POSNER, Chief Judge.

The parents of Andrew Kuehn brought this diversity suit in a federal district court in Wisconsin both on his behalf and on their own behalf against a Los Angeles hospital and its liability insurer, Wisconsin being a direct-action state. Wis. Stat. sec. 803.04(2). (The insurance company's liability is purely derivative, so we can ignore it.) The district judge granted summary judgment for the defendants on the ground that under Wisconsin's conflict of law rules, which are applicable to this diversity suit because Wisconsin is the state in which the suit was brought, the law governing the plaintiffs' tort claims is California law, under which the plaintiffs have no right to relief. So he dismissed the suit, and the plaintiffs appeal, arguing that Wisconsin's conflict of law rules make Wisconsin law, not California law, applicable to the substantive issues in the suit.

The facts are gruesome. In 1994, Andrew, aged 2, was diagnosed as having neuroblastoma, an often fatal children's cancer. His parents enrolled him in an experimental treatment program offered by Childrens Hospital of Los Angeles in cooperation with the hospital of the University of Wisconsin. The treatment required Andrew to go to Childrens Hospital to have bone marrow removed from him, the marrow to be cleansed of cancer cells by the hospital and shipped back in a vacuum-sealed container to the University of Wisconsin hospital, where the marrow would be reinserted in him. His parents flew with him to Los Angeles and the bone marrow was removed without incident at Childrens Hospital. The removal of bone marrow is a painful process that requires drilling many holes in the patient's hips. After removal the marrow was duly treated and then sent back to the University of Wisconsin hospital via Federal Express. The container in which Childrens Hospital shipped the boy's marrow was defective, however; and in addition Childrens Hospital turned the container wrong side up, which made it more likely to tip over in transit, and also took no special measures, as it could easily have done, to ensure prompt delivery. Delivery was in fact delayed; and the delay, in combination with the container's having been shipped upside down, caused the marrow to arrive in a "thawed" state in which it was unusable. The parents were told they would have to take their son back to Childrens Hospital for more of his bone marrow to be removed. They flew him back and he underwent the painful procedure a second time. Andrew's father carried the newly "harvested" marrow back to Wisconsin personally, and it was reinserted into Andrew there.

The treatment was unsuccessful and Andrew died eight months later, but no evidence has been presented that the botch of the first marrow treatment and resulting delay in the reinsertion of the cleansed marrow hastened his death. As the case comes to us, shorn of some of its original claims, it is not for wrongful death or for the expenses of the treatment but only for the pain and suffering that Andrew underwent in the second extraction of marrow, which would have been unnecessary had Childrens Hospital not been negligent, and for the emotional distress that its negligence caused his parents. There is not even a claim that the hospital's negligence, while it cannot be shown to have hastened Andrew's death, deprived him of a chance of a longer life, viewed as a legally protected interest separate from life itself. See, e.g., Doll v. Brown, 75 F.3d 1200, 1205-06 (7th Cir. 1996); Perez v. Las Vegas Medical Center, 805 P.2d 589, 592 (Nev. 1991). Although California does not recognize the loss of a chance doctrine, Dumas v. Cooney, 1 Cal.Rptr.2d 584 (App. 1991), Wisconsin recognizes a variant of it. See Ehlinger v. Sipes, 454 N.W.2d 754, 762-63 (Wis. 1990).

