In 1996 David Boim, a Jewish teenager who was both an Israeli citizen and an American citizen, living in Israel, was shot to death by two men at a bus stop near Jerusalem. His parents filed this suit four years later, alleging that his killers had been Hamas gunmen and naming as defendants Muhammad Salah plus three organizations: the Holy Land Foundation for Relief and Development, the American Muslim Society, and the Quranic Literacy Institute. (A fourth, the Islamic Associa*688tion of Palestine-National, appears to be either an alter ego of the American Muslim Society or just an alternative name for it, and need not be discussed separately. There are other defendants as well but they are not involved in the appeals.) The complaint accused the defendants of having provided financial support to Hamas before David Boim’s death and by doing so of having violated 18 U.S.C. § 2333(a), which provides that “any national of the United States injured in his or her person, property, or business by reason of an act of international terrorism, or his or her estate, survivors, or heirs, may sue therefor in any appropriate district court of the United States and shall recover threefold the damages he or she sustains and the cost of the suit, including attorney’s fees.”

The district court denied the defendants’ motion to dismiss the complaint for failure to state a claim, 127 F.Supp.2d 1002 (N.D.Ill. 2001); the defendants had argued that providing financial assistance to a terrorist group is not an act of international terrorism and therefore is not within the scope of section 2333. We authorized an interlocutory appeal, 28 U.S.C. § 1292(b), and the panel that heard the appeal affirmed the district court. Boim v. Quranic Literacy Institute, 291 F.3d 1000 (7th Cir. 2002). The case then resumed in that court. The court granted summary judgment in favor of the plaintiffs with respect to the liability of the three defendants other than the Quranic Literacy Institute. 340 F.Supp.2d 885 (N.D.Ill. 2004). A jury was convened and, after a trial lasting a week, found the Institute — which having filed a statement of “nonparticipation” attended but did not participate in the trial — liable. The jury then assessed damages of $52 million against all the defendants (including the ones not before us) jointly and severally. The amount was then trebled and attorneys’ fees added.

These defendants again appealed, this time from a final judgment. The panel vacated the judgment and directed the district court to redetermine liability. 511 F.3d 707 (7th Cir. 2007). Judge Evans agreed with the reversal as to the Holy Land Foundation but otherwise dissented.

The plaintiffs petitioned for rehearing en banc, and the full court granted the petition, primarily to consider the elements of a suit under 18 U.S.C. § 2333 against financial supporters of terrorism. The parties have filed supplemental briefs. A number of amici curiae have weighed in as well, including the Department of Justice, which has taken the side of the plaintiffs.

The first panel opinion rejected the argument that the statute does not impose liability on donors to groups that sponsor or engage in terrorism. The supplemental briefs do not revisit the issue, and at oral argument counsel for Salah and the Holy Land Foundation disclaimed reliance on their former position concerning the liability of donors. But in a letter to the court after oral argument, Salah’s counsel indicated that the disclaimer had been based solely on a belief that the doctrine of law of the case foreclosed any further consideration of the statutory issue in this court. That was a mistake. The full court can revisit any ruling by a panel. All arguments that the defendants have presented in their appeals are open today — and will be open in the Supreme Court. It is better to decide the question than to leave it hanging; why bother to address the elements of a legal claim that may not exist? Before deciding what a plaintiff must prove in order to recover from a donor under section 2333, we should decide whether the statute applies. United States National Bank of Oregon v. Insurance Agents of America, Inc., 508 U.S. *689439, 445-48, 113 S.Ct. 2173, 124 L.Ed.2d 402 (1993).

Section 2333 does not say that someone who assists in an act of international terrorism is liable; that is, it does not mention “secondary” liability, the kind that 18 U.S.C. § 2 creates by imposing criminal liability on “whoever commits an offense against the United States or aids, abets, counsels, commands, induces or procures its commission,” or “willfully causes an act to be done which if directly performed by him or another would be an offense against the United States.” See also 18 U.S.C. § 3 (accessory after the fact). The Supreme Court in Central Bank of Denver, N.A. v. First Interstate Bank of Denver, N.A., 511 U.S. 164, 114 S.Ct. 1439, 128 L.Ed.2d 119 (1994), held that section 10(b) of the Securities and Exchange Act of 1934, which prohibits securities fraud, does not reach aiding and abetting because it makes no reference to secondary liability, the kind of liability that statutes such as 18 U.S.C. §§ 2 and 3 create in criminal cases. The Court discussed the securities laws at length, but nothing in its holding turns on particular features of those laws.

So statutory silence on the subject of secondary liability means there is none; and section 2333(a) authorizes awards of damages to private parties but does not mention aiders and abettors or other secondary actors. Nevertheless the first panel opinion concluded that section 2333 does create secondary liability. It distinguished Central Bank of Denver as having involved an implied private right of action (for it was a private suit, yet section 10(b) does not purport to authorize such suits), while section 2333(a) expressly creates a private right. But as the dissenting Justices in Central Bank of Denver had pointed out, the majority’s holding was not limited to private actions. 511 U.S. at 200, 114 S.Ct. 1439. It encompassed suits by the SEC, which section 10(b) authorizes expressly.

Congress agreed with this understanding of Central Bank of Denver, for the next year it enacted 15 U.S.C. § 78t(e) to allow the SEC in section 10(b) suits to obtain relief against aiders, abettors, and others who facilitate primary violations. Stoneridge Investment Partners, LLC v. Scientific-Atlanta, Inc., — U.S. -, 128 S.Ct. 761, 771-72, 169 L.Ed.2d 627 (2008). The enactment of section 78t(e) would have been pointless had Central Bank of Denver allowed secondary liability to be imposed in suits, such as suits by the SEC under section 10(b), that the statute expressly authorizes. Years later, reaffirming Central Bank of Denver, the Supreme Court repeated that the earlier decision had not been limited to private suits under section 10(b). Stoneridge Investment Partners, LLC v. Scientific-Atlanta, Inc., supra, 128 S.Ct. at 768-69.

The first panel opinion relied on Harris Trust & Savings Bank v. Salomon Smith Barney, Inc., 530 U.S. 238, 120 S.Ct. 2180, 147 L.Ed.2d 187 (2000), an ERISA case involving an application of trust law. Trust law permits trust beneficiaries to maintain actions against third parties who have received trust assets improperly. ERISA not only does not upset this principle of trust law; it authorizes the Secretary of Labor to penalize third parties who “knowing[ly] participat[e]” in a fiduciary’s misconduct. 29 U.S.C. §§ 1106(a), 1132(i)(1)(B). Harris Trust did not cite Central Bank of Denver and did not purport to limit its holding. Stoneridge, decided eight years after Harris Trust, also did not treat Harris Trust as circumscribing Central Bank of Denver—it did not even cite Harris Trust.

To read secondary liability into section 2333(a), moreover, would enlarge the *690federal courts’ extraterritorial jurisdiction. The defendants are accused of promoting terrorist activities abroad. Congress has the power to impose liability for acts that occur abroad but have effects within the United States, F. Hoffmann-La Roche Ltd. v. Empagran S.A., 542 U.S. 155, 165, 124 S.Ct. 2359, 159 L.Ed.2d 226 (2004), but it must make the extraterritorial scope of a statute clear. Small v. United States, 544 U.S. 385, 388-89, 125 S.Ct. 1752, 161 L.Ed.2d 651 (2005); EEOC v. Arabian American Oil Co., 499 U.S. 244, 248, 111 S.Ct. 1227, 113 L.Ed.2d 274 (1991).

The first panel opinion discussed approvingly an alternative and more promising ground for bringing donors to terrorist organizations within the grasp of section 2333. The ground involves a chain of explicit statutory incorporations by reference. The first link in the chain is the statutory definition of “international terrorism” as “activities that ... involve violent acts or acts dangerous to human life that are a violation of the criminal laws of the United States,” that “appear to be intended ... to intimidate or coerce a civilian population” or “affect the conduct of a government by ... assassination,” and that “transcend national boundaries in terms of the means by which they are accomplished” or “the persons they appear intended to intimidate or coerce.” 18 U.S.C. § 2331(1). Section 2331 was enacted as part of the Federal Courts Administration Act of 1992, Pub.L. No. 102-572, § 1003(a)(3), 106 Stat. 4506, 4521. Section 2333 (having been originally enacted in 1990 and repealed for a technical reason the next year) was reenacted in , 1992 as part of that same Federal Courts Administration Act. So the two sections are part of the same statutory scheme and are to be read together. Nicholas J. Perry, “The Numerous Federal Legal Definitions of Terrorism: The Problem of Too Many Grails,” 30 J. Legis. 249, 257 (2004).

Section 2331(l)’s definition of international terrorism (amended in 2001 by the PATRIOT Act, Pub.L. No. 107-56, § 802(a)(1), 115 Stat. 272, 376, but in respects irrelevant to this case) includes not only violent acts but also “acts dangerous to human life that are a violation of the criminal laws of the United States.” Giving money to Hamas, like giving a loaded gun to a child (which also is not a violent act), is an “act dangerous to human life.” And it violates a federal criminal statute enacted in 1994 and thus before the murder of David Boim — 18 U.S.C. § 2339A(a), which provides that “whoever provides material support or resources ..., knowing or intending that they are to be used in preparation for, or in carrying out, a violation of [18 U.S.C. § 2332],” shall be guilty of a federal crime. So we go to 18 U.S.C. § 2332 and discover that it criminalizes the killing (whether classified as homicide, voluntary manslaughter, or involuntary manslaughter), conspiring to kill, or inflicting bodily injury on, any American citizen outside the United States.

By this chain of incorporations by reference (section 2333(a) to section 2331(1) to section 2339A to section 2332), we see that a donation to a terrorist group that targets Americans outside the United States may violate section 2333. Which makes good sense as a counterterrorism measure. Damages are a less effective remedy against terrorists and their organizations than against their financial angels. Terrorist organizations have been sued under section 2333, e.g., Ungar v. Palestine Liberation Organization, 402 F.3d 274 (1st Cir. 2005); Biton v. Palestinian Interim Self-Government Authority, 252 F.R.D. 1 (D.D.C. 2008); Knox v. Palestine Liberation Organization, 248 F.R.D. 420 (S.D.N.Y. 2008), but to collect a damages *691judgment against such an organization, let alone a judgment against the terrorists themselves (if they can even be identified and thus sued), is, as the first panel opinion pointed out, 291 F.3d at 1021, well-nigh impossible. These are foreign organizations and individuals, operating abroad and often covertly, and they are often impecunious as well. So difficult is it to obtain monetary relief against covert foreign organizations like these that Congress has taken to passing legislation authorizing the payment of judgments against them from U.S. Treasury funds. E.g., Victims of Trafficking and Violence Protection Act of 2000, Pub.L. No. 106-386, § 2002, 114 Stat. 1464. But that can have no deterrent or incapacitative effect, whereas suits against financiers of terrorism can cut the terrorists’ lifeline.

And whether it makes good sense or not, the imposition of civil liability through the chain of incorporations is compelled by the statutory texts — as the panel determined in its first opinion. 291 F.3d at 1012-16. But in addition the panel placed a common law aiding and abetting gloss on section 2333. The panel was worried about a timing problem: section 2339A was not passed until 1994, and the defendants’ contributions to Hamas began earlier. But that is not a serious problem on the view we take of the standard for proving causation under section 2333; we shall see that the fact of contributing to a terrorist organization rather than the amount of the contribution is the keystone of liability.

Only because this is a very old case — David Boim was killed 12 years ago — does the 1994 effective date of section 2339A, two years before his killing, present an obstacle to liability, though only with respect to Salah and possibly the Holy Land Foundation (but we are vacating the judgment against the latter anyway, as we shall explain). For there is no doubt that the other defendants made contributions after section 2339A’s effective date. Salah, however, having been arrested by Israeli authorities in 1993 and not released until 1997, did not render material support to Hamas between the effective date of section 2339A and Boim’s killing, so the judgment against him must be reversed. New future cases will be affected by the timing issue, because few such cases will involve donations that were made after section 2333 was enacted in 1990 or re-enacted in 1992 but that ceased before 1994.

In addition to providing material support after the effective date of section 2339A, a donor to terrorism, to be liable under section 2333, must have known that the money would be used in preparation for or in carrying out the killing or attempted killing of, conspiring to kill, or inflicting bodily injury on, an American citizen abroad. We know that Hamas kills Israeli Jews; and Boim was an Israeli citizen, Jewish, living in Israel, and therefore a natural target for Hamas. But we must consider the knowledge that the donor to a terrorist organization must be shown to possess in order to be liable under section 2333 and the proof required to link the donor’s act to the injury sustained by the victim. The parties have discussed both issues mainly under the rubrics of “conspiracy” and “aiding and abetting.” Although those labels are significant primarily in criminal cases, they can be used to establish tort liability, see, e.g., Halberstam v. Welch, 705 F.2d 472 (D.C.Cir. 1983); Restatement (Second) of Torts §§ 876(a), (b) (1979), and there is no impropriety in discussing them in reference to the liability of donors to terrorism under section 2333 just because that liability is primary. Primary liability in the form of material support to terrorism has the character of secondary liability. *692Through a chain of incorporations by reference, Congress has expressly imposed liability on a class of aiders and abettors.

