. After entering summary judgment in Piamba Cortes’ favor on the issue of liability, the district court conducted a trial on the issue of damages. Although Piamba Cortes’ sister was a domiciliary of Colombia, the district court’s conflict-of-laws analysis concluded that Florida compensatory damages law determined the elements of compensatory damages awarded to Piamba Cortes. In addition, the district court held that, under the Convention, American is liable for all compensatory damages and thus Florida law requiring the apportionment of liability did not apply. Finally, the district court ruled that, during the damages trial, Piamba Cortes could not introduce the factual circumstances of the crash and could not mention the court’s finding that the conduct of Flight 965’s pilots constituted willful misconduct. American appeals three issues, arguing that: (1) “willful misconduct” requires a subjective rather than an objective test, and the evidence creates a question of fact for the jury under this test; (2) conflict-of-laws principles warrant the application of the compensatory damages scheme used by the decedent’s domicile, which in this case is Colombia; and (3) the Convention operates as a “pass-through” on the issue of damages, and thus the district court should have applied Florida’s apportionment statute. Piamba Cortes cross-appeals, arguing that, during the trial on damages, she should have been able to introduce the facts of the crash and inform the jury that Flight 965’s pilots had engaged in willful misconduct as a matter of law. We consider each argument in turn. II. WARSAW CONVENTION American raises two distinct arguments related to the Warsaw Convention. First, American argues that the district court erroneously construed “willful misconduct” under the Convention to create an objective rather than a subjective test. Second, American argues that, if the subjective test is applied, a question of fact exists whether the conduct of Flight 965’s pilots constitutes willful misconduct, and therefore the entry of summary judgment on the issue of liability was inappropriate. The first issue requires us to determine whether the district court properly construed the terms of a treaty, which is a question of law that we review de novo. See Yapp v. Reno, 26 F.3d 1562, 1565 (11th Cir.1994). A. Background of the Warsaw Convention The Warsaw Convention is the commonly used name for the Convention for the Unification of Certain Rules Relating to International Transportation by Air, T.S. No. 876,137 L.N.T.S. 11, reprinted in note following 49 U.S.C.App. § 1502 (1988) (hereinafter “Warsaw Convention”), which entered into force for the United States on October 29,1934. See Butler v. Aeromexico, 774 F.2d 429, 430 n. 1 (11th Cir.1985). The Convention is the product of two international conferences, the first Paris in 1925 and the second in Warsaw in 1929. See generally Floyd v. Eastern Airlines, Inc., 872 F.2d 1462, 1467-69 (11th Cir.1989) (providing overview of the history, policies, and goals of the Warsaw Convention), rev’d on other grounds, 499 U.S. 530, 111 S.Ct. 1489, 113 L.Ed.2d 569 (1991). The Convention applies to “all international transportation of persons, baggage, or goods performed by aircraft for hire.” Warsaw Convention art. 1(1). Under Article 17 of the Convention, air carriers are “liable for damage sustained in the event of the death or wounding of a passenger ... if the accident which caused the damage so sustained took place on board the aircraft or in the course of any of the operations of embarking or disembarking.” Warsaw Convention art. 17. The Convention thus “established a presumption that air carriers are liable for damage sustained by passengers as a result of the carrier’s negligent conduct.” Floyd, 872 F.2d at 1467. In exchange for this presumption of liability, the drafters included a limitation on the amount of damages a passenger could recover from the carrier: 125,000 gold francs, or approximately $8,300. See Warsaw Convention art. 22; see also Floyd, 872 F.2d at 1467. As a companion provision to the liability cap contained in Article 22, the drafters adopted a safety valve by which passengers could escape the liability cap if they established that the air carrier’s conduct that caused their injuries constituted “wilful misconduct.” Warsaw Convention art. 25(1). Specifically, Article 25(1) provided that: The carrier shall not be entitled to avail himself of the provisions of this convention which exclude or limit his liability, if the damage is caused by his wilful misconduct or by such on as, in accordance with the law of the court to which the case is submitted, is considered to be the equivalent to wilful misconduct. Id. Consequently, in order to obtain any compensatory damages over $8,300 under the original version of the Convention, a passenger had to establish that his or her damages were the result of the air carrier’s willful misconduct. The $8,300 liability cap created by Article 22 proved to be a source of great dissatisfaction, particularly in the United States. See Floyd, 872 F.2d at 1468-69. In 1955, a conference similar to the 1929 conference in Warsaw convened at the Hague, during which delegates drafted what is known as the Hague Protocol. See Protocol to Amend the Convention for the Unification of Certain Rules Relating to International Carriage by Air Signed at Warsaw on 12 October 1929 (hereinafter “Hague Protocol”), reprinted in Lawrence B. Goldhirsch, The Warsaw Convention Annotated 265-74 (1988) (hereinafter “Gol-dhirsch”). The Hague Protocol proposed to double the liability limit contained in Article 22 to 250,000 gold francs, or approximately $16,600. See Hague Protocol art. XI, reprinted in Goldhirsch at 268; see also Floyd, 872 F.2d at 1468. In addition, delegates proposed deleting the term “willful misconduct” from Article 25 and replacing it with language