We do not think that there is any conflict between California and Wisconsin law with respect to the second claim, for negligent infliction of emotional distress. The Kuehns concede that it is barred under California law. Under that law, the plaintiff must either have a very close preexisting relationship with the defendant, which in this case would require that one of the plaintiffs have been a patient of Childrens Hospital, Burgess v. Superior Court, 831 P.2d 1197, 1201, 1203-04 (Cal. 1992); Huggins v. Longs Drug Stores California, Inc., 862 P.2d 148 (Cal. 1993); Klein v. Children's Hospital Medical Center, 54 Cal.Rptr.2d 34 (App. 1996); Mercado v. Leong, 50 Cal.Rptr.2d 569, 574 n. 1 (App. 1996), or be "present at the scene of the injury[-]producing event at the time it occurs and [be] then aware that it is causing injury to the victim." Thing v. La Chusa, 771 P.2d 814, 829 (Cal. 1989) (footnote omitted). We don't think this means that the plaintiff must have observed the act of negligence that produced the accident, which in a products-liability case might require his presence at the design or manufacture of the product, but rather that he must witness the accident and not merely the injured condition of the accident victim, Ochoa v. Superior Court, 703 P.2d 1, 8 (Cal. 1985); Golstein v. Superior Court, 273 Cal.Rptr. 270, 278 (App. 1990); In re Air Crash Disaster, 967 F.2d 1421, 1425 (9th Cir. 1992) (applying California law), and our plaintiffs witnessed only the second. Otherwise a plaintiff could recover damages although he had merely heard about the accident and had then visited the victim in the hospital. But we need not pursue the issue, given the plaintiffs' concession.

The claim of negligent infliction of emotional distress is barred under Wisconsin law as well. Wisconsin like other states has been concerned about the expansionary potential of this tort. A serious injury often has a ripple effect. A child is killed; his mother is deeply shocked by the news; if he was killed as a result of someone's negligence, the mother's shock is a consequence of that someone's negligence; why shouldn't he be liable for it? It is this sort of thinking that gave rise to the tort in the first place. See, e.g., Thing v. La Chusa, supra, 771 P.2d at 829; Purcell v. St. Paul City Ry., 50 N.W. 1034 (Minn. 1892); W. Page Keeton et al., Prosser and Keeton on the Law of Torts sec. 54, pp. 360 (5th ed. 1984). The problem, which delayed the recognition of the tort and has circumscribed its scope, is the lack of a logical stopping point. Tens of millions of Americans were shocked by the assassination of President Kennedy; should they have been allowed to join in a class action against the FBI, the CIA, and the Secret Service for negligence in failing to anticipate and neutralize the threat that Lee Harvey Oswald posed to the President? Should every case of wrongful death give rise to a claim by the victim's immediate relatives? It is difficult to believe that the expense of determining damages in such cases and of screening out the inevitable phony cases would be justified by the incremental contribution to the deterrence of wrongful conduct; and these hypothetical cases that we have put are not ones in which an award of damages is necessary to restore the plaintiff to the standard of living he enjoyed before the tort.

But rather than say that only the direct victim may sue, modern judges, either more sentimental than their predecessors or more confident of their ability to screen out phony cases, have drawn lines — arbitrary ones, but serviceable. Wisconsin's lines are slightly different from California's. Wisconsin has drawn the line around the case in which the plaintiff is a close relative of the immediate victim (also a requirement under California law), the injury to the victim is fatal or very severe, and the plaintiff either witnessed the accident or arrived at the scene of the accident immediately afterward so that in either case he was witnessing an "extraordinary event, namely the incident and injury or the scene soon after the incident with the injured victim at the scene." Bowen v. Lumbermens Mutual Casualty Co., 517 N.W.2d 432, 435 (Wis. 1994). These criteria, it is true, are not designed for, and do not appear to be applicable to, cases in which the plaintiff is not a bystander but instead someone at whom, as in La Fleur v. Mosher, 325 N.W.2d 314 (Wis. 1982), the defendant's tortious conduct was directed. See Kleinke v. Farmers Coop. Supply Shipping, 549 N.W.2d 714, 716-17 (Wis. 1996). And the plaintiffs argue that they are direct victims, not bystanders (or not just bystanders), because they had a contract with the defendant. We cannot find any Wisconsin cases on this question, though California would clearly reject the argument. Huggins v. Longs Drug Stores California, Inc., supra, 862 P.2d at 153-54. To accept it would vitiate the limitations that both states have sought to place on the tort, by using the happenstance of a contractual relation to circumvent them. The parental relation between the plaintiffs and the victim in this case is far more significant than the contractual relationship with the defendant; and as a close family relationship is a sine qua non of a bystander claim, this case is in fact a classic bystander suit, so that to dispense with the limitations on such a suit because of the adventitious presence of a contract would be to fail to take seriously the desire of these states to impose such limitations.