When a federal tort statute does not create secondary liability, so that the only defendants are primary violators, the ordinary tort requirements relating to fault, state of mind, causation, and foreseeability must be satisfied for the plaintiff to obtain a judgment. See, e.g., Bridge v. Phoenix Bond & Indemnity Co., — U.S. -, 128 S.Ct. 2131, 2141-44, 170 L.Ed.2d 1012 (2008); Stoneridge Investment Partners, LLC v. Seientific-Atlanta, Inc., supra, 128 S.Ct. at 769; Holmes v. Securities Investor Protection Corp., 503 U.S. 258, 268-69, 112 S.Ct. 1311, 117 L.Ed.2d 532 (1992); Associated General Contractors of California, Inc. v. California State Council of Carpenters, 459 U.S. 519, 103 S.Ct. 897, 74 L.Ed.2d 723 (1983). But when the primary liability is that of someone who aids someone else, so that functionally the primary violator is an aider and abettor or other secondary actor, a different set of principles comes into play. Those principles are most fully developed in the criminal context, but we must be careful in borrowing from criminal law because the state-of-mind and causation requirements in criminal cases often differ from those in civil cases. For example, because the criminal law focuses on the dangerousness of a defendant’s conduct, the requirement of proving that a criminal act caused an injury is often attenuated and sometimes dispensed with altogether, as in the statutes that impose criminal liability on providers of material support to terrorism (18 U.S.C. §§ 2339A, B, and C), which do not require proof that the material support resulted in an actual terrorist act, or that punish an attempt (e.g., 18 U.S.C. § 1113) that the intended victim may not even have noticed, so that there is no injury. The law of attempt has no counterpart in tort law, United States v. Gladish, 536 F.3d 646, 648 (7th Cir. 2008), because there is no tort without an injury. E.g., Rozenfeld v. Medical Protective Co., 73 F.3d 154, 155-56 (7th Cir. 1996); Winskunas v. Bimbaum, 23 F.3d 1264, 1267 (7th Cir. 1994).

So prudence counsels us not to halt our analysis with aiding and abetting but to go on and analyze the tort liability of providers of material support to terrorism under general principles of tort law. We begin by noting that knowledge and intent have lesser roles in tort law than in criminal law. A volitional act that causes an injury gives rise to tort liability for negligence if the injurer failed to exercise due care, period. But more is required in the case of intentional torts, and we can assume that since section 2333 provides for an automatic trebling of damages it would require proof of intentional misconduct even if the plaintiffs in this case did not have to satisfy the state-of-mind requirements of sections 2339A and 2332 (but they do).

Punitive damages are rarely if ever imposed unless the defendant is found to have engaged in deliberate wrongdoing. “Something more than the mere commission of a tort is always required for punitive damages. There must be circumstances of aggravation or outrage, such as spite or ‘malice,’ or a fraudulent or evil motive on the part of the defendant, or such a conscious and deliberate disregard of the interests of others that the conduct may be called wilful or wanton.” W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 2, pp. 9-10 (5th ed. 1984); see, e.g., Molzof v. United States, 502 U.S. 301, 305-07, 112 S.Ct. 711, 116 L.Ed.2d 731 (1992); Kemezy v. Peters, 79 F.3d 33, 35 (7th Cir. 1996). Treble damages too, not being compensatory, tend to have a punitive aim. “The very idea of treble damages reveals an intent to punish past, and to deter future, unlawful conduct.” Texas *693Industries, Inc. v. Radcliff Materials, Inc., 451 U.S. 630, 639, 101 S.Ct. 2061, 68 L.Ed.2d 500 (1981); see also Vermont Agency of Natural Resources v. United States ex rel. Stevens, 529 U.S. 765, 784-86, 120 S.Ct. 1858, 146 L.Ed.2d 836 (2000); Zelinski v. Columbia 300, Inc., 335 F.3d 633, 642 (7th Cir. 2003); Gorenstein Enterprises, Inc. v. Quality Care-USA, Inc., 874 F.2d 431, 435-36 (7th Cir. 1989); United States v. Mackby, 261 F.3d 821, 830-31 (9th Cir. 2001).

To give money to an organization that commits terrorist acts is not intentional misconduct unless one either knows that the organization engages in such acts or is deliberately indifferent to whether it does or not, meaning that one knows there is a substantial probability that the organization engages in terrorism but one does not care. “When the facts known to a person place him on notice of a risk, he cannot ignore the facts and plead ignorance of the risk.” Makor Issues & Rights, Ltd. v. Tellabs Inc., 513 F.3d 702, 704 (7th Cir. 2008). That is recklessness and equivalent to recklessness is “wantonness,” which “has been defined as the conscious doing of some act or omission of some duty under knowledge of existing conditions and conscious that from the doing of such act or omission of such duty injury will likely or probably result.” Graves v. Wildsmith, 278 Ala. 228, 177 So.2d 448, 451 (1965); see also Landers v. School District No. 203, O’Fallon, 66 Ill.App.3d 78, 22 Ill.Dec. 837, 383 N.E.2d 645 (1978). “[I]n one case we read that ‘willful and wanton misconduct approaches the degree of moral blame attached to intentional harm, since the defendant deliberately inflicts a highly unreasonable risk of harm upon others in conscious disregard of it.’ Similarly, [another case] defines ‘willful and wanton’ as exhibiting ‘an utter indifference to or conscious disregard for’ safety.” Fagocki v. Algonquin/Lake-in-the-Hills Fire Protection District, 496 F.3d 623, 627 (7th Cir. 2007) (citations omitted).

So it would not be enough to impose liability on a donor for violating section 2333, even if there were no state-of-mind requirements in sections 2339A and 2332, that the average person or a reasonable person would realize that the organization he was supporting was a terrorist organization, if the actual defendant did not realize it. That would just be negligence. But if you give a gun you know is loaded to a child, you know you are creating a substantial risk of injury and therefore your doing so is reckless and if the child shoots someone you will be liable to the victim. See Pratt v. Martineau, 69 Mass.App.Ct. 670, 870 N.E.2d 1122 (2007); Bowen v. Florida, 791 So.2d 44, 48-49 (Fla.App. 2001). That case should be distinguished from one in which the gun is given to an adult without adequately explaining the dangers — a case of negligent entrustment. To give a small child a loaded gun would be a case of criminal recklessness and therefore satisfy the state of mind requirement for liability under section 2333 and the statutes that it incorporates by reference. For the giver would know he was doing something extremely dangerous and without justification. “If the actor knows that the consequences are certain, or substantially certain, to result from his act, and still goes ahead, he is treated by the law as if he had in fact desired to produce the result.” Restatement, supra, § 8A, comment b. That you did not desire the child to shoot anyone would thus be irrelevant, not only in a tort case, see EEOC v. Illinois, 69 F.3d 167, 170 (7th Cir. 1995), but in a criminal case. United States v. Fountain, 768 F.2d 790, 798 (7th Cir. 1985); cf. United States v. Ortega, 44 F.3d 505, 508 (7th Cir. 1995).

A knowing donor to Hamas — that is, a donor who knew the aims and activities of *694the organization — would know that Hamas was gunning for Israelis (unlike some other terrorist groups, Hamas’s terrorism is limited to the territory of Palestine, including Israel; see Council on Foreign Relations, “Hamas,” www.cfr.org/publication/ 8968/, visited Nov. 16, 2008), that Americans are frequent visitors to and sojourners in Israel, that many U.S. citizens live in Israel (American Citizens Abroad, an advocacy group for expatriates, reports on the basis of State Department data that in 1999 there were about 184,000 American citizens living in Israel, accounting for about 3.1 percent of the country’s population, www.aca.ch/amabroad.pdf, visited Nov. 16, 2008), and that donations to Ha-mas, by augmenting Hamas’s resources, would enable Hamas to kill or wound, or try to kill, or conspire to kill more people in Israel. And given such foreseeable consequences, such donations would “appear to be intended ... to intimidate or coerce a civilian population” or to “affect the conduct of a government by ... assassination,” as required by section 2331(1) in order to distinguish terrorist acts from other violent crimes, though it is not a state-of-mind requirement; it is a matter of external appearance rather than subjective intent, which is internal to the inten-der.

It is true that “the word ‘recklessness’ in law covers a spectrum of meaning, ranging from gross negligence in an accident case to the conduct of a robber in shooting at a pursuing policeman without aiming carefully.” Wright v. United States, 809 F.2d 425, 427 (7th Cir. 1987). In tort law it sometimes connotes merely gross negligence and at other times requires only that the defendant have acted in the face of an unreasonable risk that he should have been aware of even if he wasn’t. But when, as in the passages we have quoted both from judicial opinions and from the Restatement, recklessness entails actual knowledge of the risk, the tort concept merges with the criminal concept, which likewise “generally permits a finding of recklessness only when a person disregards a risk of harm of which he is aware.” Farmer v. Brennan, 511 U.S. 825, 837, 114 S.Ct. 1970, 128 L.Ed.2d 811 (1994); see also Desnick v. American Broadcasting Cos., 233 F.3d 514, 517-518 (7th Cir. 2000); American Law Institute, Model Penal Code § 2.02(2)(c) (1962) (defining recklessness as “consciously disregarding] a substantial and unjustifiable risk that the material element exists or will result from his conduct”).

Critically, the criminal (like the tort) concept of recklessness is more concerned with the nature and knowledge of the risk that the defendant creates than with its magnitude. The Court in Farmer v. Brennan spoke of an “excessive” risk, a “significant” risk, a “substantial” risk, and an “intolerable” risk, 511 U.S. at 837-38, 842-43, 846, 114 S.Ct. 1970, the Model Penal Code of a “substantial and unjustifiable” risk, and the Restatement of an “unreasonable” risk, Restatement, supra, § 500, rather than assigning a minimum probability to the risk. These are relative terms; what is excessive, intolerable, etc., depends on the nature of the defendant’s conduct. Ordinarily, it is true, the risk is great in a probabilistic sense; for the greater it is, the more likely it is to materialize and so give rise to a lawsuit or a prosecution and thus be mentioned in a judicial opinion. The greater the risk, moreover, the more obvious it will be to the risk taker, enabling the trier of fact to infer the risk taker’s knowledge of the risk with greater confidence, see, e.g., Farmer v. Brennan, supra, 511 U.S. at 842, 114 S.Ct. 1970; Duckworth v. Franzen, 780 F.2d 645, 652 (7th Cir. 1985), though, as the Farmer decision emphasizes, subject to rebuttal. 511 U.S. at 837-42, 114 S.Ct. 1970.

*695But probability isn’t everything. The risk that one of the workers on a project to build a bridge or a skyscraper will be killed may be greater than the risk that a driver will be killed by someone who flings rocks from an overpass at the cars traveling on the highway beneath. But only the second risk, though smaller, is deemed excessive and therefore reckless. McNabb v. State, 887 So.2d 929, 974-75 (Ala.Crim. App. 2001). (The first risk might not even be negligent.) As we explained in United States v. Boyd, 475 F.3d 875, 877 (7th Cir. 2007) (emphasis added), “firing multiple shots from a powerful gun ... in the downtown of a large city at a time when pedestrians ... are known to be in the vicinity creates a risk of harm that, while not large in 'probabilistic terms, is ‘substantial’ relative to the gratuitousness of the defendant’s actions.... An activity is reckless when the potential harm that it creates ... is wildly disproportionate to any benefits that the activity might be expected to confer.... The emotional gratification that defendant Boyd derived from shooting into the night, though perhaps great, is not the kind of benefit that has weight in the scales when on the other side is danger to life and limb, even if the danger is limited, as it was here.” Lennon v. Metropolitan Life Ins. Co., 504 F.3d 617, 623 (6th Cir. 2007), says that the risk must be “weighed against the lack of social utility of the activity” in adjudging its reasonableness. See also Orban v. Vaughn, 123 F.3d 727, 733 (3d Cir. 1997).

So if you give a person rocks who has told you he would like to kill drivers by dropping them on cars from an overpass, and he succeeds against the odds in killing someone by this means, you are guilty of providing material support to a murderer, or equivalently of aiding and abetting — for remember that when the primary violator of a statute is someone who provides assistance to another he is functionally an aider and abettor. The mental element required to fix liability on a donor to Hamas is therefore present if the donor knows the character of that organization.

The Court also said in Farmer v. Brennan that it was no defense that “he [a particular prison official] did not know that the complainant was especially likely to be assaulted by the specific prisoner who eventually committed the assault.” 511 U.S. at 843, 114 S.Ct. 1970. That brings us to our next question — the standard of causation in a suit under section 2333.

It is “black letter” law that tort liability requires proof of causation. But like much legal shorthand, the black letter is inaccurate if treated as exceptionless. We made that point explicitly, with the aid of an example, in Maxwell v. KPMG LLP, 520 F.3d 713, 716 (7th Cir. 2008): “when two fires join and destroy the plaintiffs property and each one would have destroyed it by itself and so was not a necessary condition ... each of the firemakers (if negligent) is [nevertheless] liable to the plaintiff for having ‘caused’ the injury. Kingston v. Chicago & N.W. Ry., 191 Wis. 610, 211 N.W. 913 (Wis. 1927)” (emphasis added); see also United States v. Feliciano, 45 F.3d 1070, 1075 (7th Cir. 1995). (A “necessary condition” is another term for a “but for” cause. Maxwell v. KPMG LLP, supra, 520 F.3d at 716.)