that would allow a passenger to escape Article 22’s liability cap only if “it is proved that the damage resulted from an act or omission ... done with intent to cause damage or recklessly and with knowledge that damage would probably result.” Hague Protocol art. XIII, reprinted in Goldhirsch at 269. The Senate, however, refused to ratify, and the President did not adhere to, the Hague Protocol. The rejection of the Hague Protocol by the United States prompted the execution of the Montreal Agreement of 1966, in which air carriers agreed to enter into private contractual agreements with all passengers (created by the purchase of a ticket) to raise the liability limit to $75,000 for all international flights originating, terminating, or having a connecting point in the United States. Agreement CAB 18900, approved by Civil Aeronautics Board Order No. E-28680, May 13, 1966, 31 Fed.Reg. 7302 (1966). The execution of the Montreal Agreement set the stage for a concerted effort to update the terms of the Convention to reflect modern legal and technological standards. At another conference in Guatemala City in 1971, delegates proposed raising the liability cap to approximately $136,000. See Protocol to Amend the Convention for the Unification of Certain Rules Relating to International Carriage by Air Signed at Warsaw on 12 October 1929 as Amended by the Protocol Done at the Hague on 28 September 1955 (“the Guatemala City Protocol”), reprinted in Goldhirsch at 319-29. In 1975, delegates met once again in Montreal, Canada, and adopted a series of four protocols, known collectively as the Montreal Protocols. See Goldhirsch at 331-62 (reprinting the four Montreal Protocols). During the twenty years following the 1975 Montreal conference, the United States declined to ratify either the Guatemala City Protocol or the Montreal Protocols. On September 28, 1998, the Senate ratified the fourth of the Montreal Protocols (“Montreal Protocol No. 4”) independently of the Guatemala City Protocol and the three other Montreal Protocols, and the President signed the instrument of ratification on November 5, 1998. See El Al Israel Airlines, Ltd. v. Tsui Yuan Tseng, — U.S.-,-, 119 S.Ct. 662, 674 & n. 14, 142 L.Ed.2d 576 (1999). Montreal Protocol No. 4 replaces the term “willful misconduct” in Article 25 with the same language as contained in the Hague Protocol: [t]he limits of liability specified in ... Article 22 shall not apply if it is proved that the damage resulted from an act or omission of the carrier, his servants or agents, done with intent to cause damage or recklessly and with knowledge that damage would probably result. Additional Protocol No. 4 to Amend the Convention for the Unification of Certain Rules Relating to International Carriage by Air Signed at Warsaw on 12 October 1929 as Amended by the Protocol Done at the Hague on 28 September 1955 (hereinafter “Montreal Protocol No. 4”) art. IX, reprinted in Goldhirsch at 358. B. Application of Montreal Protocol No. I to This Case During the district court proceedings, Piamba Cortes’ ability to recover more than $75,000 hinged upon the pilots of Flight 965 engaged in willful misconduct under Article 25, thus allowing Piamba Cortes to avoid the limitations on liability contained in Article 22. While this case was on appeal, however, the United States’ adherence to Montreal Protocol No. 4 changed the language used in Article 25 to measure such conduct. Montreal Protocol No. 4 specifies that it will enter into force in a ratifying jurisdiction ninety days after the instrument of ratification is deposited with the Polish government. See S. Exec. Rep. 105-20, at 4 (1998). As a result, the Protocol entered into force in the United States on March 4, 1999. See Tsui Yuan Tseng, — U.S. at-n. 14, 119 S.Ct. at 674 n. 14. We therefore must determine whether the new language contained in Montreal Protocol No. 4 applies to this case. We first look to see whether the amendment effects a substantive change in the legal standard or merely clarifies the prior law. As we explain, if the amendment clarifies prior law rather than changing it, no concerns about retroactive application arise and the amendment is applied to the present proceeding as an accurate restatement of prior law. At first, it seems intuitively appealing to conclude that, because the new language significantly alters the text of the original Convention, the original Convention’s language may be presumed to have meant the opposite. For example, Montreal Protocol No. 4’s language includes an express requirement that the air carrier must know that damage probably will result from its conduct, so one might conclude that such a requirement was omitted from the original Convention. This intuition runs contrary to our precedent, however, which holds that an amendment con- The Department of Transportation approved the IATA Intercarrier Agreement on November 12, 1996. See D.O.T. Order 96-11-6, 1996 WL 656334, at *3 (D.O.T. Nov. 12, 1996). As of April 5, 1998, fifty-one domestic and foreign air carriers had adhered to the taming new language may be clarify existing law, to correct a misinterpretation, or to overrule wrongly decided cases. Thus, an amendment ... does not necessarily indicate that the unamended statute meant the opposite” of the language contained in the amendment. United States v. Sepulveda, 115 F.3d 882, 885 n. 5 (11th Cir.1997). Moreover, concerns about retroactive application are not implicated when an amendment that takes effect after the initiation of a lawsuit is deemed to clarify relevant law rather than effect a substantive change in the law. See Beverly Community Hosp. Ass’n v. Belshe, 132 F.3d 1259, 1265 (9th Cir.1997), cert, denied, — U.S.