So this is a bystander case; and the first condition imposed by the Wisconsin courts in such a case, that the plaintiff be a close relative of the immediate victim of the defendant's negligence, is satisfied here, but neither of the others is. If Andrew had been killed or crippled in an auto accident caused by someone's negligence, and Andrew's parents had witnessed the accident or been called to the scene before Andrew had been removed from it, they would have had a good claim against the injurer. Bowen v. Lumbermens Mutual Casualty Co., supra, 517 N.W.2d. at 435. The present case is remote from this example. The injury to Andrew from being forced to undergo the marrow extraction a second time was not severe. Once the possible effect on his longevity is set aside as unduly speculative, one is left with a procedure that is unpleasant and leaves the patient in some pain, but is not (so far as appears) dangerous, and certainly not crippling. Andrew recovered into the state in which he had been before the procedure within a week and a half of undergoing it. Of course, if one locates the source of the emotional distress to the parents not in the procedure but in the entire "surround," which includes the parents' horror of the thawing marrow, indignation at the carelessness of Childrens Hospital, and the anguish of witnessing the distress of their small child at the prospect of having to repeat the medical procedure, it becomes entirely credible that the parents' emotional distress was severe. But that is not the test. The test is whether they witnessed "either an incident causing death or serious injury or the gruesome aftermath of such an event." Id. at 444-45. They witnessed neither.

Although the claim of emotional distress thus must fail regardless of which state's law applies, there is an undeniable conflict between California and Wisconsin law with regard to the plaintiffs' other claim, the claim for damages for the pain and suffering that the second extraction inflicted on Andrew. In California, a claim for such damages does not survive the death of the victim. Cal. Civ. Proc. Code sec. 377.34; Sullivan v. Delta Air Lines, Inc., 935 P.2d 781, 783 (Cal. 1997). In Wisconsin it does. Wis. Stat. sec. 895.01(1); Wangen v. Ford Motor Co., 294 N.W.2d 437, 463 (Wis. 1980). So we must decide whether if this case had been brought in a Wisconsin state court rather than in a federal district court in the state, the state court would have applied the California or the Wisconsin rule on the survivability of claims for damages for pain and suffering. Answering this question is complicated by the fact that Wisconsin uses a laundry-list approach to conflicts questions. The list has five items: predictability of results, maintenance of interstate order (that is, not applying the law of a state that lacks a significant interest in how the case is decided), simplification of the court's task, advancement of the forum state's interests, and which state's law is better. Heath v. Zellmer, 151 N.W.2d 664, 672 (Wis. 1967); Lichter v. Fritsch, 252 N.W.2d 360, 363-64 (Wis. 1977); Employers Ins. of Wausau v. Certain Underwriters at Lloyd's, 552 N.W.2d 420, 427 (Wis.App. 1996). The items are not weighted, but the presumption is in favor of the forum state's, that is, Wisconsin's, law.

In choosing California law, the district judge emphasized the first and the last two items on Wisconsin's list. He thought that since the negligent act and injury had occurred (or at least had occurred mainly) in California, the parties would have expected California law to apply. And he thought that since Wisconsin's residents have a big stake in cancer research, and imposing tort liability on Childrens Hospital would retard such research, California's nonsurvival rule was both the better law and the one that actually served Wisconsin's interests better. The defendant adds that subjecting Childrens Hospital to the different tort rules of all 50 states (for the hospital draws its patients from all over the country) would make the hospital's legal obligations highly uncertain and therefore costly, and that California's rule on the nonsurvival of claims for pain and suffering is the better rule because it denies Andrew's parents a windfall, since it was his pain and suffering and he is dead.