The multiple-fire example and the principle that subtends it were explained at greater length in United States v. Johnson, 380 F.3d 1013, 1016 (7th Cir. 2004): “[T]wo defendants each start a fire, and the fires join and destroy the plaintiffs house; either fire, however, would have destroyed his house. Each defendant could therefore argue that he should not be hable for the damage because it would have occurred even if he had not set his fire; but the law rejects the argument.... *696[I]n the famous old ease of Cook v. Minneapolis, St. Paul & Sault Ste. Mane Ry., 98 Wis. 624, 74 N.W. 561, 564 (1898), we read that ‘it is no defense for a person against whom negligence which causes damages is established, to prove that without fault on his part the same damage would have resulted from the negligent act of the other, but each is responsible for the entire damage.’ See also Anderson v. Minneapolis, St. Paul & Sault Ste. Marie Ry., 146 Minn. 430, 179 N.W. 45, 49 (1920); Collins v. American Optometric Ass’n, 693 F.2d 636, 640 n. 4 (7th Cir. 1982); Housing 21, L.L.C. v. Atlantic Home Builders Co., 289 F.3d 1050, 1056-57 (8th Cir. 2002); Sanders v. American Body Armor & Equipment, Inc., 652 So.2d 883, 884-85 (Fla.App. 1995); Garrett v. Grant School Dist. No. 124, 139 Ill.App.3d 569, 93 Ill.Dec. 874, 487 N.E.2d 699, 706 (1985); Hart v. Browne, 103 Cal.App.3d 947, 163 Cal.Rptr. 356, 363-64 (1980); W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 41, pp. 266-67 (5th ed. 1984). The tort-feasor cannot avoid liability by pointing to an alternative unlawful cause of the damage that he inflicted.... [S]ince neither fire was a sine qua non of the plaintiffs injury, it could be argued that neither fire maker had committed a tort. Tort law rejects this conclusion for the practical reason that tortious activity that produces harm would go unsanctioned otherwise.” The Prosser treatise also recognizes the multiple-fire case as one in which the plaintiff is not required to prove “but for” causation. Keeton et al., supra, § 41, pp. 266-68; cf. Edward J. Schwartzbauer and Sidney Shindell, “Cancer and The Adjudicative Process: The Interface of Environmental Protection and Toxic Tort Law,” 14 Am. J.L. & Med. 1, 31-32 (1988).

In the fire cases the acts of each defendant are sufficient conditions of the resulting injury, though they are not necessary conditions (that is, they are not but-for causes). But in Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1 (1948), where two hunters negligently shot their rifles at the same time and a third hunter was hit by one of the bullets, it could not be determined which hunter’s gun the bullet had come from and so it could not be proved by a preponderance of the evidence that either of the shooters was the injurer in either a sufficient-condition or a necessary-condition sense; for each hunter, the probability that he had caused the injury was only 50 percent, since one of the shots had missed. Nevertheless both defendants were held jointly and severally liable to the injured person. See Restatement, supra, § 433B(3) and comment f; Smith v. Cutter Biological, 72 Haw. 416, 823 P.2d 717, 725 (1991); In re “Agent Orange” Product Liability Litigation, 597 F.Supp. 740, 822-23 (E.D.N.Y. 1984).

Similarly, if several firms spill toxic waste that finds its way into groundwater and causes damage to property but it is impossible to determine which firm’s spill caused the damage, all are liable. See, e.g., Chem-Nuclear Systems, Inc. v. Bush, 292 F.3d 254, 259-60 (D.C.Cir. 2002); United States v. Alcan Aluminum Corp., 964 F.2d 252, 267-69 (3d Cir. 1992); Michie v. Great Lakes Steel Division, 495 F.2d 213, 217-18 (6th Cir. 1974); Landers v. East Texas Salt Water Disposal Co., 151 Tex. 251, 248 S.W.2d 731, 734 (1952); Phillips Petroleum Co. v. Hardee, 189 F.2d 205, 211-12 (5th Cir. 1951); 2 Frank P. Grad, Treatise on Environmental Law § 3.02 (2007); Kenneth S. Abraham, “The Relation Between Civil Liability and Environmental Regulation: An Analytical Overview,” 41 Washburn L.J. 379, 386-87 (2002). Even if the amount of pollution caused by each party would be too slight to warrant a finding that any one of them had created a nuisance (the common law basis for treating pollution as a tort), “pol*697lution of a stream to even a slight extent becomes unreasonable [and therefore a nuisance] when similar pollution by others makes the condition of the stream approach the danger point. The single act itself becomes wrongful because it is done in the context of what others are doing.” Keeton et al., supra, § 52, p. 354.

In all these cases the requirement of proving causation is relaxed because otherwise there would be a wrong and an injury but no remedy because the court would be unable to determine which wrongdoer inflicted the injury. If “each [defendant] bears a like relationship to the event” and “each seeks to escape liability for a reason that, if recognized, would likewise protect each other defendant in the group, thus leaving the plaintiff without a remedy,” the attempt at escape fails; each is liable. Id., § 41, p. 268.

But we must consider the situation in which there is uncertainty about the causal connection between the wrongful conduct of all potential tortfeasors and the injury. Suppose in our first case that there was a third fire, of natural origin (the result of a lightning strike, perhaps), and it alone might have sufficed to destroy the plaintiffs house. One might think the law would require the plaintiff to prove that it was more likely than not that had it not been for the defendants’ negligence, his house would not have burned down — the Are of natural origin would have petered out before reaching it. Instead the law requires proof only that there was a substantial probability that the defendants’ fires (or rather either of them) were the cause. See, e.g., Anderson v. Minneapolis, St. Paul & Sault Ste. Marie Ry., supra, 179 N.W. at 46; Restatement, supra, § 432(2) (“if two forces are actively operating, one because of the actor’s negligence, the other not because of any misconduct on his part, and each of itself is sufficient to bring about harm to another, the actor’s negligence may be found to be a substantial factor in bringing it about”); see also id., illustration 3.

Our final example is Keel v. Hainline, 331 P.2d 397 (Okla. 1958). Thirty to forty junior high school students showed up one day for their music class, but the instructor failed to show so the kids began throwing wooden erasers, chalk, and even a Coke bottle at each other. One of the students was struck in the eye by an eraser, and sued. One of the defendants, Keel, apparently had not thrown anything. But he had retrieved some of the erasers after they had been thrown and had handed them back to the throwers. There was no indication that Keel had handed the eraser to the kid who threw it at the plaintiff and injured her, but the court deemed that immaterial. It was enough that Keel had participated in the wrongful activity as a whole. He thus was liable even though there was no proven, or even likely, causal connection between anything he did and the injury. “ ‘One who commands, directs, advises, encourages, procures, instigates, promotes, controls, aids, or abets a wrongful act by another has been regarded as being as responsible as the one who commits the act so as to impose liability upon the former to the same extent as if he had performed the act himself.’ ” Id. at 401. The court did not use the term “material support,” but in handing erasers to the throwers Keel was providing them with material support in a literal sense. It was enough to make him liable that he had helped to create a danger; it was immaterial that the effect of his help could not be determined — that his acts could not be found to be either a necessary or a sufficient condition of the injury.

The cases that we have discussed do not involve monetary contributions to a wrongdoer. But then criminals and other *698intentional tortfeasors do not usually solicit voluntary contributions. Terrorist organizations do. But this is just to say that terrorism is sui generis. So consider an organization solely involved in committing terrorist acts and a hundred people all of whom know the character of the organization and each of whom contributes $1,000 to it, for a total of $100,000. The organization has additional resources from other, unknown contributors of $200,000 and it uses its total resources of $300,000 to recruit, train, equip, and deploy terrorists who commit a variety of terrorist acts one of which kills an American citizen. His estate brings a suit under section 2333 against one of the knowing contributors of $1,000. The tort principles that we have reviewed would make the defendant jointly and severally liable with all those other contributors. The fact that the death could not be traced to any of the contributors (as in the example the Supreme Court gave in Farmer v. Brennan) and that some of them may have been ignorant of the mission of the organization (and therefore not liable under a statute requiring proof of intentional or reckless misconduct) would be irrelevant. The knowing contributors as a whole would have significantly enhanced the risk of terrorist acts and thus the probability that the plaintiffs decedent would be a victim, and this would be true even if Hamas had incurred a cost of more than $1,000 to kill the American, so that no defendant’s contribution was a sufficient condition of his death.

This case is only a little more difficult because Hamas is (and was at the time of David Boim’s death) engaged not only in terrorism but also in providing health, educational, and other social welfare services. The defendants other than Salah directed their support exclusively to those services. But if you give money to an organization that you know to be engaged in terrorism, the fact that you earmark it for the organization’s nonterrorist activities does not get you off the liability hook, as we noted in a related context in Hussain v. Mukasey, 518 F.3d 534, 538-39 (7th Cir. 2008); see also Singh-Kaur v. Ashcroft, 385 F.3d 293, 301 (3d Cir. 2004). The reasons are twofold. The first is the fungibility of money. If Hamas budgets $2 million for terrorism and $2 million for social services and receives a donation of $100,000 for those services, there is nothing to prevent its using that money for them while at the same time taking $100,000 out of its social services “account” and depositing it in its terrorism “account.” Kilbum v. Socialist People’s Libyan Arab Jamahiriya, 376 F.3d 1123, 1130 (D.C.Cir. 2004).

Second, Hamas’s social welfare activities reinforce its terrorist activities both directly by providing economic assistance to the families of killed, wounded, and captured Hamas fighters and making it more costly for them to defect (they would lose the material benefits that Hamas provides them), and indirectly by enhancing Ha-mas’s popularity among the Palestinian population and providing funds for indoctrinating schoolchildren. See, e.g., Justin Magouirk, “The Nefarious Helping Hand: Anti-Corruption Campaigns, Social Service Provision, and Terrorism,” 20 Terrorism & Political Violence 356 (2008); Eli Berman & David D. Laitin, “Religion, Terrorism, and Public Goods: Testing the Club Model” 7-10 (National Bureau of Econ. Research Working Paper No. 13725, 2008). Anyone who knowingly contributes to the nonviolent wing of an organization that he knows to engage in terrorism is knowingly contributing to the organization’s terrorist activities. And that is the only knowledge that can reasonably be required as a premise for liability. To require proof that the donor intended that his contribution be used for terrorism — to make a benign intent a defense — would as *699a practical matter eliminate donor liability except in cases in which the donor was foolish enough to admit his true intent. It would also create a First Amendment Catch-22, as the only basis for inferring intent would in the usual case be a defendant’s public declarations of support for the use of violence to achieve political ends.

Although liability under section 2333 is broad, to maintain perspective we note two cases that fall on the other side of the liability line. One is the easy case of a donation to an Islamic charity by an individual who does not know (and is not reckless, in the sense of strongly suspecting the truth but not caring about it) that the charity gives money to Hamas or some other terrorist organization.

The other case is that of medical (or other innocent) assistance by nongovernmental organizations such as the Red Cross and Doctors Without Borders that provide such assistance without regard to the circumstances giving rise to the need for it. Suppose an Israeli retaliatory strike at Hamas causes so many casualties that the local medical services cannot treat all of them, and Doctors Without Borders offers to assist. And suppose that many of the casualties that the doctors treat are Hamas fighters, so that Doctors Without Borders might know in advance that it would be providing medical assistance to terrorists.

However, section 2339A(b)(l) excludes “medicine” from the definition of “material resources.” And even if the word should be limited (an issue on which we take no position) to drugs and other medicines, an organization like Doctors Without Borders would not be in violation of section 2333. It would be helping not a terrorist group but individual patients, and, consistent with the Hippocratic Oath, with no questions asked about the patients’ moral virtue. It would be like a doctor who treats a person with a gunshot wound whom he knows to be a criminal. If doctors refused to treat criminals, there would be less crime. But the doctor is not himself a criminal unless, besides treating the criminal, he conceals him from the police (like Dr. Samuel Mudd, sentenced to prison for trying to help John Wilkes Booth, Lincoln’s assassin, elude capture) or violates a law requiring doctors to report wounded criminals. The same thing would be true if a hospital unaffiliated with Hamas but located in Gaza City solicited donations.

Nor would the rendering of medical assistance by the Red Cross or Doctors Without Borders to individual terrorists “appear to be intended ... to intimidate or coerce a civilian population” or “affect the conduct of a government by ... assassination,” and without such appearance there is no international terrorist act within the meaning of section 2331(1) and hence no violation of section 2333. Nor is this point limited to the rendering of medical assistance. For example, UNRWA (the United Nations Relief and Works Agency for Palestine Refugees in the Near East) renders aid to Palestinian refugees that is not limited to medical assistance to individual refugees, www.un.org/unrwa/english.html (visited Nov. 16, 2008). But so far as one can glean from its website (see id. and www.un.org/unrwa/allegations/index.html, also visited Nov. 16, 2008), it does not give money to organizations, which might be affiliates of Hamas or other terrorist groups; it claims to be very careful not to employ members of Hamas or otherwise render any direct or indirect aid to it. Id.

To the objection that the logic of our analysis would allow the imposition of liability on someone who with the requisite state of mind contributed to a terrorist organization in 1995 that killed an American abroad in 2045, we respond first that *700that is not this case — the interval here was at most two years (1994, when section 2339A was enacted, to 1996, when Boim was killed) — and second that the imposition of liability in the hypothetical case would not be as outlandish, given the character of terrorism, as one might think. (There would of course be no defense of statute of limitations, since the limitations period would not begin to run until the tort was committed, and that would not occur until the injury on which suit was based was inflicted.) Terrorism campaigns often last for many decades. Think of Ireland, Sri Lanka, the Philippines, Colombia, Kashmir — and Palestine, where Arab terrorism has been more or less continuous since 1920. Seed money for terrorism can sprout acts of violence long after the investment. In any event, whether considerations of temporal remoteness might at some point cut off liability is not an issue we need try to resolve in this case.

An issue to which the first panel opinion gave much attention (see 291 F.3d at 1021-27), but which received little attention from the parties afterward, is brought into focus by our analysis of the elements of a section 2333 violation. That is whether the First Amendment insulates financiers of terrorism from liability if they do not intend to further the illegal goals of an organization like Hamas that engages in political advocacy as well as in violence. If the financier knew that the organization to which it was giving money engaged in terrorism, penalizing him would not violate the First Amendment. Otherwise someone who during World War II gave money to the government of Nazi Germany solely in order to support its anti-smoking campaign could not have been punished for supporting a foreign enemy.