-, 119 S.Ct. 334, 142 L.Ed.2d 276 (1998); Liquilux Gas Corp. v. Martin Gas Sales, 979 F.2d 887, 890 (1st Cir. 1992); Boddie v. American Broadcasting Cos., 881 F.2d 267, 269 (6th Cir.1989); cf. Tsui Yuan Tseng, — U.S. at-, 119 S.Ct. at 667-68 (concluding that a provision in Montreal Protocol No. 4 limiting recovery for bodily injuries clarifies, but does not change, prior law under the Convention). In effect, the court applies the law as set forth in the amendment to the present proceeding because the amendment accurately restates the prior law. See Liquilux, 979 F.2d at 890 (“Clarification, effective ab initio, is a well recognized principle.”). Several factors are relevant when determining if an amendment clarifies, rather than effects a substantive change to, prior law. A significant factor is whether a conflict or ambiguity existed with respect to the interpretation of the relevant provision when the amendment was enacted. If such an ambiguity exist- IATA Intercarrier Agreement, including American. See S. Exec. Rep. No. 105-20, at 57. The parties have not argued that the IATA Intercarrier Agreement applies retroactively to this case, and we therefore do not consider the issue. ed, courts view this as an indication that a subsequent amendment is intended to clarify, rather than change, the existing law. See Liquilux, 979 F.2d at 890. Second, courts may rely upon a declaration by the enacting body that its intent is to clarify the prior enactment. See id. Courts should examine such declarations carefully, however, especially if the declarations are found in the amendment’s legislative history rather than the text of the amendment itself. See Consumer Prod. Safety Comm’n v. GTE Sylvania, Inc., 447 U.S. 102, 118 n. 13, 100 S.Ct. 2051, 2061 n. 13, 64 L.Ed.2d 766 (1980). As a general rule, “[a] mere statement in a conference report of [subsequent] legislation as to what the Committee believes an earlier statute meant is obviously less weighty” than a statement in the amendment itself. Id.; see also Pennsylvania Med. Soc’y v. Snider, 29 F.3d 886, 900 (3d Cir.1994) (attributing no value to a House committee report stating that an amendment clarifies prior law when the statement is inconsistent with a logical reading of the earlier version of the statute and with the legislative history of the earlier statute). Declarations in the subsequent legislative history nonetheless may be relevant to this analysis, especially if the legislative history is consistent with a reasonable interpretation of the prior enactment and its legislative history. See Sykes v. Columbus & Greenville Ry., 117 F.3d 287, 293-94 (5th Cir.1997) (“Although a committee report written with regard to a subsequent enactment is not legislative history with regard to a previously enacted statute, it is entitled to some consideration as a secondarily authoritative expression of expert opinion.”) (quoting Bobsee Corp. v. United States, 411 F.2d 231, 237 n. 18 (5th Cir.1969)); SEC v. Clark, 915 F.2d 439, 451-52 (9th Cir.1990) (“While a statement concerning an earlier statute by members of a subsequent legislature is of course not conclusive evidence of the meaning of the earlier statute, the later interpretation may be accorded some deference where the subsequent legislative commentary accompanies the enactment of an amendment to the earlier law.”); cf. GTE Sylvania, 447 U.S. at 118 n. 13, 100 S.Ct. at 2061 n. 13 (noting that such history is “sometimes considered relevant,” but “subsequent legislative history will rarely override a reasonable interpretation of a statute that can be gleaned from its language and legislative history prior to its enactment”). 1. Ambiguities and Conflicting Interpretations of Article 25 We first examine whether a conflict or ambiguity existed with respect to the application of Article 25 as originally enacted. The English translation of Article 25 allowed a passenger to avoid the liability cap if he could establish that the injury resulted from willful misconduct by the air carrier. We announced the Eleventh Circuit’s test for willful misconduct in Butler. 774 F.2d at 430. Adopting a test used by the District of Columbia Circuit, we held that willful misconduct “mean[s] ‘the intentional performance of an act with knowledge that the ... act will probably result in injury or damage’ or ‘reckless disregard of the consequences’ or ‘a deliberate purpose not to discharge some duty necessary to safety.’ ” Id. (quoting Koninklijke Luchtvaart Maatschappij N.V. v. Tuller, 292 F.2d 775, 778-79 (D.C.Cir.1961)). Butler thus identified three alternative ways to prove willful misconduct: (1) intentional performance of an act knowing that the act likely would result in injury or damage; (2) an action taken with “reckless disregard” of the consequences; or (3) a deliberate failure to discharge a duty necessary to safety. 774 F.2d at 430. The first prong questions whether the carrier subjectively realized at the time of the relevant action that this action was likely to result in injury or damage to its passengers. The second prong is less requiring that an air carrier “recklessly disregarded” the consequences of its acts. Without offering further explanation of the meaning of the term “reckless disregard,” we held that the facts in Butler established that the defendant’s pilot had recklessly disregarded the likelihood that his conduct would cause a plane crash and therefore had engaged in willful misconduct. Id. at 431-32. In other Eleventh Circuit precedent, the term of art “reckless disregard” has different meanings depending on the context in which it is used. For example, when examining whether an individual willfully failed to pay withholding taxes, reckless disregard requires “something less than actual knowledge” of a risk. Malloy v. United States, 17 F.3d 329, 332 (11th Cir.1994). In libel law, on the other hand, reckless disregard requires a showing that a defendant in fact knew that a published statement might be false. See Meisler v. Gannett Co., Inc., 12 F.3d 1026, 1030 (11th Cir.1994) (‘“Reckless