Foreseeability Conventions

The risk reasonably to be perceived defines the duty to be obeyed . . . .
                                                            —Palsgraf v. Long Island Railroad Co. (N.Y. 1928)

[T]here are clear judicial days on which a court can foresee forever . . . .
                                                                                          —Thing v. La Chusa (Cal. 1989)

How has the foreseeability standard survived its critics? Law relies on foreseeability to solve hard legal problems in a vast array of doctrinal fields. But for a century and more, critics have pilloried the standard as hopelessly indeterminate. Decisionmakers, observe the critics, can characterize virtually any consequence as either foreseeable or unforeseeable. It all depends on how one tells the story. This Article explains the conundrum of foreseeability’s puzzling persistence by offering a novel account of how foreseeability has flourished in fields like tort, contract, and crime. Foreseeability has survived and flourished, the Article proposes, not because it carries determinate meaning (it does not), but because lawyers, judges, and juries have established fixes or hacks—which in this Article we call foreseeability conventions—to settle what would otherwise be intractable foreseeability problems. Foreseeability conventions work because they give the concept meaning in particular fields and in discrete situations, furthering the law’s basic goals in especially thorny categories of recurring cases. We describe two types of conventions: storytelling or narrative conventions, on the one hand, and per se conventions, on the other. We offer salient illustrations, relying especially on the law of torts, showing how the law substitutes rough-hewn proxies for impenetrable foreseeability questions. In closing, we propose that the conventions strategy for resolving indeterminacy is widespread and even pervasive in the law. We observe, too, that the conventions strategy is being put to use today in solving controversial, high-profile legal problems in our age of political and cultural division—even as social fracture risks undermining the tacit agreements on which doctrinal conventions rest.

Introduction

What legal formulation is simultaneously the most indispensable to the law’s everyday operation and the most famously incoherent? The concept of foreseeability bids for the prize.

For at least a century and a half, leading commentators have relied on foreseeability to solve hard legal problems in fields like torts,1See, e.g., Palsgraf v. Long Island R.R., 162 N.E. 99 (N.Y. 1928). contracts,2See, e.g., Hadley v. Baxendale (1854) 156 Eng. Rep. 145. criminal law,3See, e.g., Model Penal Code §§ 2.03, 2.11 (Am. L. Inst., Proposed Official Draft 1962). police violence,4See, e.g., County of Los Angeles v. Mendez, 137 S. Ct. 1539 (2017); Michael Kimberly, Discovering Arrest Warrants: Intervening Police Conduct and Foreseeability, 118 Yale L.J. 177 (2008). bankruptcy,5See, e.g., 11 U.S.C. § 523(a)(6); In re Pattison, 132 B.R. 449, 450 (Bankr. D.N.M. 1991). constitutional law,6See, e.g., Bond v. United States, 529 U.S. 334 (2000) (discussing how reasonable expectations define the boundaries of Fourth Amendment rights against unreasonable searches and seizures); see also id. at 339–40 (Breyer, J., dissenting) (contending that the search in question was “entirely ‘foreseeable’” (quoting United States v. Bond, 167 F.3d 225, 227 (5th Cir. 1999))). corporations,7See, e.g., Moody v. Sec. Pac. Bus. Credit, Inc., 971 F.2d 1056, 1073 (3d Cir. 1992) (using a “reasonable foreseeability” test to assess undercapitalization in buyout); Swinney v. Keebler Co., 480 F.2d 573, 578 (4th Cir. 1973) (noting that if sellers can foresee potential fraud, they have a fiduciary duty to investigate the purchaser of the corporation). defamation,8See, e.g., Rath v. Retail Credit Co., 286 N.Y.S.2d 286, 287 (App. Div. 1968) (holding that danger of prejudice in defamation action was foreseeable). environmental law,9See, e.g., Dep’t of Transp. v. Pub. Citizen, 541 U.S. 752 (2004). antitrust,10See, e.g., FTC v. Phoebe Putney Health Sys., Inc., 568 U.S. 216 (2013); Elizabeth Trujillo, State Action Antitrust Exemption Collides with Deregulation: Rehabilitating the Foreseeability Doctrine, 11 Fordham J. Corp. & Fin. L. 349 (2006). immigration,11See, e.g., Clark v. Martinez, 543 U.S. 371 (2005); Demore v. Kim, 538 U.S. 510 (2003); Zadvydas v. Davis, 533 U.S. 678 (2001). takings,12See, e.g., Ark. Game & Fish Comm’n v. United States, 568 U.S. 23 (2012). personal jurisdiction,13See, e.g., Burger King Corp. v. Rudzewicz, 471 U.S. 462 (1985); Burnham v. Superior Court, 495 U.S. 604 (1990). patents,14See, e.g., Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co., 535 U.S. 722, 738 (2002) (“There is no reason why a narrowing amendment should be deemed to relinquish equivalents unforeseeable at the time of the amendment and beyond a fair interpretation of what was surrendered.”); Matthew J. Conigliaro, Andrew C. Greenberg & Mark A. Lemley, Foreseeability in Patent Law, 16 Berkeley Tech. L.J. 1045 (2001). and more.15See, e.g., U.S. Sent’g Guidelines Manual § 1B1.3; United States. v. Maddox, 803 F.3d 1215, 1221 (11th Cir. 2015); United States v. Davis, 689 F.3d 179, 186 (2d Cir. 2012); see also Bartsch v. Metro-Goldwyn-Mayer, Inc., 391 F.2d 150 (2d Cir. 1968) (using foreseeability as a measure of the scope of copyright licensing agreements); Shyamkrishna Balganesh, Foreseeability and Copyright Incentives, 122 Harv. L. Rev. 1569, 1609 (2009); Christina Bohannan, Copyright Harm, Foreseeability, and Fair Use, 85 Wash. U. L. Rev. 969 (2007). And for good reason. The idea of foreseeability aims to connect persons with events in a way that advances the law’s social goals and accords with moral intuitions.16See Avihay Dorfman, Foreseeability as Re-Cognition, 59 Am. J. Jurisprudence 163 (2014); Stephen R. Perry, Protected Interests and Undertakings in the Law of Negligence, 42 U. Toronto L.J. 247 (1992); see also Stephen R. Perry, Two Models of Legal Principles, 82 Iowa L. Rev. 787 (1997). “Even harmful action,” observes tort scholar David Owen, “cannot meaningfully be viewed as ‘wrong’ if the actor could not possibly have contemplated that the action might produce the harm.”17David G. Owen, Figuring Foreseeability, 44 Wake Forest L. Rev. 1277, 1277–78 (2009); see also H.L.A. Hart & Tony Honoré, Causation in the Law 255 (2d ed. 1985) (“[A] defendant is responsible for and only for such harm as he could reasonably have foreseen and prevented.”).