But it is true that “an organization is not a terrorist organization just because one of its members commits an act of armed violence without direct or indirect authorization, even if his objective was to advance the organization’s goals, though the organization might be held liable to the victim of his violent act.” Hussain v. Mukasey, supra, 518 F.3d at 538. That is the principle of NAACP v. Claiborne Hardware Co., 458 U.S. 886, 920, 102 S.Ct. 3409, 73 L.Ed.2d 1215 (1982). The defendants in the present case could not be held liable for acts of violence by members of Hamas that were not authorized by Hamas. Nor would persons be liable who gave moral rather than material support, short of incitement, to violent organizations that have political aims. As intimated earlier in this opinion, a person who gives a speech in praise of Hamas for firing rockets at Israel is exercising his freedom of speech, protected by the First Amendment. See, e.g., Communist Party of Indiana v. Whitcomb, 414 U.S. 441, 447-49, 94 S.Ct. 656, 38 L.Ed.2d 635 (1974); Brandenburg v. Ohio, 395 U.S. 444, 447-48, 89 S.Ct. 1827, 23 L.Ed.2d 430 (1969) (per curiam). But as Hamas engages in violence as a declared goal of the organization, anyone who provides material support to it, knowing the organization’s character, is punishable (provided he is enchained by the chain of statutory incorporations necessary to impose liability under section 2333) whether or not he approves of violence.

Enough about the liability standard. We have now to consider its application to the facts. That turns out to be straightforward, except with respect to one of the defendants, the Holy Land Foundation, about which we can be brief because of the thoroughness of the panel’s consideration. See 511 F.3d at 720-33. A principal basis for the district court’s finding that the Foundation had violated the statute was the court’s giving collateral estop-pel effect to findings made in Holy Land Foundation for Relief & Development v. *701Ashcroft, 219 F.Supp.2d 57 (D.D.C. 2002), affirmed, 333 F.3d 156 (D.C.Cir. 2003). The panel was unanimous that this ruling was erroneous.

In 2001 the Secretary of the Treasury determined that the Foundation “acts for or on behalf of’ Hamas, and an order freezing the Foundation’s funds was issued. The Foundation sued in the District of Columbia. The district court there found that the Secretary’s finding was not “arbitrary and capricious” (the standard of review) and upheld the blocking order. Although the court recited extensive evidence that the Foundation knew that Ha-mas was and had long been a terrorist organization, 219 F.Supp.2d at 69-75, and it appears that most or perhaps all of the evidence related to its knowledge before 1996 when David Boim was killed, the validity of the blocking order did not depend on the Foundation’s knowledge. 511 F.3d at 731; see Executive Order 13244, 66 Fed.Reg. 49079 (Sept. 23, 2001); Garry W. Jenkins, “Soft Power, Strategic Security, and International Philanthropy,” 85 N. Car. L.Rev. 773, 808-09 (2007); Jennifer Lynn Bell, “Terrorist Abuse of Non-Profits and Charities: A Proactive Approach to Preventing Terrorist Financing,” 17 Kan. J.L. & Public Policy 450, 458-59 (2008). If someone is giving money to an organization that the government knows to be a terrorist organization, any subsequent gift can be blocked whether or not the donor knows (or agrees with the government concerning) the nature of the recipient.

Even if the decision of the district court in the District of Columbia were read as finding that the Foundation knew that Ha-mas was a terrorist organization (and, as the court also found, that the Holy Land Foundation made contributions to Hamas after the effective date of 18 U.S.C. § 2339A, 219 F.Supp.2d at 70-71), such a finding would not have been essential to the judgment upholding the blocking order — and essentiality is at the heart of collateral estoppel. Arizona v. California, 530 U.S. 392, 414, 120 S.Ct. 2304, 147 L.Ed.2d 374 (2000); Montana v. United States, 440 U.S. 147, 159, 99 S.Ct. 970, 59 L.Ed.2d 210 (1979); H-D Michigan, Inc. v. Top Quality Service, Inc., 496 F.3d 755, 760 (7th Cir. 2007); Central Hudson Gas & Electric Corp. v. Empresa Naviera Santa S.A., 56 F.3d 359, 368 (2d Cir. 1995); Restatement (Second) of Judgments § 27 (1982). If a finding is unnecessary to the judgment, the appellant has no reason to challenge it and if he does the appellate court has no reason to review it because it is irrelevant to the appeal — and so the appellant would not have his (full) day in court.

So the judgment against the Foundation must be reversed and the case against it remanded for further proceedings to determine its liability. The judgment against Salah must also be reversed, as we explained earlier. Regarding the remaining defendants, the American Muslim Society and the Quranic Literacy Institute, the judgment of the district court was in our view correct. The activities of the American Muslim Society are discussed at length in the district court’s second opinion. See 340 F.Supp.2d at 906-13. There we learn that while its activities included donating money to the Holy Land Foundation, there was much else besides. Moreover, the fact that the Foundation may not have known that Hamas was a terrorist organization (implausible as that is) would not exonerate the American Muslim Society, which did know and in giving money to the Foundation was deliberately tunneling money to Hamas. The funnel doesn’t have to know what it’s doing to be an effective funnel.

Nor should donors to terrorism be able to escape liability because terrorists and *702their supporters launder donations through a chain of intermediate organizations. Donor A gives to innocent-appearing organization B which gives to innocent-appearing organization C which gives to Hamas. As long as A either knows or is reckless in failing to discover that donations to B end up with Hamas, A is liable. Equally important, however, if this knowledge requirement is not satisfied, the donor is not liable. And as the temporal chain lengthens, the likelihood that a donor has or should know of the donee’s connection to terrorism shrinks. But to set the knowledge and causal requirement higher than we have done in this opinion would be to invite money laundering, the proliferation of affiliated organizations, and two-track terrorism (killing plus welfare). Donor liability would be eviscerated, and the statute would be a dead letter.

With regard to the Quranic Literacy Institute, the district court, after denying the Institute’s motion for summary judgment, 340 F.Supp.2d at 929, submitted the case against the Institute to a jury trial but instructed the jury that Hamas was responsible for the murder of David Boim. The jury was left to decide whether the Institute had knowingly provided material support to Hamas. The jury found the Institute liable. By deciding not to participate in the trial, the Institute waived any objection it might have had to the jury instructions or the jury’s findings.

In any event, the only factual determination underlying the judgment against the Institute, as against the American Muslim Society, that might be questioned — and was by the panel — was the determination, made by the district court on summary judgment, that Hamas had been responsible for the murder. The panel thought that the district judge had considered inadmissible evidence that the two terrorists who shot Boim were in fact members of Hamas.

Here is the panel’s critique of the principal though not only evidence of their membership:

To show that the murder of David Boim was the work of Hamas, the Boims submitted the declaration of Dr. Ruven [sic ] Paz, a former member of the Israeli security community who describes himself as an expert in terrorism and counter-terrorism, Islamic movements in the Arab and Islamic world, Palestinian Islamic groups, and Palestinian society and politics. Based on his review of various exhibits submitted in connection with this case, his independent research, and his knowledge of how Hamas and other Islamic terror organizations operate, Paz concluded that Hinawi and Al-Sharif had murdered David Boim, that Hinawi and Al-Sharif were members of Hamas at the time they killed Boim, and that Hamas itself had accepted responsibility for the murder....

In concluding that Al-Sharif was a member of Hamas and that Hamas had taken responsibility for the murder, Paz relied heavily on information set forth on certain websites that he attributed to Hamas. Paz explained that Hamas publicly acknowledges its terrorist acts and identifies its “martyrs” as a way to promote itself and to recruit new members. According to Paz, internet websites are a means by which Hamas disseminates such information. Paz’s declaration asserts that scholars, journalists, and law enforcement routinely rely on the website postings of terrorist organizations for what they reveal about the activities of those organizations. Looking to certain websites whose content he asserts is controlled by Hamas, Paz found statements indicating that Hamas had taken responsibility for the Beit-El attack that *703took David Boim’s life and that Al-Sharif was one of the participants in this attack. Paz repeated these statements in his declaration.

Paz’s reliance upon, and his recounting of, internet website postings demand a certain caution in evaluating his prospective testimony. Such postings would not be admissible into evidence for their truth absent proper authentication, and this would typically require some type of proof that the postings were actually made by the individual or organization to which they are being attributed — in this case, Hamas — as opposed to others with access to the website. Paz’s declaration identifies the websites from which he quotes as ones controlled by Hamas, but it does not describe the basis for his conclusion, and consequently his declaration does not permit any independent assessment of the purported links between these sites and Hamas and the source of the postings that he recounts. Of course, the rules of evidence do not limit what type of information an expert may rely upon in reaching his opinion; even if that information would not otherwise be admissible in a court proceeding, an expert witness may rely upon it so long as it is the type of information on which others in the field reasonably rely. Indeed, Rule 703 now expressly permits the expert to disclose such information to the jury, provided the court is satisfied that its helpfulness in evaluating the expert’s opinion substantially outweighs its prejudicial effect. Nonetheless, a judge must take care that the expert is not being used as a vehicle for circumventing the rule against hearsay. Where, as here, the expert appears to be relying to a great extent on web postings to establish a particular fact, and where as a result the factfinder would be unable to evaluate the soundness of his conclusion without hearing the evidence he relied on, we believe the expert must lay out, in greater detail than Paz did, the basis for his conclusion that these websites are in fact controlled by Hamas and that the postings he cites can reasonably and reliably be attributed to Hamas.

Paz’s conclusion that Hinawi was responsible for the murder of David Boim was based in significant part on two documents related to Hinawi’s trial and sentencing by a Palestinian Authority tribunal: (1) a set of notes prepared by a U.S. foreign service officer who attended Hinawi’s trial in February 1998, and (2) an Arabic-language document purporting to be the written verdict reflecting Hinawi’s conviction and sentence. The foreign service officer’s notes indicate that Hinawi was tried in open proceedings for participating in a terrorist act and acting as an accomplice in the killing of David Boim, that he was afforded counsel by the tribunal, that he contended in his defense that his friend AI-Sharif was the gunman and that Al-Sharif exploited his friendship with Hi-nawi by asking him to drive the car, and that he was convicted on both charges and sentenced to ten years. Paz’s declaration accepts these documents as genuine and relies principally on them for the proposition that Hinawi participated in David Boim’s murder and was convicted by the Palestinian Authority tribunal for the same.

Once again we have concerns about whether the record as it stands lays an appropriate foundation for these documents. We can assume that the report of a U.S. government official who, in the course of his duties, observed a trial in a foreign tribunal may constitute proof of what occurred in that proceeding. We also have no doubt that a properly authenticated, official report of a judgment *704issued by a foreign tribunal constitutes adequate proof of that judgment. The difficulty we have with Paz’s reliance upon these documents is that they have not been properly authenticated. The foreign service officer’s notes are unsigned and reveal nothing about the circumstances under which they were prepared. The document that we are told is the official verdict is entirely in Arabic, is not readily evident as an official document, and is unaccompanied by an English translation. There is a single cover note, on the letterhead of the U.S. Consulate General in Jerusalem, which accompanies these documents and explains what they are. But the cover note itself is unsigned and does not even identify its author. This is unacceptable. We assume that Paz knows more about these documents and that he would not have relied upon them if he had doubts about their authenticity. But given that Paz relies almost exclusively on these documents as proof of Hinawi’s complicity in Boim’s murder, and because a factfinder could not evaluate the soundness of Paz’s conclusion without knowing what these documents say, an appropriate foundation must be laid for these documents before the conclusions that Paz has drawn from these documents may be admitted.

511 F.3d at 752-54 (citations omitted).

We accept the panel majority’s description of the infirmities of the evidence on which Reuven Paz (formerly research director of Shin Bet, Israel’s domestic security agency) based his expert opinion. But we do not agree that the district court abused its discretion in allowing the opinion into evidence. As the quoted passage acknowledges (albeit grudgingly, in its warning against using an expert witness “as a vehicle for circumventing the rule against hearsay”), an expert is not limited to relying on admissible evidence in forming his opinion. Fed.R.Evid. 703; Wendler & Ezra, P.C. v. American Int’l Group, Inc., 521 F.3d 790, 791 (7th Cir. 2008); In re James Wilson Associates, 965 F.2d 160, 172-73 (7th Cir. 1992); United States v. Locascio, 6 F.3d 924, 938 (2d Cir. 1993). That would be a crippling limitation because experts don’t characteristically base their expert judgments on legally admissible evidence; the rules of evidence are not intended for the guidance of experts. Biologists do not study animal behavior by placing animals under oath, and students of terrorism do not arrive at their assessments solely or even primarily by studying the records of judicial proceedings. Notice, moreover, that there was no need for the plaintiffs to prove that both Al-Sherif and Hinawi were complicit in Boim’s death; if either was complicit and a member of Hamas, that is enough to fix responsibility on Hamas for killing Boim.

In dissenting from the panel’s ruling Judge Evans offered an assessment of Paz’s evidence (see 511 F.3d at 758) that we find persuasive. An expert on terrorism in the Arab world, fluent in Arabic, Paz explained that the websites of Islamic movements and Islamic terrorist organizations have long been accepted by security experts as valid, important, and indeed indispensable sources of information. Terrorist organizations rely on the web to deliver their messages to their adherents and the general public. The United States Institute for Peace, a nonpartisan federal institution created by Congress, published an extensive report, submitted to the district court along with Paz’s declaration, on the use of the Internet by terrorists. And — critically—the defendants presented no evidence to contradict Paz: no evidence that the killing of Boim was not a Hamas hit. Had they thought Paz had mistranslated the Arabic judgment against Hinawi, they could have provided the district court *705with their own translation. Had they doubted that Paz can identify a Hamas website (he gave the web addresses of several of them), they could have presented testimony to that effect. Paz’s 12-page declaration is detailed, concrete, and backed up by a host of exhibits. The district court did not abuse its discretion in admitting his evidence; and with it in the record and nothing on the other side the court had no choice but to enter summary judgment for the plaintiffs with respect to Hamas’s responsibility for the Boim killing.