disregard,’ for purposes of proving actual malice, is shown if the defendant entertained serious doubts as to the truth of his publication.”) (internal quotation marks omitted). We therefore find little guidance under Butler and other Eleventh Circuit precedent as to the precise meaning of reckless disregard in the context of cases filed under the Convention. A similar lack of clarity exists in our sister circuits when applying the term “willful misconduct” in Warsaw Convention cases. Despite the fact that all circuits to address the issue have used the term “reckless disregard” when applying Article 25, see, e.g., Shah v. Pan Am. World Servs., Inc., 148 F.3d 84, 93 (2d Cir.1998) (carrier must have acted either (1) with knowledge that its actions would result in injury or death, or (2) in conscious or reckless disregard of the fact that death or injury would be the probable consequences of its actions), cert. denied, — U.S.-, 119 S.Ct. 1033, 143 L.Ed.2d 42 (1999); Koirala v. Thai Airways Int’l, Ltd., 126 F.3d 1205, 1209-10 (9th Cir.1997) (air carrier must intentionally perform an act, or fail to perform an act, with knowledge that it probably will result in injury or harm, or intentionally perform an act in some manner as to imply a reckless disregard of the consequences of its performance); Saba v. Compagnie Nationale Air France, 78 F.3d 664, 666 (D.C.Cir.1996) (same); In re Air Crash Disaster, 86 F.3d 498, 544 (6th Cir.1996) (same), the courts have employed a “know-it-when-we-see-it” approach rather than articulating precisely what is meant by reckless disregard. See, e.g., Saba, 78 F.3d at 667 (stating that, although the D.C. Circuit clearly has equated willful misconduct with reckless disregard, the court “never ha[s] been very clear as to what we meant by reckless disregard”); see also Perry S. Bechky, Mismanagement and Misinterpretation: U.S. Judicial Implementation of the Warsaw Convention in Air Disaster Litigation, 60 J. Air L. & Com. 455, 501-02 (1995) (U.S. courts have “grappled” with question whether reckless disregard envisions a subjective or objective test, but have “generally regarded ‘willful misconduct’ as equivalent to recklessness or gross negligence”). Notwithstanding the fact that courts have avoided precise definitions of reckless disregard in Warsaw Convention eases, it is possible to make inferences regarding the analysis the courts employed in their decisions. For example, in 1961 the District of Columbia Circuit upheld multiple findings of willful misconduct based on a theory of reckless disregard without finding that the air carrier or its pilots subjectively knew that their conduct likely would harm passengers. See Tuller, 292 F.2d at 779-80. The Second Circuit, on the other hand, admonished a district court for “concluding that the Second Circuit does not require knowledge that damage would probably result” when assessing willful misconduct under a theory of reckless disregard. See Berner v. British Commonwealth Pac. Airlines, Ltd., 346 F.2d 532, 536 (2d Cir.1965). The Berner court further observed that “[t]here must be a realization of the probability of injury from the conduct, and a disregard of the probable consequences of such conduct.” Id. at 537 (quoting Grey v. American Airlines, Inc., 227 F.2d 282, 285 (2d Cir.1955)). Later Second Circuit opinions continue to require a showing that the carrier knew that its actions placed its passengers at risk. See, e.g., Republic Nat. Bank v. Eastern Airlines, Inc., 815 F.2d 232, 239 (2d Cir.1987) (“factors must be established indicating that such a loss is likely to occur and that defendant was aware of the probability”). The opinion that has attracted the most recent attention in this area is Saba, in which the District of Columbia Circuit endeavored to clarify the ambiguity surrounding that circuit’s definitions of reckless disregard and willful misconduct. Saba adopts a definition of willful misconduct that is the same in all practical respects as the definitions adopted in Tuller and Butler: “[wjillful misconduct is the intentional performance of an act with knowledge that the act will probably result in an injury or damage, or in some manner as to imply reckless disregard of the consequences of its performance.” 78 F.3d at 666. The district court in Saba concluded that the carrier engaged in willful misconduct by recklessly disregarding an obvious risk — specifically, the possibility that inadequate packing procedures would expose cargo to water damage when the cargo was left outside in the rain. Id. at 670. The circuit court reversed, holding that reckless disregard cannot be based solely upon a finding that the risk was obvious. Id. at 669-70. Rather, the plaintiff must prove that the carrier subjectively knew that the carpets would be exposed to rain or that the packing procedures used would create a grave risk of water damage to the cargo if it rained. Id. at 670. The District of Columbia Circuit labeled this analysis a subjective test because it “requires a showing of a subjective state of mind.” Id. at 668. The court acknowledged that the carrier’s subjective state of mind may be established solely by inferences taken from circumstantial evidence; the inferences thus act as “a legitimate substitution for intent to do the proscribed act because, if shown, it is a proxy for that forbidden intent.” Id. Saba’s test thus is satisfied if evidence allows an inference that the carrier “must have been aware” of a risk created by its conduct. Id. at 669. The court observed that its test is more stringent than an objective test that inquires only if the carrier “should have known” about the danger, because the test cannot be satisfied by showing merely “an extreme departure from standards of ordinary care.” Id. The holdings of Saba, Berner, and Tul-ler thus are not readily reconcilable and do not provide a clear definition