For almost as long as the law has depended on foreseeability, critics have pilloried its application as meaningless and hopelessly indeterminate.18Michael S. Moore, Placing Blame: A Theory of the Criminal Law 363–64 (1997); Thomas C. Galligan, Jr., A Primer on the Patterns of Negligence, 53 La. L. Rev. 1509, 1523 (1993) (denouncing “foreseeable” and “unforeseeable” as “magic mumbo jumbo” used “to obfuscate the policies that were really at the heart of their decisions”); Leon Green, Foreseeability in Negligence Law, 61 Colum. L. Rev. 1401, 1421 (1961) (describing foreseeability as an “argumentative word[] . . . capable of absorbing any meaning”); Patrick J. Kelley, Restating Duty, Breach, and Proximate Cause in Negligence Law: Descriptive Theory and the Rule of Law, 54 Vand. L. Rev. 1039, 1046 (2001) (contending that foreseeability is “so open-ended [that it] can be used to explain any decision”); Joshua Knobe & Scott Shapiro, Proximate Cause Explained: An Essay in Experimental Jurisprudence, 88 U. Chi. L. Rev. 165, 234 (2021); (criticizing foreseeability as “warp[ed]” and “render[ed] meaningless”); Benjamin C. Zipursky, The Many Faces of Foreseeability, 10 Kan. J.L. & Pub. Pol’y 156, 156 (2000) (describing foreseeability as “undoubtedly a muddle”). Decisionmakers, say the critics, can characterize virtually any consequence as either foreseeable or unforeseeable.19See infra Section II.B. It all depends on how one tells the story. Abstract and general descriptions of the relevant events yield findings of foreseeability. Concrete and granular descriptions of the relevant events produce findings of unforeseeability.20See infra Section II.B. Change the level of abstraction, and events shift smoothly from within the ambit of reasonable foresight to beyond the foreseeability horizon and back again.21Predictably, arbitrariness in application often leads to discrimination. See Leslie Bender & Perette Lawrence, Is Tort Law Male?: Foreseeability Analysis and Property Managers’ Liability for Third Party Rapes of Residents, 69 Chi.-Kent L. Rev. 313, 320 (1993); Martha Chamallas, Vicarious Liability in Torts: The Sex Exception, 48 Val. U. L. Rev. 133, 169–70 (2013) (contending that courts “classify sexual misconduct as unforeseeable or unexpected, despite its prevalence”); Justin D. Levinson & Kaiping Peng, Different Torts for Different Cohorts: A Cultural Psychological Critique of Tort Law’s Actual Cause and Foreseeability Inquiries, 13 S. Cal. Interdisc. L.J. 195 (2004).