Which state's law applies under the applicable conflict of law rules is a pure question of law. Gramercy Mills, Inc. v. Wolens, 63 F.3d 569, 572 (7th Cir. 1995); NL Industries, Inc. v. Commercial Union Ins. Co., 65 F.3d 314, 319 (3d Cir. 1995); Putnam Resources v. Pateman, 958 F.2d 448, 466 (1st Cir. 1992). We therefore cannot give any deference to the answer given by the district judge; the Supreme Court has held that a federal court of appeals is not to give weight to a determination of state law by a district judge even if he is sitting in that state and so is presumed to be especially familiar with its laws. Salve Regina College v. Russell, 499 U.S. 225 (1991); see also Stratmeyer v. United States, 67 F.3d 1340, 1345 (7th Cir. 1995); Control Data Corp. v. S.C.S.C. Corp., 53 F.3d 930, 939 (8th Cir. 1995). We must give our independent view and it is that Wisconsin's method of resolving conflicts points to Wisconsin, not California, law to govern this case. Because both states have substantial interests at stake — Wisconsin because it is the home of the tort victim, California because it is both the home of the defendant and the place where the defendant's careless acts occurred — the "interstate order" criterion drops out. Heath v. Zellmer, supra, 151 N.W.2d at 674. As for simplifying the judicial task, all that comes to mind is that of course if California law applies, barring the plaintiffs' only remaining claim and thus requiring the dismissal of their suit, the judicial task will be simplified, because it is easier to dismiss a case at the outset than to proceed to the summary judgment stage and perhaps to trial. We cannot imagine that such a consideration, systematically favoring defendants, is entitled to significant weight; it was given none in Lichter v. Fritsch, supra, 252 N.W.2d at 364.

That leaves us with three criteria to consider. The first criterion, predictability, might seem to point unerringly to California, because that's where the medical procedure that caused the child's pain and suffering was performed. Closer consideration shows the compass wobbling. The defendant is arguing in effect for lex loci delicti — the law of the place of the tort should govern. That was the reigning rule in the early part of this century, Zelinger v. State Sand Gravel Co., 156 N.W.2d 466, 468 (Wis. 1968); Albert A. Ehrenzweig, Conflicts in a Nutshell sec. 76-1, pp. 213-14 (3d ed. 1974), and had the virtue of relative certainty, and therefore predictability. It persists in attenuated form in some states, Ingersoll v. Klein, 262 N.E.2d 593, 595-96 (Ill. 1970); Rice v. Nova Biomedical Corp., 38 F.3d 909, 915 (7th Cir. 1994); Restatement (Second) of Conflict of Laws sec. 145(2)(a) (1969), but not, so far as we can find, in Wisconsin — and even if it did, this would not help the defendant. The place of the tort, for conflicts and other purposes, is not where the negligent act occurs but where the injury is felt. Rice v. Nova Biomedical Corp., supra, 38 F.3d at 915; Rozenfeld v. Medical Protective Co., 73 F.3d 154, 155 (7th Cir. 1996); Robert A. Leflar, American Conflicts Law sec. 133, p. 267 (3d ed. 1977). The negligent act occurred not at the time of the second extraction, which was conducted without negligence, but when the first extraction of marrow was shipped. The injury to Andrew occurred when that marrow arrived in unusable form; for at that point he had been deprived of cells that he needed in order to have a chance of extending his life. And he was in Wisconsin then. The return to California for the second procedure amounted to an effort by the tortfeasor to mitigate the harm by replacing the product that it had damaged in shipment. The pain and suffering from the second procedure occurred in California (where the marrow was extracted) and Wisconsin (where it was reinserted), but was not inflicted tortiously; the tortious injury had occurred in Wisconsin, earlier, when the botched shipment deprived Andrew of the marrow and thus necessitated the second extraction and reinsertion.

Suppose you broke your leg in Illinois but were sent across the state line to Wisconsin to have it set. If the setting of the bone though done without any negligence was painful, the pain would be deemed an incident of the injury in Illinois, and Illinois would be the locus delicti even if the only damages you were seeking were for the pain and suffering of the setting. It would be like a case in which the plaintiff, injured in one state, is treated in his state of domicile; the state of injury would still, we take it, be the locus delicti, though we cannot find a case on the question.