of what is meant by reckless disregard in Convention cases. An examination of the interpretations of willful misconduct by other countries that are signatories to the Warsaw Convention reveals a lack of uniformity similar to that contained in United States case law. Cf. Zicherman v. Korean Air Lines Co., Ltd., 516 U.S. 217, 226, 116 S.Ct. 629, 634, 133 L.Ed.2d 596 (1996) (“Because a treaty ratified by the United States is not only the law of this land, [cit.], but also an agreement among sovereign powers, we have traditionally considered as aids to its interpretation ... the post-ratification understanding of the contracting parties”). According to one collection of case law interpreting the Warsaw Convention, foreign jurisdictions have adopted both subjective and objective tests for willful misconduct, with France, Germany, Greece, and Korea adopting objective tests and Switzerland adopting a subjective test. See Golhirseh at 121 (collecting cases). This brief discussion addresses only a handful of the published cases that have construed willful misconduct under Article 25. Even this limited review, however, reveals a body of law that frequently is inconsistent and that provides a vague and nebulous definition of willful misconduct, rendering it difficult to apply. Under these circumstances, this ambiguity supports a conclusion that Montreal Protocol No. 4 clarifies, rather than effects a substantive change to, Article 25. 2. Declarations of Intent Concerning Montreal Protocol No. I Montreal Protocol No. 4 contains no statements concerning an intent to clarify or change prior law. We therefore will examine the drafting and legislative history of the amended language to discern the intentions behind enacting the amendment. As we have explained, Montreal Protocol No. 4 adopts the Hague Protocol’s substituted language for Article 25; we therefore will look first to the drafting history of the Hague Protocol, where the negotiations surrounding the adoption of this language occurred. The delegates at the 1955 Hague Conference began their deliberations with a draft proposal that narrowed Article 25 to allow unlimited liability only where the carrier committed a “deliberate act or omission ... done with intent to cause damage.” ICAO Doc. 7686 LC/140, Vol. II, Documents 99 (1956). The Norwegian delegation proposed an amendment to Article 25 that would force the carrier to bear unlimited liability if “the act or omission was committed either with the intention to cause damage or recklessly by not caring whether or not damage was likely to result.” Id. at 174. The Norwegian delegate explained this language to mean that “the person in question understood that there might be damage caused by his act or omission, but, nevertheless, he took the position of saying: T am quite indifferent as to whether damage will occur or not.’ ” ICAO Doc 7686-LC/140 Vol. I, Minutes 196 (1956) (hereinafter “Hague Conference Minutes”) (emphasis added). Other countries interpreted the proposal the same way. See id. at 194-96 (statements of Spanish and British delegates). The Dutch delegation observed that, if the delegates intended to include a requirement that the carrier subjectively realize that damage likely would result from its actions, the Norwegian proposal’s failure to specify the requirement expressly left the proposal open to the same interpretive problems encountered with the language used in the original Convention. See id. at 197-98. The Dutch delegation thus proposed replacing the phrase “not caring whether or not damage was likely to result” with the phrase “with full realization of the reckless character of his or their conduct and of the danger that damage would result.” Id. at 198. After several other delegates concurred with the Dutch proposal, the conference finally adopted the phrase “with knowledge that damage would probably result.” See id. at 198-206. Significantly, both the Norwegian proposal and the final adopted language reflected an effort to retain the same standard of conduct that had been adopted under the text of the Convention. See id. at 196 (statement by British delegate that “[o]ne of the most important elements in cases decided by courts on the Warsaw Convention was that it must be shown that the servants or agents of the operator ... had knowledge of the probable consequences of their acts”); id. at 197 (statement by United States’ delegate that the proposed alteration to Article 26, “although different in drafting from the text of the present Article 25, continued as nearly as possible to establish the same rules of law as existed in American jurisprudence”). We next turn to the legislative history surrounding the United States’ adherence to Montreal Protocol No. 4, which supports this interpretation. According to a report prepared by the Senate Committee on Foreign Relations, the language that replaces willful misconduct in Article 25 “does not modify the scope of the standard ... [but rather serves as] a clarifying response to the difficulties that arose from differing translations of the text” of the original Convention. S. Exec. Rep. No. 105-20, at 15 (1998). In the view of the Senate Committee, the Protocol replaces the term “willful misconduct” with “the common law definition of ‘willful misconduct.’” Id. The State Department, in response to questions from members of the Senate Committee, explained further that the language contained in Montreal Protocol No. 4 “is merely an alternative interpretation of the original French text [of the Convention], developed to harmonize the various legal interpretations that had developed from the original.” Id. at 47. Use of the. term “willful misconduct” in the original Convention resulted, in the State Department’s view, in a: discrepancy between common and civil law concerning the nature of conduct required to remove limits on liability. Because the