dissenting:
One of the complications of diversity jurisdiction is that the federal courts are called upon to make pronouncements upon state law that may later prove wrong. On the highly unusual facts before us, this wrongful death case presents a question of New York state law that will control the outcome of this appeal, as to which there is no controlling precedent. See Local Rule § 0.27; N.Y. Comp.Codes R. & Regs. Tit. 22, § 500.17(a) (McKinney’s 2000 N.Y. Rules of Court). Because I believe the question of state law should be certified to the New York State Court of Appeals, I respectfully dissent.
The majority believes that two decisions of the New York Court of Appeals, Turcotte v. Fell, 68 N.Y.2d 432, 510 N.Y.S.2d 49, 502 N.E.2d 964 (1986), and Morgan v. State, 90 N.Y.2d 471, 662 N.Y.S.2d 421, 685 N.E.2d 202 (1997), govern the outcome of this appeal. According to the majority, these cases require us to hold that when Goodlett participated in the air race, he assumed any risks that “flow from such participation,” Morgan, 90 N.Y.2d at 484, 662 N.Y.S.2d 421, 685 N.E.2d 202, including the risk of collision after the race was over. However, Turcotte explains that among the factors to be considered in assessing whether-such a plaintiff assumed the risk are “the ultimate purpose of the game and the method or methods of winning it[, and] the relationship of the defendant’s conduct to the game’s ultimate purpose, especially his conduct with respect to rules and customs whose purpose it is to enhance the safety of the participants.” Turcotte, 68 N.Y.2d at 440, 510 N.Y.S.2d 49, 502 N.E.2d 964. The district court judge, with over a decade of experience interpreting New York law as a state court judge, concluded that the doctrine of primary assumption of the risk would have barred plaintiffs action had the collision occurred during the race itself. However, because the collision occurred after the race finished, the judge concluded that the doctrine was inapplicable, and that Good-lett’s assumption of the risk, if any, went to the issue of comparative fault, which was a question for the jury. Thus, the judge noted that
The competition was over. The flying in a constricted narrow low area at high speed was concluded. The pilots were getting ready to land. A normal procedure with an airplane. They were in the same order as at the conclusion of the race, very reasonable and logical and apparently safe.
The judge further noted that “it is a different, completely different aura after the race is over. There is no more competition, and they are maneuvering to land, which happens many times.” The judge obviously believed that after the race was won and its ultimate purpose achieved, the competitors’ interests are no longer ad*40verse, both are under a duty to conclude the competition safely, and the pilots no longer assumed the risks inherent in the competition.1 On this theory, the judge refused to take from the jury the issues before it. The jury then found that Good-lett was 60% at fault and .Kalishek 40% at fault for the accident, which resulted in a judgment for plaintiff against Kalishek of $390,213.
The majority is “confident” that the New York Court of Appeals would hold that the present action is barred by the doctrine of primary assumption of the risk. I do not share that confidence. Indeed, it is not at all clear to me that the judge erred in concluding that the case should go to the jury. It is certainly true, as the majority notes, that all pilots were aware of the risks of participation in air races. However, the evidence suggests that a different set of expectations applied to post-race flight: (1) according to the National Transportation Safety Board, there have been no airplane accidents after the end of a race; (2) defendant Kalishek admitted that he was never aware of any risk of a post-race collision; and (3) Kalishek’s expert testified that he was not aware of any post-race collisions either. After all, the Association’s guide for new pilots warns that, “The race pilot’s greatest fear is a mid-air collision between two race-planes during a race.” (Emphasis added.)
I would certify to the New York Court of Appeals the question whether, on the facts before us, Goodlett assumed the risk of collision after the race was over, thereby completely barring this action by his executrix. If we do not certify the issue now, the New York Court of Appeals can at any time in the future reject the majority’s interpretation of New York law, as it has occasionally done in the past. Compare, e.g., DeWeerth v. Baldinger, 836 F.2d 103, 109-10 (2d Cir.1987), with Solomon R. Guggenheim Found. v. Lubell, 77 N.Y.2d 311, 318, 567 N.Y.S.2d 623, 569 N.E.2d 426 (1991); compare also Factors Etc., Inc. v. Pro Arts, Inc., 652 F.2d 278, 283 (2d Cir.1981) (interpreting Tennessee law), with Elvis Presley Int’l Mem. Found. v. Crowell, 733 S.W.2d 89, 97 (Tenn.Ct.App.1987) (rejecting this court’s interpretation). If it does so, the decedent’s estate will have been denied the benefit of the ruling of the state’s highest court on a controlling issue of state law. See DeWeerth v. Baldinger, 38 F.3d 1266, 1273-74 (2d Cir.1994) (rejecting motion for relief from the judgment based on this Court’s error in interpreting New York law); see also id. at 1273-74 (“The very nature of diversity jurisdiction leaves open the possibility that a state court will subsequently disagree with a federal court’s interpretation of state law.”).
One of the basic reasons for a certification statute is to avoid the possibility of our misinterpreting state law. See Recommendation of the Law Revision Commission to the 1984 Legislature: Relating to Certification of Questions of Law to the Courts of Appeals, 1984 N.Y. Laws 2976, 2979 (McKinney’s 1985) (“[T]he federal courts have sometimes failed to predict accurately the state position.”) (citing Henry J. Friendly, Federal Jurisdiction: A General View 141-42 (1973)); see also Friendly at 143 (“All such cases [in which state law is unclear] are pregnant with the possibility of injustice.”). Of course, should the New York Court of Appeals for any reason find it inappropriate to answer the question of state law, it is free to decline to answer, as it has done in the past. See, e.g., Tunick v. Safir, 94 N.Y.2d 709, 709 N.Y.S.2d 881, 731 N.E.2d 597 (2000), declining to answer question certi*41fied by Tunick v. Safir, 209 F.3d 67 (2d Cir.2000).
I see no persuasive reason why, in a situation where the stakes are high and the law is arguably unclear, we should not try to make sure that we properly understand New York law before proceeding further. For these reasons, I decline to join the majority opinion and respectfully dissent.
. Morgan does not refute the "finish line” distinction drawn by the district court because the plaintiff in Morgan did not sue a fellow bobsledder, but the owner of the bobsled track. See Morgan, 90 N.Y.2d at 481, 662 N.Y.S.2d 421, 685 N.E.2d 202. While the relationship between a sledder and a track operator remains the same before, during, and after the race, the relationship between competitors changes greatly once the competition is over.