Yet, to date, foreseeability has survived its critics. If simple case counting is the measure, foreseeability has flourished.22See Benjamin C. Zipursky, Foreseeability in Breach, Duty, and Proximate Cause, 44 Wake Forest L. Rev. 1247, 1259-63 (2009) (“[A]lmost every jurisdiction does treat foreseeability as a significant factor (and frequently the most significant factor) in analyzing whether the duty element is met in a negligence claim.”). Even as we write this, new foreseeability problems press upon some of the most important legal questions of our time. Can a protest leader be held liable for injuries resulting from a protest?23See Mckesson v. Doe (Mckesson V), 141 S. Ct. 48 (2020); Doe v. Mckesson (Mckesson VII), 339 So. 3d 524 (La. 2022). Foreseeability is guiding tort claims arising out of the coronavirus pandemic, such as those against cruise lines24See Hugo Martín, Death on the High Seas: Cruise Passengers Face Head Winds with COVID Lawsuits, L.A. Times (Nov. 3, 2021), https://www.latimes.com/​business/​story/​2021-11-03/​covid-cruises-lawsuits-maritime-law-princess-battles [https://perma.cc/​8SE3-6L5R]. and healthcare providers.25J. David Brittingham, Kyle R. Bunnell & Paige N. Johnson, Tort-Focused Legal Considerations for Healthcare Providers and Product Manufacturers in the COVID-19 Landscape, Nat’l L. Rev. (Mar. 19, 2020), https://www.natlawreview.com/​article/​tort-focused-legal-considerations-healthcare-providers-and-product-manufacturers [https://perma.cc/​326A-PDQC]; see also Nathan P. Nasrallah & DeAngelo A. LaVette, COVID-19 and Ivermectin Lawsuits, Am. Bar Ass’n (Oct. 28, 2021), https://www.americanbar.org/​groups/​litigation/​committees/​mass-torts/​articles/​2021/​winter2022-covid-19-and-ivermectin-lawsuits [https://perma.cc/​EG6K-VMW3]. Foreseeability will determine whether or not the pandemic qualifies as an “act of God” for those seeking to avoid contractual obligations.26See United States v. Brooks-Callaway Co., 318 U.S. 120 (1943); David J. Ball et al., Contractual Performance in the Age of Coronavirus: Force Majeure, Impossibility and Other Considerations, Nat’l L. Rev. (Mar. 18, 2020), https://www.natlawreview.com/​article/​contractual-performance-age-coronavirus-force-majeure-impossibility-and-other [https://perma.cc/​BQ4Q-P9U4].

In this Article, we propose that foreseeability has survived and flourished not because it carries determinate meaning in the abstract (it does not), but because social and professional conventions have proliferated to further the law’s basic goals in particularly thorny categories of recurring cases. Lawyers, judges, and juries have established fixes or hacks, which in this Article we call foreseeability conventions, to settle what would otherwise be intractable foreseeability problems.

We will describe two basic types of conventions here. Some foreseeability conventions are what we will call storytelling conventions or narrative conventions. Such conventions instruct decisionmakers about the appropriate level of generality at which to articulate the foreseeability inquiry. Narrative conventions are like maps: they are guides on how courts should navigate the many possible paths to recounting the facts of the cases before them. Other foreseeability conventions are categorical: they are what we call here per se conventions. These latter conventions, which arise in a variety of crucial and regularly repeating domains, label events as foreseeable or unforeseeable as a matter of law. They offer not maps or guides but one-size-fits-all solutions to the thorny foreseeability questions at issue.

Much of the discussion here is quite abstract, but two vivid illustrations help us bring out the key features of the analysis. The classic (if underappreciated) decision in a case called Kinsman Transit27In re Kinsman Transit Co. (Kinsman Transit II), 388 F.2d 821 (2d Cir. 1968); In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708 (2d Cir. 1964). and the high-stakes litigation odyssey arising out of Louisiana police officers’ dispute with Black Lives Matter (BLM) leader DeRay Mckesson28Doe v. Mckesson (Mckesson II), 935 F.3d 253, 261 (5th Cir. 2019); Mckesson V, 141 S. Ct. 48; Doe v. Mckesson (Mckesson VI), 2 F.4th 502, 503 (5th Cir. 2021), certifying questions to 320 So. 3d 416 (La. 2021); see also Doe v. Mckesson (Mckesson IV), 947 F.3d 874, 879 (5th Cir. 2020) (en banc) (Higginson, J., dissenting). put the production of foreseeability conventions in bold relief. In Part I of the Article, we begin with a prelude on the Kinsman Transit cases and focus especially on an opinion in the case by Judge Henry Friendly that offers a vivid illustration of the foreseeability problem and of the conventional solutions to which we mean to draw attention here—improbably and unforeseeably, the decision also made possible the wild ride of the modern New York Yankees franchise.29See infra note 38 and accompanying text.

In what follows, we trace the law’s use of such foreseeability conventions. Part II of the Article introduces the reader to the foreseeability controversy as it developed in modern tort doctrine and proposes our conventions-based solution. In Part III, we lay out a series of doctrinal and situational conventions that offer jurists stable footing in the foreseeability quagmire. We show that distinctive conventions—typically of the storytelling variety—have emerged to resolve knotty foreseeability problems in accord with the basic goals of general doctrinal domains like contract, crime, and tort. Other conventions—often, though not always, of the per se variety—arise to handle foreseeability problems in discrete and regularly recurring social situations. In these ways, and very likely in others not explored here, solutions to the foreseeability quagmire typically rest on more or less stable agreements among lawyers and judges to advance the basic goals of the law by means of one or another convention. In particular, we reinterpret a number of well-known doctrines in contract and (mostly) tort as establishing precisely such conventions. We map a set of rules and doctrinal guidelines designed to shape otherwise intractable foreseeability inquiries of the kind that Judge Friendly worked successfully to manage in Kinsman Transit.