concept of willful misconduct came to have different connotations in the civil and common law systems, the drafters [of the amended language] ... replaced the legal standards with a description of the conduct itself. Id. The State Department therefore concluded that “this change does not modify the scope of the standard,” id., and that “the amendment to Article 25 will have no practical effect on the rights of claimants in cases under the Warsaw Convention.” Id. at 53. The Senate Committee and State Department premised their views upon the express requirement in Montreal Protocol No. 4 that a passenger must prove that the carrier knew its conduct would likely result in damage. Among other things, the Senate report quotes two Second Circuit opinions that require passengers seeking to establish willful misconduct to prove knowledge on the part of the carrier that injury likely would result from its actions. See S. Exec. Rep. 105-20, at 53 (quoting Pekelis v. Transcontinental & W. Air, Inc., 187 F.2d 122 (2d Cir.1951), and Grey v. American Airlines, Inc., 227 F.2d 282 (2d Cir.1955)). In fact, in Grey, the Second Circuit emphasized that “[tjhere must be a realization of the probability of injury, from the conduct, and a disregard of the probable consequences of such conduct.” 227 F.2d at 285. In sum, the Senate Committee and the State Department concluded that Article 25 always has required a passenger to prove knowledge on the part of the air carrier that its conduct would likely result in damage, and that Montreal Protocol No. 4 clarifies the existing law to codify expressly this requirement. We pay close attention to this conclusion, as “[rjespect is ordinarily due the reasonable views of the Executive Branch concerning the meaning of an international treaty.” Tsui Yuan Tseng, — U.S. at-, 119 S.Ct. at 671. At the same time, we are mindful of the admonition that subsequent legislative history purporting to clarify prior law should be viewed skeptically if it is inconsistent with a reasonable interpretation of the text and legislative history of the earlier enactment. See GTE Sylvania, 447 U.S. at 118 n. 13, 100 S.Ct. at 2061 n. 13. We therefore will examine the language and drafting history of the original Convention to ensure that the subsequent declarations of intent are consistent with the earlier law. The delegates to the 1929 Warsaw conference drafted the Convention in French, and used the word “dol” to describe the level of misconduct that allows a passenger to bypass Article 22’s liability cap. Although the term “dol” has “no precise analogue in the English language,” Second International Conference on Private Aeronautical Law, Oct. 4-12, 1929, Warsaw, Minutes at v. (Robert C. Horner & Didier Legrez trans.1975) (hereinafter “Warsaw Minutes”), “[i]t implies an act or omission that was done intentionally to cause a harm.” Goldhirsch at 121. The use of the word “dol” resulted from the drafters’ dissatisfaction with the phrase “intentional illicit act,” which had been included in an earlier draft of Article 25. See Warsaw Minutes at 265. Initially, the delegates could not agree on proposed substitutions to replace “intentional illicit act.” The German delegation suggested the term “faute lorde,” which the delegates equated with a common-law gross negligence standard, see id. at 58-59, 61, 278, 290, while the British delegation proposed to limit the exemption to acts committed deliberately for the purpose of injury. See id. at 298. The Brazilian delegate, Alcibiades Peca-nha, presciently observed that the competing proposals raised the question whether the air carrier’s conduct was to be measured by an objective or subjective standard, and consequently endorsed a compromise approach. See id. at 61. Ultimately, the delegates rejected the inclusion of “faute lorde” and retained the French word “dol,” adding that a court may apply the legal equivalent of “dol” as defined by the law of the forum jurisdiction. Warsaw Convention art. 25(1). As noted by British delegate Sir Alfred Dennis, the adopted language reflects the delegates’ agreement that “dol” was to be translated into English as willful misconduct, see Warsaw Minutes at 213, which Sir Dennis defined as “covering] not only deliberate acts but also careless acts done without regard for the consequences.” Id. at 59-60. The drafting history thus reveals that conferees rejected an effort to define willful misconduct to encompass gross negligence. Although Sir Dennis’ characterization of willful misconduct suggests that the standard may be satisfied without establishing that the carrier knew its actions placed its passengers at risk, the language adopted by the conference does not expressly embody this characterization. We therefore find the 1929 drafting history to be ambiguous in this respect. In sum, the recent legislative history surrounding Montreal Protocol No. 4 is consistent with a reasonable interpretation of the original text, and the drafting history for the original Convention does not suggest otherwise. Given the uniform and clear statements of those who enacted and adopted the amended language contained in Montreal Protocol No. 4, we find these statements to be persuasive indicators that the Protocol clarifies, rather than effects a substantive change to, existing law. C. Summary of Law For the reasons that we have explained, we conclude that Montreal Protocol No. 4 clarifies the definition of willful misconduct under Article 25, rather than effecting a substantive change in the law. The amended language provides a more precise articulation of the standard, requiring a passenger to prove that the carrier, or its servants or agents, acted: (1) “with intent to cause damage,” or (2) “recklessly and with knowledge that damage would probably result.” Montreal Protocol No. 4, art. IX, reprinted in Goldhirsch at 358. This definition of the standard replaces the less precise articulation set forth in Butler, including the reckless disregard standard employed by the district court. Under the clarified standard, we no longer inquire as to reckless disregard, but rather examine whether the pilots of Flight 965, at a minimum, acted recklessly and with knowledge that their conduct likely would result in damage. Before applying the Protocol’s clarified definition of the standard to this case, we believe it is necessary to comment upon the type of evidence that may be used to satisfy the standard. For this task, we refer to Farmer v. Brennan, 511 U.S. 825, 114 S.Ct. 1970, 128 L.Ed.2d 811 (1994), where the Court announced its standard for “deliberate indifference” in Eighth Amendment cases. 