To summarize, the judgment of the district court is affirmed except with respect to (1) Salah, as to whom the judgment is reversed with instructions to enter judgment in his favor; (2) the Holy Land Foundation, as to which the judgment is reversed and the case remanded for further proceedings consistent with this opinion; and (3) the award of attorneys’ fees— for we adopt the panel’s criticisms of that award, 511 F.3d at 749-50, and anyway the award will have to be adjusted because of the further proceedings on remand that we are ordering.

AFFIRMED IN PART, REVERSED IN PART, AND Remanded.

with whom WILLIAMS, Circuit Judge, joins, concurring in part and dissenting in part.1

At this late stage in the litigation, we are now turning to a fundamental question: Are we going to evaluate claims for terrorism-inflicted injuries using traditional legal standards, or are we going to rewrite tort law on the ground that “terrorism is sui generis ”? Ante at 698. My colleagues in the majority have opted to “relaxf ]” — I would say eliminate — the basic tort requirement that causation be proven, believing that “otherwise there would be a wrong and an injury but no remedy because the court would be unable to determine which wrongdoer inflicted the injury.” Ante at 697. The choice is a false one. The panel took pains to identify a number of ways in which the plaintiffs might establish a causal link between the defendants’ financial contributions to (and other support for) Hamas and the murder of David Boim. Boim II, 511 F.3d at 741-43. It is not the case that the plaintiffs were unable show causation, it is rather that they did not even make an attempt; and that was the purpose of the panel’s decision to remand the case.2 But rather than requiring the plaintiffs to present evidence of causation and allowing the factfin-der to determine whether causation has been shown, the majority simply deems it a given, declaring as a matter of law that any money knowingly given to a terrorist organization like Hamas is a cause of terrorist activity, period. This sweeping rule of liability leaves no role for the factfinder to distinguish between those individuals and organizations who directly and purposely finance terrorism from those who are many steps removed from terrorist activity and whose aid has, at most, an indirect, uncertain, and unintended effect on terrorist activity. The majority’s approach treats all financial support provided to a terrorist organization and its affiliates as support for terrorism, regardless of whether the money is given to the terrorist organization itself, to a charitable entity controlled by that organization, or to an intermediary organization, and regardless of what the money is actually used to do.

*706The majority’s opinion is remarkable in two additional respects. By treating all those who provide money and other aid to Hamas as primarily rather than secondarily liable — along with those who actually commit terrorist acts — the majority eliminates any need for proof that the aid was given with the intent to further Hamas’s terrorist agenda. Besides eliminating yet another way for the factfinder to distinguish between those who deliberately aid terrorism from those who do so inadvertently, this poses a genuine threat to First Amendment freedoms. Finally, the majority sustains the entry of summary judgment on a basic factual question — Did Hamas kill David Boim? — based on an expert’s affidavit that both relies upon and repeats multiple examples of hearsay. Rather than sustain the panel’s unexceptional demand that the expert’s sources be proven reliable, consistent with Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 597, 113 S.Ct. 2786, 2799, 125 L.Ed.2d 469 (1993), the majority gives its blessing to circumventing the rules of evidence altogether.

Thus, although I concur in the decision to remand for further proceedings as to HLF, I otherwise dissent from the court’s decision.

1.

One point of clarification at the outset. The majority’s opinion reads as though the defendants were writing checks to Hamas, perhaps with a notation on the memo line that read “for humanitarian purposes.” If indeed the defendants were directing money into a central Hamas fund out of which all Hamas expenses — whether for humanitarian or teiTorist activities — were paid, it would be easy to see that the defendants were supporting Hamas’s terrorism even if their contributions were earmarked for charity. In fact, the case is not as simple as that. For example, much of the money that defendant HLF provided to Hamas apparently was directed not to Hamas per se but to a variety of zakat committees and other charitable entities, including a hospital in Gaza, that were controlled by Ha-mas. See Holy Land Found, for Relief & Dev. v. Ashcroft, 219 F.Supp.2d 57, 70-71 (D.D.C. 2002), j. aff'd, 333 F.3d 156 (D.C.Cir. 2003).3 I gather that this is a distinction without a difference in the majority’s view, and certainly I agree that if the zakat committees and other recipients of HLF’s funding were mere fronts for Hamas or were used to launder donations targeted for Hamas generally, then those donations ought to be treated as if they were direct donations to Hamas itself.4 But to the extent that these Hamas subsidiary organizations actually were engaged solely in humanitarian work and HLF was sending its money to those subsidiaries to support that work, HLF is one or more significant steps removed from the direct financing of terrorism and the case for HLF’s liability for terrorism is, in my view, a much less compelling one. Defendant AMS is yet another step removed, in that AMS is alleged to have contributed money not to Hamas but to HLF.

Moreover, the type of support that can give rise to civil liability is not limited to financial support. As the panel discussed

*707in Boim I, civil liability under section 2333(a) can result from the provision of “material support or resources” to terrorism and to terrorist organizations as prohibited by 18 U.S.C. §§ 2339A and 2339B, see 291 F.3d at 1012-17, and “material support or resources” is defined broadly to include not only weapons and money but “any property, tangible or intangible, or service,” including such things as lodging, expert advice, training, and personnel. § 2339A(b)(1). Notably, the plaintiffs have sought to hold AMS liable, and the district court found it liable, not simply for the financial support it provided to HLF, but for various types of pro-Hamas advocacy, such as hosting Hamas speakers at its conferences, publishing sympathetic editorials in its newsletter, and the like. See Boim v. Quranic Literacy Inst., 340 F.Supp.2d 885, 908-13 (N.D.Ill. 2004).

So the majority’s rule has the potential to sweep within its reach not only those who write checks to Hamas and the organizations that it controls but also individuals and groups who support Hamas and its affiliates in myriad other ways, including those who advocate on Hama’s behalf. My point is not that there is no case to be made for imposing liability on such supporters for Hamas’s terrorist acts. My point is simply that the basis for their liability is not nearly as clean and straightforward as it might seem from the majority’s opinion.

2.

The majority has chosen to evaluate the prospective liability of the defendants in this case through the lens of primary liability, reasoning that those who provide financial and other aid to terrorist organizations are themselves engaging in terrorism and thus may be held liable on the same basis as those who actually commit terrorist acts.5 In formulating its theory of primary liability, the majority relies in part upon section 2331(l)’s definition of “international terrorism” and partly upon section 2339A(a)’s criminal proscription against providing material support or resources to terrorists. Treating the defendants as primarily rather than secondarily liable enables the majority to accomplish two things: First, it compensates for what the majority believes was Congress’s failure in section 2333(a) to authorize the imposition of secondary liability on those who aid or abet terrorist acts or conspire with terrorists. Second, it eliminates any need for proof of a defendant’s intent to support terrorism; a defendant’s knowledge that it is providing aid to an organization that engages in terrorism is deemed enough to hold that defendant liable for the organization’s terrorist acts.

*708For the reasons outlined in the Boim I opinion, I continue to believe that Congress when it enacted section 2333(a) subjected to civil liability not only those who engage in terrorism but also those who aid or abet terrorism. 291 F.3d at 1016-21. The government as an amicus curiae has expressed agreement with that view. The secondary liability framework is a much more natural fit for what the defendants here are alleged to have done and as I shall discuss below, the elements of aiding and abetting serve a useful function in distinguishing between those who intend to aid terrorism and those who do not.

But even if I am wrong about the availability of secondary liability under section 2333(a), I have my doubts about the viability of the majority’s theory of primary liability. For there are conceptual problems with this approach, particularly as it is applied in this case. These problems may help to explain why the plaintiffs have long since abandoned any theory of primary liability and have relied solely on theories of secondary liability in this appeal. And it makes it all the more extraordinary that this court has gone out on a limb to craft a liability standard that none of the parties has advocated.

The majority first posits that the defendants’ alleged conduct falls within section 2331(l)’s definition of “international terrorism,” ante at 689-90, but the fit is by no means perfect. In full, the statutory definition of the term reads as follows:

[T]he term “international terrorism” means activities that—

(A) involve violent acts or acts dangerous to human life that are a violation of the criminal laws of the United States or of any State, or that would be a criminal violation if committed within the jurisdiction of the United States or of any State;

(B) appear to be intended—

(i) to intimidate or coerce a civilian population;

(ii) to influence the policy of a government by intimidation or coercion;

(iii) to affect the conduct of a government by mass destruction, assassination, or kidnapping; and

(C)occur primarily outside the territorial jurisdiction of the United States, or transcend national boundaries in terms of the means by which they are accomplished, the persons they appear intended to intimidate or coerce, or the locale in which their perpetrators operate or seek asylumf.]

18 U.S.C. § 2331(1). The language of this definition certainly is broad enough to reach beyond bomb-throwers and shooters to include those who provide direct and intentional support to terrorists: someone who ships arms to a terrorist organization, for example, easily could be thought to be engaging in activity that “involve[s] violent acts or acts dangerous to human life” as set forth in section 2331(1)(A). See Boim I, 291 F.3d at 1014-15. But it is far from clear that sending money to a Hamas-controlled charitable organization, for example, is on par with that type of direct support for terrorism. It may be, as the majority posits, that donations to Hamas’s humanitarian wing indirectly aid its terrorism by freeing up other funds for terrorism, by giving cover to Hamas, and by otherwise enhancing Hamas’s image. But it is difficult if not implausible to characterize donations that are earmarked and used for humanitarian work as violent or life-threatening acts as referenced in section 2331(1)(A). Nor is it evident (to say the least) that financially supporting a Ha-mas-affiliated charity is an act that “appeals] to be intended” to have the sorts of coercive or intimidating effects on govern*709ment policy or upon a civilian population as described in section 2331(1)(B).

It may be more plausible to say, as the majority does, that one who provides financial support to Hamas, even to its charitable subsidiaries, is “providing] material support or resources” to Hamas’s terrorist acts in violation of section 2339A(a) by increasing the heft of Hamas’s purse. See ante at 690-91. But that theory too has its problems. The language of section 2339A(a) requires that the material support or resources be given with the knowledge or intent that they “are to be used in preparation for, or in carrying out” one of a number of specified crimes, including as relevant here the killing of American citizens. (Emphasis mine.); see ante at 690, citing 18 U.S.C. § 2332. In other words, the donor must at least know that the financial or other support he lends to Ha-mas will be used to commit terrorist acts. In Boim I, the panel agreed that giving money to Hamas with the purpose of financing its terrorism would both violate section 2339A(a) and give rise to civil liability under section 2333. 291 F.3d at 1012-16. But at that early stage of this litigation, the Boims had a straightforward and direct theory that Hamas’s American contributors (including HLF) intended for their money be used to support terrorism, that the zakat committees and other humanitarian organizations to which these contributors were sending their money were mere fronts for Hamas, and that the money received by these front organizations was laundered and funneled into Ha-mas’s coffers to fund terrorist activity, including the attack that took David Boim’s life. See id. at 1004. That theory was consistent with the express terms of section 2339A(a). But that is no longer the Boims’ theory (they have long since abandoned it in favor of aiding and abetting and conspiracy), nor is it the majority’s. The majority posits that any money given to a Hamas affiliate, even if it is given with a benign intent and even if it is actually put to charitable use, furthers Hamas’s terrorism in one way or another. Ante at 698. Even if that is so, not all donors will know or intend that their contributions will be used to commit the sorts of criminal acts identified in section 2339A(a). And what the statute proscribes is the knowing or intentional support of specific terrorist acts, not the knowing support of a terrorist organization. If nothing else, the defendants’ contributions to charitable organizations controlled by Hamas would present a factual question as to whether the defendants knew that they were supporting the murder of American citizens or any of the other crimes listed in section 2339A(a).

3.

Causation, as the majority acknowledges, is a staple of tort law, ante at 695, and yet the majority relieves the plaintiffs of any obligation to demonstrate a causal link between whatever support the defendants provided to Hamas and Hamas’s terrorist activities (let alone David Boim’s murder in particular). Instead, the majority simply declares as a matter of law that any money given to an organization like Hamas that engages in both terrorism and legitimate, humanitarian activity, necessarily enables its terrorism, regardless of the purpose for which the money was given or the channel through which the organization received it. “Anyone who knowingly contributes to the nonviolent wing of an organization that he knows to engage in terrorism is knowingly contributing to the organization’s terrorist activities.” Ante at 698. This is judicial activism at its most plain. The majority offers no rationale for relieving the plaintiffs of the burden of showing causation, and there is none that I can discern. The panel in Boim II expressly disavowed any requirement that *710the Boims link specific donations or other acts of support to David Boim’s murder in particular. 511 F.3d at 741. But it did insist on proof that the types of support the defendants were alleged to have given Hamas were, in fact, a cause of Hamas’s terrorism. Id. at 741-43. The panel outlined multiple ways in which the plaintiffs might show that support given to Hamas, even donations to its humanitarian activities, furthers its terrorist agenda, such that it could be considered a cause of David Boim’s murder. Id. Someone familiar with Hamas’s financial structure, or with the financing of terrorism generally, presumably could provide that sort of testimony. But the majority is not even conditioning liability on expert opinion that might link the various types of support provided to Hamas with its terrorist acts. Expert testimony as to the ways in which even aid to Hamas’s humanitarian wing enables terrorism would be subject to adversarial testing and the judgment of the factfinder based on the totality of the evidence put before the court. But rather than subject the notion of causation to those checks, the majority, acting as though we ourselves are experts, simply declares causation to be a given that cannot be challenged. Liability under the majority’s announced rule is sweeping: one who gives money to any Hamas entity, even if it is a small donation to help buy an x-ray machine for a Hamas hospital, is liable from now until the end of time for any terrorist act that Hamas might thereafter commit against an American citizen outside of the United States. (The majority itself acknowledges that under its approach a contribution to a terrorist organization in 1995 might render the donor liable for the murder of an American citizen committed by that organization fifty years later. Ante at 700.) This type of across-the-board judgment is out of place in the realm of torts. As an appellate court, it is our job to articulate a framework of liability under the statute and thereafter leave it to the parties to present evidence pursuant to that framework and to the factfinder to determine whether or not liability has been established. Where it is open to question, as I believe it is, whether even humanitarian support given to Hamas, to its charitable subsidiary, or to a hospital or other institution that receives funding from Hamas, actually contributes to Hamas’s terrorist activities, it should be left to factfinding in individual cases (subject, of course, to appellate review) to evaluate, based on the evidence presented in those cases, what types of support to Hamas and its affiliated entities actually cause terrorism. Cf. Thorogood v. Sears, Roebuck & Co., 547 F.3d 742, 744-45 (7th Cir. 2008) (where the claims of multiple plaintiffs present complex factual questions, it is preferable to let those claims be resolved via individual lawsuits, so that the aggregate outcome fairly reflects the uncertainty of the plaintiffs’ claims, rather than risk error by having the issue resolved on a class-wide basis by a single trier of fact).