The idea of foreseeability conventions faces challenges, too. For one thing, sophisticated lawyer-philosopher critics argue that versions of the conventional account we offer here cannot and do not resolve foreseeability’s difficulties. Part IV sets out objections to the conventionality idea from philosophers who contend, among other things, that there are too many available ways of framing any given case for the conventions to give traction to the foreseeability inquiry. We defend the conventions hypothesis by reference to the particular fashion in which the bilateral structure of common law litigation makes foreseeability a practical working tool for the law. In conclusion, we observe that the hotly contested, much-discussed, and seemingly interminable Mckesson case in Louisiana appeared poised to adopt a conventions solution to an otherwise grave foreseeability question—at least until the Louisiana Supreme Court erred early last year when it misunderstood the logic of the law’s conventional approach. The Conclusion notes, too, an even graver difficulty for foreseeability conventions than the philosophers’ objections. In an age of fractured politics and social disruption, the social and professional settlements on which foreseeability conventions rest are at risk. Lawyers, judges, and jurors increasingly disagree about the goals underlying the law’s conventions. Hacking these new foreseeability crises—and building new conventions for unsettled times—will be the work of jurists for years to come.

I. Prelude—The Kinsman Transit Case Study

Like so many challenging foreseeability cases, Kinsman Transit arose out of a freakish set of events. An unusual weather pattern in late January 1959 caused the frozen Buffalo River to flood, loosing ice flows downriver toward Lake Erie.30Kinsman Transit I, 338 F.2d at 711–12. Shortly before midnight on the night of the 21st, a 500-foot cargo ship known as the MacGilvray Shiras, owned by the Kinsman Transit Company and loaded with grain, broke loose from its moorings under the pressure of the ice and the current.31Id. at 712. Kinsman Transit and the dock owner had failed to secure the vessel properly, and a botched emergency effort to deploy the anchors made matters worse.32Id. The Shiras careened downriver, striking a second large cargo ship called the Tewksbury, whose shipkeeper had abandoned his station to watch television with his girlfriend.33Id. Swollen currents carried the giant vessels downriver, until they crashed into an unopened drawbridge whose operator had left his shift early to visit a nearby tavern.34Id. at 713. The collision toppled both towers of the drawbridge, injuring two drawbridge workers and nearby property.35Id. More damage quickly ensued. The Shiras and the Tewksbury had not only struck the unopened drawbridge. They had also fixed themselves into its structure in the shape of a wedge. The resulting wall blocked the river’s flow like a kind of instant dam. Water and ice backed up, flooding industrial facilities for three miles upstream, doing vast amounts of damage to the suddenly submerged mid-century city.36Id.

The case is fairly bristled with knotty foreseeability questions of the kind that trouble the law all the time in less spectacular cases. What about losses to Buffalo businesses that had relied on being able to cross the now-inoperable Michigan Avenue bridge? Were they foreseeable? How about losses to vessels trapped in the Buffalo harbor and therefore unable to go about their business? Did the owners of cargo in the hulls of those vessels suffer foreseeable damages? What of the flooding victims with vast damages miles upriver from the bridge catastrophe? Were their damages a foreseeable result of the cascading series of improbable errors? Baseball fans might pose a question of their own. The Steinbrenner family, which owned the Kinsman Transit Company, bought the New York Yankees in 1973.37                See Bill Madden, Steinbrenner: The Last Lion of Baseball 21–45 (2010). Who could have foreseen that baseball history hung in the balance in the great Buffalo River flood of 1959?38See Joe Nocera, Was Steinbrenner Just Lucky?, N.Y. Times (July 16, 2010), https://www.nytimes.com/​2010/​07/​17/​business/​17nocera.html (last visited Dec. 20, 2022). For nearly four decades, beginning in 1973, George Steinbrenner would serve as the controversial owner of the team, presiding over the restoration of its storied legacy and the resumption of its World Series-winning ways. See Maury Allen, All Roads Lead to October: Boss Steinbrenner’s 25-Year Reign over the New York Yankees (2000); Marty Appel, Pinstripe Empire: The New York Yankees from Before the Babe to After the Boss (2012); Buster Olney, The Last Night of the Yankee Dynasty: The Game, the Team, and the Cost of Greatness (2004); Bill Pennington, Chumps to Champs: How the Worst Teams in Yankees History Led to the ‘90s Dynasty (2019). The 1959 flood, however, nearly drowned Steinbrenner’s baseball future before it had even started. The family faced bankruptcy if it was held responsible for the costs of the Buffalo River disaster. Only an obscure and highly technical point rescued George and the Steinbrenner family from responsibility for the costs. Federal law capped (as it caps now) the liability of a vessel owner at the value of the vessel unless the owners themselves were personally negligent. See 46 U.S.C. § 183. Plaintiffs in the case—shoreline business owners, grain contractors, the drawbridge owner, nearby businesses, and injured workers—insisted that George was personally negligent. He ought to have been in Buffalo, not in Florida, and to have taken care to oversee the mooring of the vessel, they said. See Kinsman Transit I, 338 F.2d at 714. Judge Friendly, however, concluded that Steinbrenner’s conduct, even if negligent, did not undo the cap on the Kinsman Transit firm’s liability. Why? Judge Friendly found that George knew nothing about vessels. He had joined his father’s firm only two years before. He had no apparent knowledge of cargo ships or moorings, and (crucially) “nothing different would have been done” had he been on the scene. See id. at 715 (“‘Where a vessel is held in corporate ownership, the imputation of “privity or knowledge” to the corporate owner will be made if a corporate officer sufficiently high in the hierarchy of management is chargeable with the requisite knowledge or is himself responsible on a negligence rationale. How high is “sufficiently high” will depend on the facts of particular cases.’ Henry and George Steinbrenner were ‘sufficiently high,’ but George had no knowledge of the mooring, and Henry had none save for Davies’ report on his return from Buffalo and the United States Salvage certificate, both of which were reassuring rather than the reverse. They were not negligent in assigning the task to Davies, whose competence was established. . . . [T]here is every indication that nothing different would have been done if George Steinbrenner had been on the scene during the final mooring as he had entrusted the operation to one admittedly more competent to oversee it than he was.” (citation omitted)). Spared bankruptcy for the flooding of nearly all of industrial Buffalo (spared because of his ignorance and incompetence, no less), George was in a position to buy the New York Yankees in 1973, just five years after the final resolution of the case.