511 U.S. at 837, 114 S.Ct. at 1979. The Farmer Court’s discussion is helpful here because the Court explained in great detail how a plaintiff may prove that a defendant subjectively knew that his or her actions would likely result in harm to the plaintiff. See id. at 836-44, 114 S.Ct. at 1978-82; see also Saba, 78 F.3d at 669 (analogizing the Farmer Court’s discussion to a plaintiffs claim that an air carrier engaged in willful misconduct under Article 25). Farmer begins its analysis by dividing the legal definition of recklessness into two distinct standards, one measured objectively and the other subjectively. 511 U.S. at 836-37, 114 S.Ct. at 1978-79. An objective test, according to Farmer, examines whether an actor acts or fails to discharge a duty to act “in the face of an unjustifiably high risk of harm that is either known or so obvious that it should be known.” Id at 836, 114 S.Ct. at 1978. A subjective test, on the other hand, asks whether an actor has disregarded a risk of harm of which he is aware. Id at 836-38, 114 S.Ct. at 1978-79. Stated differently, to satisfy the subjective test the actor “must both be aware of facts from which the inference could be drawn that a substantial risk of serious harm exists, and he must also draw the inference.” Id at 837, 114 S.Ct. at 1979. Because the language set forth in Montreal Protocol No. 4 also requires a plaintiff to establish that the carrier knows that its conduct likely will result in damage—in other words, that the carrier has drawn an inference that a risk of harm exists—Article 25 creates what Farmer defines as a subjective test. Under Farmer, establishing knowledge on the part of the actor need not be accomplished solely by direct evidence; a factfin-der is permitted to infer from circumstantial evidence that the actor actually drew the inference that the circumstances posed a substantial risk of harm. Id at 842, 114 S.Ct. at 1981. Indeed, it is possible to premise this inference on “the very fact that the risk was obvious.” Id Obviousness of the risk thus plays a role in both the subjective and the objective tests. The difference, when the tests are put into practice, is a fine one. The objective test is satisfied if a grave risk is sufficiently obvious, because the person “should have” been aware of the risk regardless of whether he actually recognized it. See Saba, 78 F.3d at 669. The subjective test, on the other hand, precludes a finding of liability if the factfinder concludes that, even though a grave risk is obvious, no inference can be made that the actor actually became aware of the risk. See Farmer, 511 U.S. at 844, 114 S.Ct. at 1982 (“That a trier of fact may infer knowledge from the obvious ... does not mean that it must do so.”). In this way, a plaintiff may rely solely upon circumstantial evidence related to the obviousness of a grave risk to satisfy both tests, but the subjective test is satisfied only if the circumstances also permit an inference that the actor “must have known” about the risk. Id at 842-43, 114 S.Ct. at 1981-82. “It is not enough merely to find that a reasonable person would have known, or that the defendant should have known,” of the risk. Id at 843 n. 8,114 S.Ct. at 1982 n. 8. Thus, while an objective test asks whether an actor “should have known” of an obvious risk, the subjective test requires, at a minimum, a showing that the actor “must have known” of the risk. Cf. Spruce v. Sargent, 149 F.3d 783, 786 (8th Cir.1998). D. Entry of Summary Judgment Against American Having determined the proper test to be applied, we now turn to American’s argument that the district court improperly entered summary judgment in favor of the passengers on the issue of willful misconduct. Without' having the benefit of Montreal Protocol No. 4’s more precise language, the district court applied a test for willful misconduct that is at odds with the conclusions contained in our opinion. Acting perhaps out of an abundance of caution, however, the district court held in the alternative that, because the pilots of Flight 965 decided to continue descending even though they knew they were off course in a dangerously mountainous region, Piamba Cortes was entitled to summary judgment on the issue of willful misconduct even under a subjective test. If correct, the district court’s decision may be affirmed on this ground without requiring a remand. We review a district court’s entry of summary judgment de novo. See City of Tuscaloosa v. Hateros Chems, Inc., 158 F.3d 548, 556 (11th Cir.1998). Summary judgment is appropriate only if no genuine issue of material fact exists and the moving party is entitled to judgment as a matter of law. See Fed.R.Civ.P. 56(c). In assessing whether the movant has met its burden of demonstrating the absence of a genuine issue of fact, the court must view the evidence and all factual inferences in the light most favorable to the party opposing the motion. See Reynolds v. Bridgestone/Firestone, Inc., 989 F.2d 465, 469 (11th Cir.1993). An issue of fact is genuine, thus barring the entry of summary judgment, unless “the record taken as a whole could not lead a rational trier of fact to find for the non-moving party.” Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 1356, 89 L.Ed.2d 538 (1986). Piamba Cortes argues that the pilots of Flight 965 knew that, by continuing their descent even though they knew the plane was off course in a mountainous region, their conduct likely would result in harm to the passengers. To justify entering summary judgment in her favor on this claim, Piamba Cortes must show that no genuine dispute exists with respect to three distinct factual issues: (1) the pilots of Flight 965 knew that the plane was significantly off course in a mountainous region; (2) the pilots knowingly elected to continue descending the plane; and (3) the pilots knew that descending the plane under such circumstances would likely result in harm to the passengers. The second and third issues are easily resolved in Piamba Cortes’ favor. The record is replete with convincing circumstantial evidence that both Tafuri and Williams knew that continuing a descent in a mountainous region when the aircraft was significantly off course would create a risk of danger to the passengers. Most compelling, however, is the fact that American did not argue to the contrary in the district court and does not argue to the contrary on appeal. Tafuri and Williams’ awareness that such conduct would create a risk of injury to passengers therefore is not at issue here. The same is true with respect to the pilots’ knowledge that the plane was descending. Statements by the pilots recorded by the cockpit voice recorder reveal that the pilots were actively monitoring the plane’s altitude during the descent. Furthermore, American once again poses no argument to the contrary on appeal. The final, and most difficult, question is whether the pilots in fact knew that the aircraft was off course while it was descending. We narrow this inquiry to reflect the fact that, in order to enter summary judgment in favor of the passengers, the pilots must have known that the aircraft was “significantly” off course — in other words, at risk of leaving the valley while the pilots continued their descent. If the pilots believed that the plane was slightly off course, yet well within the safety of the valley, a factfinder reasonably might infer that the pilots were not actually aware that their actions probably would result in injury to the passengers. On the other hand, the pilots need not have realized that the aircraft had flown all the way out of the valley to realize that their conduct placed the passengers at risk of harm. As American concedes, if the phots knew the plane was significantly off course, the only reasonable inference a factfinder could make is that the pilots knew the plane was at risk of crashing into the mountains. Consequently, entry of summary judgment in favor of Piamba Cortes hinges upon a narrow finding that a reasonable juror could only infer that Ta-furi and Williams realized the aircraft was significantly off course during the time they continued their descent. The district court concluded that the only plausible inference to be taken from the evidence is that the pilots knew that they had strayed significantly away from the published arrival path. We agree with the district court that a reasonable factfin-der must conclude that the pilots knew they were off course. We disagree, however, that the only reasonable inference is that the pilots knew they were significantly off course; to the contrary, even though more plausible interpretations suggest otherwise, a factfinder reasonably might conclude that the pilots believed they were near enough to the published arrival path that they did not realize they had placed the passengers at risk of harm. The district court set forth a detailed and thorough account of the circumstantial evidence supporting a finding that the pilots realized they were not on the published flight arrival path to Cali. See generally In re Crash Near Cali, 985 F.Supp. at 1138-43. This occurred after the pilots mistakenly entered “Romeo” instead of “Rozo” into the FMC and the plane turned east toward the mountains. According to the district court, the pilots likely would have realized that, in light of the amount of time they had been flying east, the aircraft in fact was significantly off the published course to the Tulua waypoint. Nonetheless, the evidence did not show conclusively that the pilots perceived the amount of time that they had been flying east toward the mountains, thus allowing a reasonable inference that the pilots believed the aircraft to be near the published arrival path even after the turn. Id. at 1140. The circumstances changed, however, at the moment marked 21:38:54 on the cockpit voice recorder, when the pilots realized that the aircraft was headed in the wrong direction and must turn to the right to intercept the proper course. As sum-marked by the district court, the pilots’ statements — such as “[wjhere’re we going” and “we got fucked up here didn’t we?”— added with expert testimony describing this portion of the flight — allow only one reasonable inference: that the pilots realized that the plane was not on the published arrival path. The evidence is not equally compelling, however, with respect to the pilots’ knowledge of the extent that they had traveled off course. We agree with the district court that, given the pilots’ statements on the cockpit voice recorder, a highly plausible inference to be taken from the evidence is that “the pilots did not even know precisely where they were in the sky.” Id. at 1142. Nonetheless, at one point during the phots’ discussion of the aircraft’s location Tafuri says, “You’re okay, you’re in good shape now,” which the district court concedes “can be read as an indication that Tafuri, at least, believed the plane was on or very near the published route.” Id. at 1143. The district court evaluated this statement by Tafuri