The majority’s decision to carve out an exception to its sweeping liability rule for non-governmental organizations like the Red Cross and Doctors Without Borders who provide humanitarian aid to individuals affiliated with Hamas lays bare the weakness of the rule’s analytical underpinnings.6 Providing medical care on the bat*711tlefield to individuals that one knows are Hamas terrorists (see ante at 699) undoubtedly would have the effect of aiding Hamas’s terrorism — patching up an injured terrorist enables him to strike again. I do not doubt that such aid could be given for noble and compassionate reasons, but neither do I doubt that from the standpoint of the Israelis whom Hamas targets, the knowing provision of medical care to individual terrorists could be and would be understood as aid to terrorism. One can also imagine scenarios in which medical aid could be provided for ignoble and devious reasons. Cf. United States v. Alvarez-Machain, 504 U.S. 655, 657, 112 S.Ct. 2188, 2190, 119 L.Ed.2d 441 (1992) (physician indicted for participating in the kidnap and murder of agent of Drug Enforcement Administration by helping to prolong captured agent’s life so that others could continue to interrogate and torture him). Yet, for no apparent reason other than our own sense that organizations like the Red Cross and Doctors Without Borders are good and do good, the majority simply declares them exempt from the broad liability standard that it has announced. Ante at 699. On the other hand, any other individual or organization that gives to a Hamas-controlled charity is deemed liable, regardless of whether the money is given with a humanitarian purpose and regardless of whether the money is, in fact, put to humanitarian use. So one cannot fund the construction of a Hamas hospital, buy the hospital an x-ray machine, or volunteer her medical services to the hospital, because this is not providing direct aid to individuals in the manner of the Red Cross. My colleagues reason that there is a distinction between providing aid to an individual, even if he is terrorist, and aid to a terrorist organization. Ante at 699. But to my mind, that is a distinction without a difference when one knows that the individual being aided is engaged in terrorism (or is recklessly indifferent to that possibility). For example, the majority notes that one way in which Hamas uses its social welfare activities to reinforce its terrorist agenda is by providing economic aid to the families of killed, wounded, or captured Hamas terrorists, which ensures the continued loyalty of these family members to Hamas. Ante at 698. In that respect, one who donates money to Hamas in order to fund such payments thus could be thought to be promoting terrorism. Yet, the same could be said of a donor who instead makes payments directly to the family members of terrorists rather than giving the money to Hamas. Indeed, that is exactly what HLF is alleged to have done (among other things). See Boim II, 511 F.3d at 722; Holy Land Found, for Relief & Dev., 219 F.Supp.2d at 71-73. So providing this type of aid to individuals, rather than to Hamas, would be accomplishing the same end, notwithstanding the fact that the donor was giving aid to individuals rather than to a terrorist organization. See Singh-Kaur v. Ashcroft, 385 F.3d 293, 301 (3d Cir. 2004) (providing food and shelter to militant Sikhs who had committed or planned to commit terrorist acts constituted material support for terrorism). The distinction between aiding an organization and aiding individual members of that organization does not hold up.

It is only the majority’s sweeping rule of liability that puts humanitarian organizations like Doctors Without Borders in peril and that forces the majority to carve out *712an unprincipled exemption for such organizations. If a plaintiff were required to establish a donor’s intent to aid terrorism, along with a causal link between the aid provided and terrorist activity, then the factfinder would be able to draw reasoned, pragmatic distinctions (subject, of course, to appellate review) between those defendants who are truly enabling terrorism and those who are not.

4.

The secondary liability framework that we outlined in Boim I, and on which the plaintiffs built their entire ease against the defendants, provides a more grounded and effective way of identifying and distinguishing between the types of support and supporters that actually aid terrorism and those that do not. As the panel recognized, those who aid and abet Hamas’s terrorism can be held liable to the same extent as those who commit the terrorist acts. Boim I, 291 F.3d at 1016-21. But in addition to showing knowledge of Hamas’s terrorist activity and the provision of financial or other support to Hamas, aiding and abetting would require proof of an intent to help Hamas’s terrorist activities succeed. Id. at 1021, 1023.

Proof of intent would serve two important functions. First, it would serve to single out the most culpable of Hamas’s financiers and other supporters by focusing on those who actually mean to contribute to its terrorist program, as opposed to those who may unwittingly aid Hamas’s terrorism by donating to its charitable arm. I think it would be possible to infer the intent to further terrorism in a number of scenarios. Donations to Hamas itself have been a crime since 1997, for example, when Hamas was formally designated a foreign terrorist organization pursuant to 8 U.S.C. § 1189, see § 2339B(a) and (g)(6); and so a prohibited donation in the wake of that designation would be prima facie proof of one’s intent to further terrorism.7 The same could be said of donations to zakat committees and other organizations that themselves have been formally designated as terrorist organizations based on their links with Hamas. On the other hand, a factfinder confronted with evidence that a donor gave only to a non-designated, Hamas-controlled hospital for the purpose of funding the medical services provided by that hospital would be free to conclude that the donor had a benign intent and did not aid or abet Hamas’s terrorism even if, in the abstract, one might believe that furthering Hamas’s humanitarian activity enhances its image and thereby supports its violent activities. The ability of the factfinder to draw such distinctions is important, given the difficulty there might be in deciding, under the majority’s standard, what constitutes a terrorist organization and what constitutes the knowing provision of support to such an organization. Organizations that openly embrace terrorism as their declared goal are easy to categorize as terrorist organizations. But what about organizations that engage in terrorism but disclaim responsibility? Or organizations whose members frequently engage in terrorist acts with implicit but not explicit approval from the organizations themselves? And what are we to make of charitable entities that are affiliated with such organizations? *713Or charitable entities that receive some but not all of their funding from such organizations — a hospital that receives contributions from Hamas but is not controlled by it, for example? I am not sure just how far the majority’s liability rule extends. Insisting on proof of a donor’s intent to support terrorism would help to confirm the donor’s culpability in instances where the terrorist nature of the organization receiving aid is less clear than it would be if a donor were making out a check payable to Hamas. It would also serve as a principled way to exempt organizations like the Red Cross and Doctors Without Borders, who engage in humanitarian work that may incidentally or tangentially aid individual terrorists or terrorist organizations, but who have no intent to aid terrorist activity.

The intent requirement would also play a vital role in protecting the First Amendment rights of those accused of facilitating Hamas’s terrorism. The possibility that a section 2333(a) suit might implicate First Amendment rights is not an abstract one. Even to the extent that such a suit is based on the money that a defendant has contributed to an organization that engages in terrorism, the defendant’s First Amendment rights must be accounted for, given that donating money to an organization, though it is not speech in and of itself, is one way to express affinity with that organization and to help give voice to the viewpoints that organization espouses. See Buckley v. Valeo, 424 U.S. 1, 65-66, 96 S.Ct. 612, 657, 46 L.Ed.2d 659 (1976) (per curiam) (“The right to join together ‘for the advancement of beliefs and ideas’ is diluted if it does not include the right to pool money through contributions, for funds are often essential if ‘advocacy’ is to be truly or optimally ‘effective.’ ”) (quoting NAACP v. Alabama, 357 U.S. 449, 460, 78 S.Ct. 1163, 1170, 2 L.Ed.2d 1488 (1958)); see also Citizens Against Rent Control v. Berkeley, 454 U.S. 290, 102 S.Ct. 434, 70 L.Ed.2d 492 (1981). Certainly, given the government’s paramount interest in battling terrorism, the government may prospectively ban, and even criminalize, donations to an organization that it deems a terrorist organization. See § 2339B(a); Boim I, 291 F.3d at 1027; Humanitarian Law Project v. Reno, 205 F.3d 1130, 1135 (9th Cir. 2000). Hamas was so designated in 1997, the year after David Boim was murdered. See n. 7, supra. But when an organization engages in both legal and illegal activities and donations to that organization have not been prohibited, a donor may not be held civilly liable for the organization’s illegal activity based solely on his contributions, for to do so would infringe upon the defendant’s First Amendment freedoms. In re Asbestos School Litigation, 46 F.3d 1284, 1290 (3d Cir. 1994) (Alito, J.).

And money is not the only type of support that the defendants are alleged to have provided Hamas. One need only look again at the conduct for which AMS was held liable by the district court: hosting Hamas speakers at its conferences, publishing pro-Hamas articles and editorials in his newsletters, rallying support for HLF when it was declared a terrorist organization, and so forth. 340 F.Supp.2d at 908-13. All of that conduct involves pure speech. See ante at 700; Boim I, 291 F.3d at 1026.

And so the First Amendment is very much implicated by this case. Both through their contributions of money to Hamas and its subsidiary organizations, and (in the case of AMS) through their advocacy on behalf of Hamas, the defendants have demonstrated an affiliation with and affinity for Hamas. But NAACP v. Claiborne Hardware Co., 458 U.S. 886, 102 S.Ct. 3409, 73 L.Ed.2d 1215 (1982), holds that an individual may not be held *714civilly liable for his mere association with an organization whose members engage in illegal acts.

Civil liability may not be imposed merely because an individual belonged to a group, some members of which committed acts of violence. For liability to be imposed by reason of association alone, it is necessary to establish that the group itself possessed unlawful goals and that the individual held a specific intent to further those illegal aims.

Id. at 920, 102 S.Ct. at 3429 (footnote omitted). Moreover, an individual’s intent vis-a-vis an organization that holds both lawful and unlawful purposes “must be judged ‘according to the strictest law,’ ”

for “otherwise there is a danger that one in sympathy with the legitimate aims of such an organization, but not specifically intending to accomplish them by resort to violence, might be punished for his adherence to lawful and constitutionally protected purposes, because of other and unprotected purposes which he does not necessarily share.”

Id. at 919, 102 S.Ct. at 3429 (quoting Noto v. United States, 367 U.S. 290, 299-300, 81 S.Ct. 1517, 1521, 6 L.Ed.2d 836 (1961)). The panel in Boim I recognized that the aiding and abetting standard is consistent with the rule announced in Claiborne Hardware in that it conditions liability on proof that a defendant knew of the organization’s illegal purposes and had the intent to further those purposes when that defendant joined and/or aided the organization. 291 F.3d at 1023-24. By contrast, the majority’s approach requires no proof of an intent to further Hamas’s activities; so long as a donor to Hamas or its affiliate knows that Hamas engages in terrorism, the donor is liable for any terrorist act committed by Hamas against an American citizen regardless of the purpose behind the donation.

The majority suggests that the rule of Claiborne Hardware does not apply because violence is a stated goal of Hamas rather than something a few rogue members happen to engage in without its approval.

The defendants in the present case could not be held liable for acts of violence by members of Hamas that were not authorized by Hamas.... But as Hamas engages in violence as a declared goal of the organization, anyone who provides material support to it, knowing the organization’s character, is punishable ... whether or not he approves of violence.

Ante at 700 (emphasis in original). But this holding is directly contrary to Claiborne Hardware, which requires proof of a defendant’s intent to further violence even when violence is a goal that the organization embraces. See 458 U.S. at 920, 102 S.Ct. at 3429 (“For liability to be imposed by reason of association alone, it is necessary to establish that the group itself possessed unlawful goals and that the individual held a specific intent to further those illegal aims.”) (emphasis added). See Scales v. United States, 367 U.S. 203, 229, 81 S.Ct. 1469, 1486, 6 L.Ed.2d 782 (1961) (individual may be convicted for active membership in organization that advocates violent overthrow of U.S. government so long as there is “clear proof that a defendant ‘specifically intend[s] to accomplish [the aims of the organization] by resort to violence.’ ”) (quoting Noto v. United States, 367 U.S. at 299, 81 S.Ct. at 1522); see also Communist Party of Indiana v. Whitcomb, 414 U.S. 441, 447-49, 94 S.Ct. 656, 661-62, 38 L.Ed.2d 635 (1974) (government may not forbid advocacy of lawbreaking or use of force unless it is inciting imminent lawless action) (citing Brandenburg v. Ohio, 395 U.S. 444, 447-48, 89 S.Ct. 1827, 1829-30, 23 L.Ed.2d 430 (1969)); Elfbrandt v. Russell, 384 U.S. 11, 15-18, 86 *715S.Ct. 1238, 1240-41, 16 L.Ed.2d 321 (1966). Certainly I agree that someone who gives money or other support to Hamas knowing that it will be used for terrorist activity — a violation of section 2339A(a) — can be held civilly liable for that activity, but in that case one’s intent could readily be inferred. But to impose liability based on aid that may have been given — and, in fact, used-for humanitarian purposes is to do exactly what Claiborne Hardware proscribes: punish the supporter “for his adherence to [an organization’s] lawful and constitutionally protected purposes, because of other and unprotected purposes which he does not necessarily share.” 458 U.S. at 919, 102 S.Ct. at 3429 (quoting Noto).