The law has grave difficulty with such open-ended foreseeability questions. Who is to say what is foreseeable and what is not, especially in the hard cases where it matters? The truth is that the answer all depends on how one asks the question. We will have more to say shortly about why foreseeability questions are so difficult. For now, the important thing to see is that the law turned out to have tools for evading the problem—for reducing imponderable and open-ended foreseeability issues to manageable inquiries. Judge Friendly deployed a handful of per se conventions to resolve certain questions. He ruled, for example, that a patient whose doctor could not reach them across the impassable Michigan Avenue bridge had no cause of action against the Kinsman Transit defendants.39Kinsman Transit I, 338 F.2d at 725. A subsequent panel of the Second Circuit ruled similarly that economic losses to a plaintiff, whose grain had been stored on a vessel downstream of the now-inoperable bridge, were not recoverable.40In re Kinsman Transit Co. (Kinsman Transit II), 388 F.2d 821, 824–25 (2d Cir. 1968). Were damages in such cases foreseeable? Well, perhaps. And perhaps not. It depends how one frames the narrative. But Judge Friendly and the Second Circuit adopted hard-and-fast categorical rules—per se conventions—to skirt the intractable foreseeability questions altogether. A categorical rule of no-duty to parties outside the impact zone of the accident took care of the claims arising out of the inconvenience of the bridge. The long-standing rule barring all pure economic loss claims resolved the claims of the grain owners.41See generally Herbert Bernstein, Civil Liability for Pure Economic Loss Under American Tort Law, 46 Am. J. Compar. L. 111 (1998).

What we call storytelling conventions helped resolve another tricky problem in the case by directing Judge Friendly to ask about the foreseeability of flooding damages to property owners along the river according to a particular script. The question, Judge Friendly ruled, was not whether the flooding damages that occurred to riparian property-owner plaintiffs were foreseeable in the abstract; rather, a narrative foreseeability convention, which below we will call the Palsgraf convention, instructed Judge Friendly to ask more simply whether the plaintiffs were foreseeable victims of the defendants’ negligence.42Kinsman Transit I, 338 F.2d at 722; see also Kinsman Transit II, 388 F.2d at 824 & n.4. Since fortuitously, all the riparian property-owner plaintiffs were located downstream of the Shiras’s moorage, Judge Friendly could say that the foreseeability requirement—so articulated—was met. And when the defendants pressed for a broader, open-ended foreseeability test that would limit damages to those that had taken place in a foreseeable manner, Judge Friendly gently demurred, choosing instead to frame the case in such a way as to hold off ill-defined foreseeability problems.43See Kinsman Transit I, 338 F.2d at 722–23; see also Kinsman Transit II, 388 F.2d at 824.

The tools at hand for Judge Friendly in Kinsman Transit were hard won in the law. They developed over time and in response to the grave conceptual crisis that foreseeability tests pose. They brought resolution to an otherwise unruly and even unmanageable case. And they allowed a leading jurist to advance basic goals of sensible loss allocation such as spreading, compensation, and deterrence. Kinsman Transit puts the problem on vivid display. But if we are to understand the tools that Judge Friendly deployed so skillfully, if we are to grasp the way in which the law today reduces foreseeability questions to conventions, we have to start long before, in the century of the modern common law’s beginning, when the foreseeability standard first emerged in modern doctrine.