Since the tort occurred in Wisconsin rather than in California, the defendant cannot get any mileage out of the doctrine of lex loci delicti even if it retains some force in Wisconsin, of which we cannot find any evidence. This does not resolve the issue of predictability. But that issue can be resolved by noting that if Childrens Hospital, a sophisticated enterprise, had wanted predictability of outcome it would undoubtedly have included in its contract with the Kuehns pursuant to which Andrew was enrolled in the experimental bone marrow extraction and reinsertion program a clause specifying California as the state whose law would apply in the event of a suit growing out of the experimental treatment. Ever since informed consent became the rage, hospitals have been expert at including elaborate disclaimers in their contracts with patients. Choice of law clauses are common and when reasonable are enforced. E.g., Bush v. National School Studios, Inc., 407 N.W.2d 883, 886-87 (Wis. 1987); S.A. Healy Co. v. Milwaukee Metropolitan Sewerage District, 50 F.3d 476, 478 (7th Cir. 1995) (applying Wisconsin law); Nedlloyd Lines B.V. v. Superior Court, 834 P.2d 1148, 1151 (Cal. 1992); Home Ins. Co. v. Chicago Northwestern Transportation Co., 56 F.3d 763, 766 (7th Cir. 1995).

One can, it is true, find cases that say that contractual choice of law provisions govern only contractual disputes and not torts. E.g., Lazard Freres Co. v. Protective Life Ins. Co., 108 F.3d 1531, 1540 (2d Cir. 1997). But what the cases actually hold is that such a provision will not be construed to govern tort as well as contract disputes unless it is clear that this is what the parties intended. E.g., Krock v. Lipsay, 97 F.3d 640, 645 (2d Cir. 1996); Thompson Wallace of Memphis, Inc. v. Falconwood Corp., 100 F.3d 429, 432-33 (5th Cir. 1996); Caton v. Leach Corp., 896 F.2d 939, 942-43 (5th Cir. 1990). When it is clear, the provision is enforced. Turtur v. Rothschild Registry Int'l, Inc., 26 F.3d 304, 309-10 (2d Cir. 1994); Moses v. Business Card Express, Inc., 929 F.2d 1131, 1139-40 (6th Cir. 1991). These cases do not involve personal injury; but that is not a critical distinction. Even waivers of negligence are often enforced in personal-injury cases. Keeton et al., supra, sec. 68, pp. 482-84. And the issue here is not waiver but merely which state's tort law shall apply. It would have been difficult to accuse Childrens Hospital of overreaching or unconscionability had it stipulated in its contract with the Kuehns that California law would apply to any malpractice suit arising out of the contractually specified services, which were to be performed largely in that state. Thus, all the hospital had to do if it wanted to spare itself the uncertainty of having to defend under the law of a different state was to specify clearly and comprehensively in its contract that California law would apply to torts as well as contract disputes, and it did not do so.

A defendant's failure to negotiate a choice of law provision when as in this case the defendant has a written contract with the potential plaintiff makes the claim that applying the law of another state would unsettle the defendant's legal obligations ring hollow. It may indeed unsettle them but evidently the defendant does not much care. It is not as if Childrens Hospital could not have foreseen that it might be sued in another state — specifically Wisconsin. Its cooperative arrangement with the University of Wisconsin in the treatment program in which Andrew was enrolled clearly made it suable there. There is no suggestion, moreover, that exposing Childrens Hospital to the different tort laws of the different states would subject it to conflicting obligations. The only difference between California and Wisconsin tort law, so far as bears on this case at any rate, is in the scope of liability for negligence, not in the standard of care. It is not as if California had required one method of packing and shipping bone marrow and Wisconsin another.