Given that the majority’s analysis requires no proof of that any of the defendants intended to support Hamas’s terrorism, it is inconsistent with the Supreme Court’s First Amendment jurisprudence. Although the majority suggests that an intent requirement would, as a practical matter, eliminate donor liability except in those few cases where a donor declared his intent to support terrorism, ante at 698, that certainly is not true in other areas of the law where proof of a defendant’s intent is required. As we often note in employment discrimination and a wide variety of other cases, there is rarely direct proof of a defendant’s intent, and yet intent can be proved circumstantially. See, e.g., Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 141, 120 S.Ct. 2097, 2105, 147 L.Ed.2d 105 (2000) (age discrimination); Amrhein v. Health Care Serv. Corp., 546 F.3d 854, 858 (7th Cir. 2008) (Title VII retaliation); United States v. Roberts, 534 F.3d 560, 571 (7th Cir. 2008) (wire fraud); United States v. Patterson, 348 F.3d 218, 225-26 (7th Cir. 2003) (narcotics conspiracy), abrogated on other grounds by Blakely v. Washington, 542 U.S. 296, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004); Toushin v. Comm’r of Internal Revenue, 223 F.3d 642, 647 (7th Cir. 2000) (tax fraud); United States v. Rose, 12 F.3d 1414, 1417, 1420 (7th Cir. 1994) (aiding and abetting the transportation and receipt of a stolen motor vehicle). Moreover, should there be evidence that a defendant has made statements in support of the use of violence to achieve political ends, relying on such statements as proof that the defendant provided financial or other aid to a terrorist organization with the intent to support its terrorist activities would not, as the majority suggests, ante at 698-99, pose a First Amendment problem. See Wisconsin v. Mitchell, 508 U.S. 476, 489, 113 S.Ct. 2194, 2201, 124 L.Ed.2d 436 (1993) (“The First Amendment ... does not prohibit the evidentiary use of speech to establish the elements of a crime or to prove motive or intent.”).

5.

Finally, the majority treats Dr. Paz’s affidavit as sufficient evidence that Hamas was responsible for David Boim’s murder. Although the majority recognizes that Paz relied on a variety of unauthenticated electronic and documentary sources for his conclusion, it nonetheless deems his affidavit admissible and sufficient to sustain summary judgment for the plaintiffs on this point because an expert is free in forming his opinion to rely on evidence that would not be admissible in court. Ante at 704. But the panel’s principal point was that Dr. Paz’s conclusion as to who killed David Boim is meaningless without reference to the websites and documents that he so heavily relied upon in forming his opinion, and yet allowing Paz to recount what those sources say without establishing their authenticity and trustworthiness would contradict the basic requirement that expert opinion have “a reliable foundation,” Daubert v. Merrell Dow Pharmaceuticals, supra, 509 U.S. at 597, *716113 S.Ct. at 2799; see also Fed.R.Evid. 703. Paz’s opinion is based exclusively on what these websites and documents say; he has no personal knowledge of who killed David Boim. So if these sources are not genuine or say something other than what he has represented, then his opinion is worthless. No expert worth his salt would base his opinion on internet and documentary sources without assuring himself that they are reliable — that a website thought to be a Hamas site is, in fact, a website controlled by Hamas and authorized to make representations on its behalf, for example, or that what purports to be the written judgment of a foreign tribunal is actually that. But Paz’s affidavit does not describe any such efforts that he made, and there is no other evidence in the record that establishes the authenticity and reliability of the websites and documents whose contents he recounts.

The glaring lack of any information confirming the authenticity and accuracy of Paz’s sources raises obvious doubts about the reliability of his opinion. To cite just a few examples: For the proposition that Hinawi killed David Boim, Paz relies on a document in Arabic that purports to be the written judgment reflecting Hinawi’s conviction and sentence before a Palestinian Authority tribunal, along with the notes of a U.S. State Department employee who observed Hinawi’s trial. Here is the cover letter accompanying and describing both the trial notes and the judgment (Figure 1), followed by the judgment form (Figure 2):

Figure 1: Cover letter

[[Image here]]

*717Figure 2: Hinawi Judgment

[[Image here]]

No translation of the Arabic-language judgment has been provided (it could be an advertisement for all I know), and neither the judgment nor the notes of the foreign service officer have been authenticated in any meaningful way by the cover letter, which does not even identify the letter’s author. We have absolutely no way to know, given the current state of the record, whether these documents are what Paz says they are, and thus no way of assessing the reliability of his conclusions. As a final example, here is one of the web pages Paz relied on as evidence that Ha-mas took responsibility for David Boim’s murder:

*718Figure 3: web page

[[Image here]]

The selective translation obviously makes it impossible for the reader to independently evaluate the context and meaning of what Paz is relying on. Notwithstanding these infirmities, the majority is content not only to deem Paz’s opinion admissible, but to sustain the entry of summary judgment against the defendants on this point. The defendants cannot be faulted for failing to refute Paz’s conclusions, see ante at 704-05, for the party opposing summary judgment is not required to rebut factual propositions on which the movant bears the burden of proof and that the movant has not properly supported in the first instance. See Celotex Corp. v. Catrett, 477 U.S. 317, 331, 106 S.Ct. 2548, 2557, 91 L.Ed.2d 265 (1986) (Brennan, J., dissenting) (citing 10A C. Wright, A. Miller, & M. Kane, Federal Practice & Procedure § 2727 (2d ed. 1983)); L & W, Inc. v. Shertech, Inc., 471 F.3d 1311, 1318 (Fed.Cir. 2006); Black v. M & W Gear Co., 269 F.3d 1220, 1238 (10th Cir. 2001). In any other sort of case, this sort of sloppiness would not be tolerated, and we certainly would not sustain the entry of summary judgment based on such shaky evidence.8

6.

The murder of David Boim was an unspeakably brutal and senseless act, and I can only imagine the pain it has caused his parents. Terrorism is a scourge, but it is our responsibility to ask whether it presents so unique a threat as to justify the abandonment of such time-honored tort requirements as causation. Our own re*719sponse to a threat can sometimes pose as much of a threat to our civil liberties and the rule of law as the threat itself. See, e.g., Korematsu v. United States, 323 U.S. 214, 65 S.Ct. 193, 89 L.Ed. 194 (1944). The panel’s opinion in Boim II took a conservative approach, fully consistent with precedent, that insisted on proof that the defendant’s actions were a cause of Hamas’s terrorism, proof that the defendants intended to support terrorism, and admissible evidence to support such basic factual points as whether Hamas was responsible for David Boim’s murder. This en banc court, by contrast, relieves the plaintiffs of all of these obligations, following a path that portends sweeping liability for those individuals and groups who give their support to the humanitarian activities and affiliates of terrorist organizations but who may have no intent to support terrorism and whose actual link to terrorism has never been evaluated by a factfinder. I stand by the approach taken by the Boim I and Boim II panel.

7.

For all of the reasons set forth above and in the panel’s Boim II opinion, I would remand for further proceedings as to all four defendants, including Salah. I would require the plaintiffs on remand to demonstrate that any financial or other support the defendants have given to Hamas and Hamas-affiliated entities was in some way a cause of Hamas’s terrorism. I would also insist the plaintiffs set forth a more complete evidentiary foundation for the proposition that Hamas killed David Boim.

concurring in part and dissenting in part.1

This is a heart-breaking case. No parent can fail to empathize with Joyce and Stanley Boim, who lost their son to the evil of terrorism just as he was on the brink of all of life’s promise. Nothing can bring David Boim back, but the Boims have taken advantage of a statute that Congress passed that was designed to provide some degree of accountability for those who commit such awful acts. See 18 U.S.C. § 2333(a). In Boim v. Quranic Literacy Inst. & Holy Land Found., 291 F.3d 1000 (7th Cir. 2002) (“Boim I”), this court decided that the set of possible defendants in such an action includes not only the direct actors (here, Amjad Hinawi and Khalil Tawfiq Al-Sharif) and the organization to which they belonged and that directed their actions (here, said to be Hamas), but also organizations that aid and abet the former two. When all is said and done, the en banc majority has reaffirmed the latter ruling, though it does so under a slightly different rubric. But, in our zeal to bring justice to bereaved parents, we must not lose sight of the need to prove liability on the facts that are presented to the court. Assumptions and generalizations are no substitute for proof. Particularly because, unfortunately, this probably will not be the last case brought by a victim of international terrorism, it is crucial that we be as clear as we can in fleshing out the statutory requirements and that we do not rush to judgment. Because I do not agree with the majority’s articulation and application of some of the governing legal standards, and I find too many central facts to be in dispute, I am still of the view that this case needs to be remanded for further proceedings.

I begin, however, by underscoring that I agree with the en banc majority’s analysis on a number of points. First, throughout *720the proceedings before this court, we have unanimously rejected the district court’s decision to give collateral estoppel effect to the findings in the case that was litigated in the District of Columbia, Holy Land Found. for Relief & Dev. v. Ashcroft, 219 F.Supp.2d 57 (D.D.C. 2002), affirmed, 333 F.3d 156 (D.C.Cir. 2003) (“HLF v. Ashcroft”). See Boim v. Holy Land Foundation for Relief and Dev., 511 F.3d 707, 720-33 (7th Cir. 2007) (“Boim II”). We all agree that it was error to grant summary judgment in favor of the plaintiffs against the Holy Land Foundation for Relief and Development (“HLF”), and that further proceedings are required. Second, under the new analysis that the en banc majority has undertaken, which uses “a chain of explicit statutory incorporations by reference,” ante, at 690, it was error to grant summary judgment in favor of the plaintiffs against Muhammad Salah. Again, we all agree that there are problems with Salah’s part of the case. The en banc majority is reversing the finding of liability outright because Salah could not have rendered material support to Hamas between the effective date of 18 U.S.C. § 2339A, September 13, 1994, and the date of David Boim’s murder, May 13, 1996, because he was in Israeli custody between January 1993 and November 1997. Ante, at 691. In fact, the Boim II panel majority took a less absolute approach. It found that the district court erred in concluding that Sa-lah’s liability could be established only by showing that (1) he knew of Hamas’s terrorist activities, (2) he desired to help those activities succeed, and (3) through his participation in the Hamas conspiracy, acts of co-conspirators sufficed to show that he engaged in some act of helping to bring about Boim’s murder. Rather than reversing outright, as the en banc majority has done, the Boim II panel majority would have reversed the summary judgment in the plaintiffs’ favor and remanded to give plaintiffs the opportunity to identify “evidence that would permit a reasonable factfinder to find that Salah’s actions on behalf of Hamas in some way caused or contributed to David Boim’s death.” Boim II, 511 F.3d at 748.

I am persuaded by the en banc majority’s statutory analysis that the correct result is reversal of the finding against Sa-lah, rather than a remand for further proceedings. Its careful exegesis of the way that the governing statutes in this area work together demonstrates why the furnishing of material assistance is a ground for liability under 18 U.S.C. § 2333. I thus do not dissent from the en banc court’s decision that the judgment against Salah must be reversed.

It is the en banc majority’s analysis of the cases against the Quranic Literacy Institute (“QLI”) and the American Muslim Society (“AMS”) (along with the Islamic Association of Palestine) that I find problematic. I continue to believe that the decisions in Boim I and Boim II correctly found that Congress intended, in passing 18 U.S.C. § 2333, to create an intentional tort, that it meant to “extend civil liability for acts of international terrorism to the full reaches of traditional tort law,” Boim I, 291 F.3d at 1010, that nothing in Central Bank of Denver, N.A. v. First Interstate Bank of Denver, N.A., 511 U.S. 164, 114 S.Ct. 1439, 128 L.Ed.2d 119 (1994), suggests that Congress lacks the power to do so when it wishes, and finally that § 2333 does impose secondary liability on those who aid and abet acts of terrorism. The en banc majority expresses doubts about this holding, although in the end it neither adopts it nor rejects it. Instead, it turns to “an alternative and more promising ground for bringing donors to terrorist organizations within the grasp of section 2333.” Ante, at 690.

*721Working through a chain of statutes— from § 2333(a) (treble damages action for person injured by an act of international terrorism), to § 2331(1) (definition of international terrorism), to § 2339A (providing material support for something that violates a federal criminal law is itself a crime), to § 2332 (criminalizing the killing of any American citizen outside the United States) — the en banc majority concludes that there is primary liability under § 2333(a) for someone who donates money “to a terrorist group that targets Americans outside the United States.” Ante, at 690. The en banc majority then establishes several criteria for the claim it has recognized: (1) it is the fact of contributing to a terrorist organization, not the amount of the contribution, that is the key to liability, ante, at 691; and (2) there is a knowledge requirement, to the effect that the donor-defendant must have known that the money would be used “in preparation for or in carrying out the killing or attempted killing of, conspiring to kill, or inflicting bodily injury on, an American citizen abroad.” Ante, at 691. At that point, however, the en banc majority announces that its theory does not establish primary liability after all — instead, a claim based on material support “has the character of secondary liability. Through a chain of incorpo-rations by reference, Congress has expressly imposed liability on a class of aiders and abettors.” Id.

I would have thought that this was exactly the conclusion that the Boim I panel reached. By labeling its theory as one of primary liability, the en banc majority is apparently trying to reap the advantages of both kinds of theories. It acknowledges that in order to prove a primary liability case, the plaintiffs would need to establish “the ordinary tort requirements relating to fault, state of mind, causation, and foreseeability.” Ante, at 692. But, it says, those requirements do not apply here, because “functionally the primary violator is an aider and abettor or other secondary actor.” Ante, at 692.

I believe that the following is a fair summary of the formal requirements that the en banc majority has announced for proving a case under § 2333:

1. Act requirement: the defendant must have provided material assistance, in the form of money or other acts, directly or indirectly, to an organization that commits terrorist acts.