II. Foreseeability and Its Discontents

Unforeseeable events are ones for which we are typically inclined to think actors bear no responsibility. And so, it is hardly surprising that foreseeability appears as a central doctrinal concept in the modern law of tort, contract, and crime. Yet for a century and more, critics have pointed to its glaring defects.

A. Triumph and Anxiety

Canonical nineteenth-century cases made the foresight concept central to foundational categories like contract and tort. The famous 1854 contract decision Hadley v. Baxendale ruled that a party may recover only those damages that “may reasonably be supposed to have been in the contemplation of both parties at the time they made the contract, as the probable result of the breach of it.”44Hadley v. Baxendale (1854) 156 Eng. Rep. 145, 145, 151–52. In the modern law of torts, foreseeability emerged early on, too, as in the English courts, which in the 1850s held that liability did not extend to injuries “which could by no possibility have been foreseen,”45Greenland v. Chaplin (1850) 155 Eng. Rep. 104, 106 (affirming verdict for plaintiff injured in steamboat accident on ground that plaintiff was not obliged to “anticipate and guard against that which no reasonable man would expect to occur”). and allowed tort claims only when the “mischief which happened . . . could have been foreseen.”46Cornman v. E. Cntys. Ry. Co. (1859) 157 Eng. Rep. 1050, 1052 (rejecting tort claim of man struck by weighing machine tipped by crowd of railroad passengers on the ground that the “proof is wanting that the mischief which happened was one which could have been foreseen”). In the American common law, Oliver Wendell Holmes, Jr. consolidated the concept by making it the animating principle of his 1881 classic, The Common Law.47See Oliver Wendell Holmes, Jr., The Common Law 56–57 (1881). It was only when a person “fails to exercise the foresight of which he is capable,” id. at 109, or fails to foresee “whatever a prudent and intelligent man would have foreseen,” id. at 146–47, that the law imposes a sanction on him. The “power of avoiding the evil complained of [is] a condition of liability,” and “[t]here is no such power where the evil cannot be foreseen.” Id. at 95; see also David Rosenberg, The Hidden Holmes: His Theory of Torts in History (1995).

By the turn of the twentieth-century, common law doctrine in torts proceeded not as a story about whether to deploy foreseeability,48See Thomas G. Shearman & Amasa A. Redfield, A Treatise on the Law of Negligence § 6 (1869) (noting that negligence “does not require from any man superhuman wisdom or foresight”); Thomas M. Cooley, A Treatise on the Law of Torts or the Wrongs Which Arise Independent of Contract 79 (2d ed. 1888) (describing recoverable injuries as those “that should have been foreseen by ordinary forecast”); Francis Wharton, A Treatise on the Law of Negligence § 15 (2d ed. 1878) (defining negligent injuries as those where the relevant actor “may foresee a probable danger”); see also Thomas William Saunders, The Law of Negligence 119 (E. Blackwood Wright ed., 2d ed. 1898) (stating that the duty of care “arises whenever a reasonable and prudent man would foresee . . . th[e] danger of injury”); Horace Smith, A Treatise on the Law of Negligence 1 (1880) (noting that one would be liable for actions that “a reasonably careful man would foresee might be productive of injury”); Morton Barrows, Handbook on the Law of Negligence 11 (1900) (“[T]he courts seemingly hold that the result [of a harmful action] must be so intimately connected with the cause, in a direct and natural sequence of events, that a man of ordinary prudence and intelligence would actually have foreseen some injurious result, although not necessarily the one that did ensue.”); id. at 16. but as a raucous doctrinal debate about where and how—and even how often—the concept would be used.49See generally W. Jonathan Cardi, The Hidden Legacy of Palsgraf: Modern Duty Law in Microcosm, 91 B.U. L. Rev. 1873 (2011) [hereinafter Cardi, The Hidden Legacy]; W. Jonathan Cardi, Purging Foreseeability, 58 Vand. L. Rev. 739 (2005) [hereinafter Cardi, Purging Foreseeability]; Dilan A. Esper & Gregory C. Keating, Abusing “Duty”, 79 S. Cal. L. Rev. 265, 314–20 (2006); Harry G. Fuerst, Foreseeability in American and English Law, 14 Clev.-Marshall L. Rev. 552 (1965); John C.P. Goldberg & Benjamin C. Zipursky, The Restatement (Third) and Duty in Negligence Law, 54 Vand. L. Rev. 657, 727–28 (2001); John C.P. Goldberg & Benjamin C. Zipursky, The Moral of MacPherson, 146 U. Pa. L. Rev. 1733, 1742, 1771–73 (1998); Green, supra note 18; Joseph W. Little, Palsgraf Revisited (Again), 6 Pierce L. Rev. 75 (2007); Zipursky, supra note 22, at 1249, 1268–70 (explaining that the Restatement (Third) of Torts writers “applaud the role of foreseeability in breach and undercut it in duty,” and then taking issue with these applications); Zipursky, supra note 18, at 158 (“[F]oreseeability does not play a single role within negligence law, but a complex variety of connected roles.”). Some treated the question of whether a party ought reasonably to have foreseen a risk as an inquiry into breach of the duty of care.50E.g., Cornman v. E. Cntys. Ry. Co. (1859) 157 Eng. Rep. 1050. Other jurists analyzed foreseeability as a tool for determining the extent of the damages.51Saunders asserted that individuals were only responsible for the “consequences that may reasonably be expected to result” from their actions. Saunders, supra note 48, at 151. In Frederick Pollock’s view, foreseeability—what the reasonable person could “be expected to foresee as likely to follow”—defined the outer limits of liability. Frederick Pollock, The Law of Torts: A Treatise on the Principles of Obligations Arising from Civil Wrongs in the Common Law 21 (1887). Still others deployed foreseeability to determine whether one party owed a duty of care to another.52See Palsgraf v. Long Island R.R., 162 N.E. 99 (N.Y. 1928).