So considerations of predictability do not strongly favor California law in this case, and the other considerations strongly favor Wisconsin law. Even if one thought that imposing slightly greater liability for negligence on Childrens Hospital would have a perceptible effect on the vigor of cancer research in the United States, which is exceedingly unlikely even apart from the fact that the negligence was in the shipping of a good rather than in the course of medical research, the injury to Wisconsin's residents from this imperceptible diminution in research would be outweighed by the state's interest in obtaining for its residents the measure of relief that the state believes appropriate in tort cases. Or so the Wisconsin courts would probably think. See Heath v. Zellmer, supra, 151 N.W.2d at 674; Lichter v. Fritsch, supra, 252 N.W.2d at 364; Spearing v. National Iron Co., 770 F.2d 87, 92 (7th Cir. 1985) (applying Wisconsin law); James A. White, Comment, "Stacking the Deck: Wisconsin's Application of Leflar's Choice-Influencing Considerations to Torts Choice-of-Law Cases," 1985 Wis. L. Rev. 401, 432. This means that the application of Wisconsin law will advance the forum state's interests. And it is not argued, as it perhaps might be, that a state which prefers its own interests to those of another state, rather than treating all states' interests impartially, is behaving questionably under the commerce, privileges and immunities, or other federalist clauses of the Constitution.

California's rule (nonsurvival) is not the better rule, moreover, but the worse rule or at least the more archaic one. The needs of cancer research cannot logically have any weight in the choice. That would imply that whether the Wisconsin courts applied California law rather than their own would depend on whether the defendant was thought to be engaged in work of peculiar merit — an approach that would eliminate all predictability and greatly complicate the judicial task. It would also put the Wisconsin courts and their federal surrogates in the role of selectively reviving the tort immunity of charitable institutions, even though the immunity has been rejected in Wisconsin as elsewhere. Kojis v. Doctors Hospital, 107 N.W.2d 131 (Wis. 1961); see also Keeton et al., supra, sec. 133, p. 1070; but cf. Note, "The Quality of Mercy: `Charitable Torts' and Their Continuing Immunity," 100 Harv. L. Rev. 1382, 1383-86 (1987).

The defendant is correct that if the victim is dead the award of damages for his pain and suffering constitutes a windfall — the award is to someone other than the victim. But that logic is too powerful; it implies that all tort suits should abate with the death of the victim, not just suits seeking damages for pain and suffering. That was the common law rule, but it was changed in the nineteenth century, T.A. Smedley, "Wrongful Death — Bases of the Common Law Rules," 13 Vand. L. Rev. 605, 624 (1960); Keeton et al., supra, secs. 125A, 126, pp. 940-42, and its survival (the survival of nonsurvival) in California with respect to pain and suffering is quaintly vestigial, rather than its being the better law and the law of the future. A few states retain this vestige, 3 Stuart M. Speiser, Charles F. Krause Juanita M. Madole, Recovery for Wrongful Death and Injury sec. 14:8, p. 33 n. 90 (3d ed. 1992), but we cannot find any articulated justification for it. Although the California legislature did in 1961 reject a recommendation by the state's law review commission to abolish it, Grimshaw v. Ford Motor Co., 174 Cal.Rptr. 348, 394 n. 24 (App. 1981), we do not know why the legislature rejected the recommendation. So it is not as if what we are describing as a vestige is actually the first stage in a new wave of tort reform.

The objection to making a tort suit abate with the death of the victim is that it gives the injurer an incentive to make a clean kill and reduces the deterrent effect of tort law by eliminating any tort sanction for inflicting fatal injuries. The objection is diminished when the rule of abatement is limited as California has done to one item of damages. But it reappears in that situation as an objection to drawing arbitrary distinctions with less cause than in the case of negligent infliction of emotional distress with which we began. So strongly do the other considerations besides predictability favor Wisconsin law in this case that the application of that law was predictable — thus completing the sweep. So saying, we acknowledge our inability to understand the precise bearing of predictability in a conflicts analysis, save perhaps in a contract case in which the law has changed since the parties made their contract. One predicts the resolution of a conflicts issue by applying the conflicts rule, so to consider predictability as an additional component of the rule is circular. No matter. It is reasonably clear that Wisconsin law is applicable. The dismissal of the emotional distress claim is affirmed, but the dismissal of the pain and suffering claim is reversed and the case remanded for further proceedings consistent with this opinion.

AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.