2. State of mind requirement: the defendant must either know that the donee organization (or the ultimate recipient of the assistance) engages in such acts, or the defendant must be deliberately indifferent to whether or not it does so.

3. Causation: there is no requirement of showing classic “but-for” causation, nor, apparently, is there even a requirement of showing that the defendant’s action would have been sufficient to support the primary actor’s unlawful activities or any limitation on remoteness of liability.

There is little to criticize in the first of these criteria, as an abstract matter. The second may also pass muster, again as an abstract matter. For both of these, my problem with the en banc majority’s opinion lies more in the way that they are applied to these facts, as I explain further below, than in their formal scope. With respect to the third requirement, there is both a theoretical problem and a problem with the application, and so I begin with that.

The en banc majority asserts that its position on causation is supported by a number of cases that it discusses. Those cases, however, do not go as far as the en banc majority claims, nor am I familiar *722with anything else in the law of torts that does so. It is important here to be precise once again about areas of agreement and areas of disagreement. The en banc majority is quite right to point out that literal “but-for” causation cannot be shown in certain cases, and in those cases, the courts have accepted substitutes for the “but-for” showing. Thus, in the case where there are two independent acts, and either one alone would have brought about the injury, a defendant who was responsible for one of those acts cannot defeat liability by pointing out that the other one would have been enough to create the harm by itself. That is the principle illustrated by Kingston v. Chicago & N.W. Ry. Co., 191 Wis. 610, 211 N.W. 913 (1927), discussed in our decision in Maxwell v. KPMG LLP, 520 F.3d 713, 716 (7th Cir. 2008). It is also the principle endorsed by the most recent draft of the American Law Institute’s Restatement (Third) of the Law of Torts: Liability for Physical and Emotional Harm, § 27 (“Restatement (Third) of Torts”), which says “[i]f multiple acts occur, each of which alone would have been a factual cause under § 26 of the physical harm at the same time in the absence of the other act(s), each act is regarded as a factual cause of the harm.” This is a far cry from saying that cause need not be proven if there are multiple sufficient causes; the ALI’s draft acknowledges simply that some harms may be overdetermined, and in those cases, cause can be proven by demonstrating that the defendant’s tortious conduct was sufficient to produce the harm. Maxwell, cited by the en banc majority, illustrates this principle as well as anything: “There are also cases in which a condition that is not necessary, but is sufficient, is deemed the cause of an injury, as when two fires join and destroy the plaintiffs property and each one would have destroyed it by itself and so was not a necessary condition; yet each of the fire-makers (if negligent) is liable to the plaintiff for having ‘caused’ the injury.” 520 F.3d at 716. The key word here is “sufficient”: the plaintiff cannot win without showing that the defendant’s act would have been sufficient to cause the injury, even though it may be the case that other acts might also have been sufficient.

The other examples the en banc majority uses fit the rule articulated in Restatement (Third) § 27. Thus, if there were two wrongful causes and a third innocent one (two arsonists plus a lightning strike, for example), any of which would have caused the injury at issue, the person responsible for one of the wrongful acts cannot take refuge in the fact that other sufficient causes were also present. Or if, as in the classic case of Summers v. Tice, 33 Cal.2d 80, 199 P.2d 1 (1948), there are two possible causes, either of which would have been sufficient to cause the harm (a bullet from each of two guns, either one of which would have sufficed to harm the third party), once again sufficient cause has been proven even if necessary cause cannot be. Ditto with the en banc majority’s example of several firms that spill toxic waste that finds its way into groundwater and damages property. Even if the damage is slight, that wrongful act is sufficient for liability. Any remaining uncertainties can be resolved through rules on apportionment of damages.

In the end, the en banc majority is reduced to relying on a case where a roomful of junior high school students erupted into a melee and a bystander student was seriously hurt. See Keel v. Hainline, 331 P.2d 397 (Okla. 1958). A closer look at the facts of that case is useful. Approximately 35 to 40 students were in their music classroom one day, but because the instructor failed to show up on time, they were unsupervised for about a half hour. *723Here is the court’s description of what unfolded:

During the absence of the instructor, several of the male students indulged in what they termed “horse play”. This activity consisted of throwing wooden blackboard erasers, chalk, cardboard drum covers, and, in one instance, a “coke” bottle, at each other. It appears that two or three of the defendants went to the north end of the class room and the remaining defendants went to the south end of the room. From vantage points behind the blackboard on the north end and the piano on the south end, they threw the erasers and chalk back and forth at one another. This activity was carried on for a period of some 30 minutes, and terminated only when an eraser, thrown by defendant [Larry] Jennings, struck plaintiff in the eye, shattering her eye glasses, and resulting in the loss of the use of such eye.

331 P.2d at 398-99. The defendants to whom the court refers were six boys — two or three at one end of the room, the rest at the other end of the room. Robert Keel, the plaintiff-in-error, was in one of those groups — the facts do not mention whether Keel was on Jennings’s “team” or the other one. The court first found that what it characterized as “the willful and deliberate throwing of wooden blackboard erasers at other persons in a class room containing 35 to 40 students” was wrongful conduct, because it amounted to an assault and battery. Id. at 399. The intent of the actors was immaterial. Addressing Keel’s argument that there was no evidence that he aided or abetted Jennings in the final throw that injured the plaintiff, the court said:

It is undisputed that defendant Keel participated in the wrongful activity engaged in by the other defendants of throwing wooden blackboard erasers at each other back and forth across a class room containing 35 to 40 students, although most of the testimony indicates that defendant Keel’s participation was limited to the retrieving of such erasers and handing them to other defendants for further throwing. Keel aided and abetted the wrongful throwing by procuring and supplying to the throwers the articles to be thrown. It is immaterial whether defendant Keel aided, abetted or encouraged defendant Jennings in throwing the eraser in such a manner as to injure Burge, or not, since it is virtually undisputed that defendant Keel aided, abetted or encouraged the wrongful activity of throwing wooden erasers at other persons, which resulted in the injury to Burge.

331 P.2d at 400 (emphasis added). The en banc majority reads this as a holding that Keel was liable “even though there was no proven, or even likely, causal connection between anything he did and the injury.” Ante, at 697. But that reading entirely ignores the perspective that the Oklahoma court adopted. Keel and the other five boys jointly created a dangerous situation in the classroom. By acting together, they greatly enhanced the risk of harm to the other students in the room. So viewed, there is a readily observable causal link between the collective action of the six boys and the harm the plaintiff suffered. Whether we call Keel’s contribution “material support” or something else does not matter — the point here is that the Oklahoma court did not dispense with the requirement of proving causation.

So, too, must we insist on proof that QLI’s and AMS’s actions amounted to at least a sufficient cause of the terrorist act that killed David Boim, even if, on these facts, there were multiple such causes. The Boim II panel majority opinion outlines ways in which this might be done. I would summarize these approaches to the *724causation element as follows: there must be proof (1) that the actual recipient organization received a non-trivial amount of money from either QLI or AMS, and (2) that the recipient was, itself, sufficiently affiliated with Hamas that those dollars indirectly supported Hamas’s terrorist mission. Because money is fungible, the combination of the link to Hamas and the receipt of an amount that would have been sufficient to finance the shooting at the Beit El bus stop would be enough to show that the “material assistance” of giving money caused the terrorist act that took David Boim’s life. (There is no allegation here that either QLI or AMS directly tunneled money to Hamas; had there been, this obviously would have sufficed as well.)

Another reason why I find it ill-advised to exempt plaintiffs suing under § 2333 on a “material assistance” theory from showing causation is that this approach also appears to eliminate the need to show what was classically called “proximate cause.” As the Proposed Final Draft to the Restatement (Third) of Torts points out, that term is imprecise at best. See Restatement (Third) of Torts, ch. 6, Special Note on Proximate Cause. The new Restatement refers to this concept as “scope of liability,” in recognition of the fact that “[t]ort law does not impose liability on an actor for all harm factually caused by the actor’s tortious conduct.” Id. At some point, the harm is simply too remote from the original tortious act to justify holding the actor responsible for it. It may be the case that the boundaries of liability are wider for intentional torts, see Restatement (Third) of Torts § 33, but that does not mean that they are limitless. In part, this reflects the reality that as the temporal or factual chain between the tor-tious act and the harm becomes ever longer, the likelihood of intervening or superseding causes becomes greater. See generally Restatement (Third) of Torts § 34. The en banc majority freely concedes that there are no limits at all to its rule, and that a donor who gave funds to an organization affiliated with Hamas in 1995 might still be liable under § 2333 half a century later, in 2045. I see no warrant for assuming that § 2333, unlike the rest of tort law, contains no scope-of-liability limitations. I note as well that such an open-ended rule would be in serious tension with the general four-year statute of limitations Congress has passed for civil actions based on statutes passed after 1990 (like this one). See 28 U.S.C. § 1658.

The scope of the causation element is not my only concern about the en banc majority’s opinion. My other problem is with its application of the principles that, at a high level of generality, state the law correctly. As I noted earlier, the plaintiffs must prove that the defendant provided material assistance to an organization that commits terrorist acts. But what does it take to qualify as such an organization? The Boims did not sue Hamas, nor does their case rely on the proposition that QLI or AMS sent money directly to Hamas. We must decide how far down the chain of affiliates, in this shadowy world, the statute was designed to reach, and how deeply Hamas must be embedded in the recipient organization. QLI and AMS argue strenuously that at worst they sent money to charitable organizations with some kind of link to Hamas. Some might have been analogous to wholly owned subsidiaries; some might have been analogous to joint ventures; some might have been independent entities that accepted funding from Hamas as well as other more reputable organizations. The record throws little light on these matters, because the district court thought them irrelevant. As I understand the en banc majority opinion, it is saying that even if an independent day care center receives $1 from organization *725H known to be affiliated with Hamas, not only the day care center but also anyone who gave to H is liable for all acts of terrorism by Hamas operatives from that time forward against any and all Americans who are outside the United States.

That is a proposition of frightening, and I believe unwise, breadth. The en banc majority has tried to carve out humanitarian non-governmental organizations like the American Red Cross and Doctors Without Borders, which (fortuitously) may also benefit from a “medical services” exemption in the statute. But I am not sure that it has succeeded. Those worthy organizations are not the only ones committed to nondiscriminatory treatment of all needy human beings. The United Nations High Commissioner for Refugees sponsors many programs designed to assist people in war-torn areas. The United Nations Relief and Works Agency for Palestine Refugees in the Near East (“UNRWA”) has been in existence since 1950. See http://www.unhcr.org/partners/ PARTNERS/48fdeced20.html (last visited November 11, 2008). It describes itself as “the main provider of basic services — education, health, relief and social services — to over 4.1 million registered Palestine refugees in the Middle East.” Id. The odds are strong that some of the agencies that UN-RWA helps may also receive assistance from Hamas. The en banc majority does not tell us whether, if QLI or AMS also happens to give money to such an agency, the donor has violated § 2333 by doing so.

The en banc majority also slides over the statutory requirement (derived from its chain of statutory connections) that the entity providing material assistance must know that the donee plans to commit terrorist acts against U.S. citizens. Ante, at 691-92, 693. All that is necessary, we are told, is that

a donor [to Hamas — and presumably to another organization with an adequate link to Hamas, whatever that may be] who knew the aims and activities of the organization [only Hamas? or the affiliated recipient?] — would know that Ha-mas was gunning for Israelis, that Americans are frequent visitors to and sojourners in Israel, that some Israeli citizens have U.S. citizenship as well, and that donations to Hamas, by augmenting Hamas’s resources, would enable Hamas to kill or wound, or try to kill, or conspire to kill more people in Israel.

Ante, at 693-94. This is awfully vague. Americans travel, and are known to travel, to every country on the face of the globe— they even go to places like Antarctica that are not even countries. If one could, it would be more realistic and sound as a legal matter simply to hold that it makes no difference whether or not the terrorist acts that the organization commits are directed toward Americans. The only problem with such a holding — which otherwise would be a routine application of the doctrine of transferred intent — is that the statutory basis for a tort action under § 2333 depends upon a finding that the material support violated U.S. federal criminal law, and that here the crime in question is the killing of an American citizen outside the United States. In my view, given the language of the statutes that Congress has passed thus far, we are required to take a more restricted view of § 2333. A statute focusing on extraterritorial killings of Americans would still be a strong tool against terrorist activities and organizations that threaten vital U.S. interests. A1 Qaeda, for example, trumpets its intent to target Americans whenever and wherever it can. If the plaintiffs could show both that Hamas has done the same thing and further that Hamas’s intent should be attributed to the donee organiza*726tion (recalling once again that neither QLI nor AMS gives money directly to Hamas), then a § 2333 claim may proceed; otherwise, it may not. Put differently, I find it difficult to read § 2333 as creating a claim against an organization that has, in effect, declared war on the entirety of civilization.

The Boim II opinion explains the problems with a finding, on the present record, that Hamas was indeed responsible for David Boim’s murder. That finding rests entirely on the affidavit submitted by Dr. Reuven Paz. The majority accepts that affidavit as adequate, noting only the uncontroversial point that experts are allowed to rely on hearsay and other inadmissible evidence. See Fed.R.Evid. 703 (expert may rely on facts or data “reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject”). No one doubts this. The panel’s point in Boim II was that, at least since Daubert v. Morrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and the revision of FED. R. EVID. 702, there must nevertheless be a solid foundation for the expert’s opinion. Rule 702 puts the point this way: the expert may offer an opinion “if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.” It is these threshold criteria that are at issue. No one is saying that these requirements cannot be met in this case, or in any other case involving international teiTorism. They just have not been satisfied yet, and so QLI and AMS should have won a remand on this basis as well.

For these reasons, I would remand for further proceedings on the claims against QLI and AMS. I concur in the en banc majority’s opinion insofar as it reverses the judgment against Salah and it remands for further proceedings on the claims against HLF.