Yet an equally important story of foreseeability in twentieth-century doctrine was its rapidly changing scope, which introduced a new note of anxiety to the law’s reliance on the foresight standard. Early cases had adopted a startlingly narrow definition of the foreseeable. In Ryan v. New York Central Railroad Co., for example, the New York Court of Appeals held that the spread of a fire to a second or third structure was not an “expected result” of negligently igniting the first.5335 N.Y. 210, 212 (1866); id. at 216 (“A man may insure his own house or his own furniture, but he cannot insure his neighbor’s building or furniture, for the reason that he has no interest in them.”). Similarly, the leading early products liability decision in Huset v. J.I. Case Threshing Machine Co. ruled that a manufacturer could not reasonably foresee injury to a third party other than the party to whom it sold the product.54120 F. 865, 867 (8th Cir. 1903) (“[W]hen . . . a manufacturer constructs a car or a carriage, for the owner thereof, under a special contract with him, an injury to any other person than the owner for whom the article is built and to whom it is delivered cannot ordinarily be foreseen or reasonably anticipated . . . .”). But cases like Ryan and Huset soon came to be seen as mistaken anachronisms.55Shearman & Redfield, supra note 48, § 30 (“[T]hese decisions have been overruled everywhere else, and are practically overruled in New York, where they originated . . . .” (footnote omitted)). Among jurists today, such cases are widely derided. See, e.g., John C.P. Goldberg, Tort in Three Dimensions, 38 Pepp. L. Rev 321, 325 (2011); Timothy D. Lytton, Responsibility for Human Suffering: Awareness, Participation, and the Frontiers of Tort Law, 78 Cornell L. Rev. 470, 490–91 (1993); Mari Matsuda, On Causation, 100 Colum. L. Rev. 2195, 2200 (2000). In New York, the fire rule has shown surprising durability. See Dan B. Dobbs, Paul T. Hayden & Ellen M. Bublick, The Law of Torts § 206 (2d ed. 2022); 59 N.Y. Jur. 2d Explosives and Fires § 70 (2022). The increased scale of modern industrial operations and the logic of insurance markets belied the courts’ truncated conception of foreseeability.56John Fabian Witt, The Accidental Republic: Crippled Workingmen, Destitute Widows, and the Remaking of American Law 140–41 (2004); John Fabian Witt, Speedy Fred Taylor and the Ironies of Enterprise Liability, 103 Colum. L. Rev. 1 (2003); see also W. Page Keeton, Dan B. Dobbs, Robert E. Keeton & David G. Owen, Prosser and Keeton on Torts 668, 682 (5th ed. 1984); Francis H. Bohlen, The Basis of Affirmative Obligations in the Law of Tort, 53 Am. L. Reg. 337, 351 (1905); Richard A. Epstein, Products Liability as an Insurance Market, 14 J. Legal Stud. 645, 658–59 (1985); Fleming James, Jr., Scope of Duty in Negligence Cases, 47 Nw. U. L. Rev. 778, 798 (1953). The expansion of insurance markets in particular, and the accompanying actuarial point of view, helped propel an epistemological transformation in foreseeability standards.57Kenneth S. Abraham, The Liability Century: Insurance and Tort Law from the Progressive Era to 9/11, at 179 (2008); see Tom Baker, Risk, Insurance, and the Social Construction of Responsibility, in Embracing Risk: The Changing Culture of Insurance and Responsibility 33, 33–34, 43, 48 (Tom Baker & Jonathan Simon eds., 2002); Jonathan Simon, The Ideological Effects of Actuarial Practices, 22 L. & Soc’y Rev. 771 (1988); see also In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708, 725–26 (2d Cir. 1964) (“[A]s society has come to rely increasingly on insurance . . . , the point [at which a loss becomes unforeseeable] may lie further off than a century ago.”). And for nearly a century, foreseeability in tort has expanded inexorably.58See generally Peter W. Huber, Liability: The Legal Revolution and Its Consequences (1990); G. Edward White, Tort Law in America: An Intellectual History (expanded ed. 2003). See also Kenneth S. Abraham & G. Edward White, Tort Law and the Construction of Change: Studies in the Inevitability of History (2022); Wex S. Malone, Damage Suits and the Contagious Principle of Workmen’s Compensation, 9 Nat’l Ass’n Claimants’ Comp. Att’ys L.J. 20 (1952); Robert L. Rabin, Tort Law in Transition: Tracing the Patterns of Sociolegal Change, 23 Val. U. L. Rev. 1, 26 (1988). In a modern economy of large institutions, of interconnections, and of social operations at scale, observers found it easy to foresee harms that had once seemed distant and improbable.59See Ulrich Beck, Risk Society: Towards a New Modernity (Mike Featherstone ed., Mark Ritter trans., Sage Publications 1992) (1986); Tom Baker & Jonathan Simon, Embracing Risk, in Embracing Risk: The Changing Culture of Insurance and Responsibility, supra note 57, at 1; Francois Ewald, The Return of Descartes’s Malicious Demon: An Outline of a Philosophy of Precaution, in Embracing Risk: The Changing Culture of Insurance and Responsibility, supra note 57, at 273; Samuel Issacharoff & John Fabian Witt, The Inevitability of Aggregate Settlement: An Institutional Account of American Tort Law, 57 Vand. L. Rev. 1571 (2004); Jonathan Simon, The Emergence of a Risk Society: Insurance, Law, and the State, Socialist Rev., Sept. 1, 1987, at 61. Then-Judge Benjamin N. Cardozo said as much in MacPherson v. Buick Motor Co. when he explained that product injury to third parties “is to be foreseen not merely as a possible, but as an almost inevitable result.”60MacPherson v. Buick Motor Co., 111 N.E. 1050, 1054–55 (N.Y. 1916). Viewed at scale, expanded foreseeability seemed capable of advancing socially important goals like deterring unwarranted risks61See Guido Calabresi, Some Thoughts on Risk Distribution and the Law of Torts, 70 Yale L.J. 499, 529 (1961). and spreading the losses from accidents.62See Fleming James, Jr., Accident Liability: Some Wartime Developments, 55 Yale L.J. 365, 381–83 (1946). Accordingly, a mass of twentieth-century cases, led by a set of California cases, carried forward the basic idea that scale and the actuarial point of view and social goals had expanded foreseeability far beyond the constricted vision of Ryan and Huset.63See Tarasoff v. Regents of the Univ. of Cal., 551 P.2d 334, 343 (Cal. 1976) (overruling old principle that “one person owed no duty to control the conduct of another” on grounds that therapists may foresee injuries to third parties arising out of dangers posed by their patients); Dillon v. Legg, 441 P.2d 912 (Cal. 1968) (abolishing old rule that negligent infliction of emotional distress was unforeseeable in the absence of physical impact); Rowland v. Christian, 443 P.2d. 561, 567–68 (Cal. 1968) (adopting open-ended foreseeability standard in landowner and occupier cases on grounds that even unusual harms such as certain harms to trespassers might, “in a particular case,” be foreseeable); Escola v. Coca Cola Bottling Co., 150 P.2d 436, 440–41 (Cal. 1944) (Traynor, J., concurring) (“[T]he manufacturer can anticipate some hazards and guard against the recurrence of others, as the public cannot.”). Outside of California, see Larsen v. Gen. Motors Corp., 391 F.2d 495, 501–02 (8th Cir. 1968) (describing “collisions and injury-producing impacts” for third parties as “inevitable contingenc[ies] of normal automobile use”); Spencer v. Madsen, 142 F.2d 820 (10th Cir. 1944) (holding trailer manufacturer liable for injury to third party); Hall v. E.I. du Pont de Nemours & Co., 345 F. Supp. 353, 362 (E.D.N.Y. 1972) (noting that injuries to children in accidents involving blasting caps were foreseeable due to statistical probability); Outwater v. Miller, 153 N.Y.S.2d 708, 711 (Sup. Ct. 1956) (describing product injuries as “almost inevitable”); White Sewing Mach. Co. v. Feisel, 162 N.E. 633 (Ohio Ct. App. 1927) (holding sewing machine manufacturer liable for injury to third party). In case after case, newly open-ended foreseeability inquiries replaced narrower, situation-specific rules.64See Robert L. Rabin, The Historical Development of the Fault Principle: A Reinterpretation, 15 Ga. L. Rev. 925 (1981).

And therein lay a new crisis for the foreseeability standard. Mid-twentieth-century tort law’s increased reliance on foreseeability came hand-in-hand with a newly expanded conception of what might count as foreseeable—which is to say: jurists came to rely on the concept at precisely a moment in which its indeterminacy was laid bare.

B. The Skeptics

Even as tort law relied on the foreseeability concept to do more and more work, at least three different kinds of skeptics emerged. A first group of critics objects on policy grounds to the open-ended authority for juries that such a test seems to produce.65E.g., Amaya v. Home Ice, Fuel & Supply Co., 379 P.2d 513, 520–22 (Cal. 1963), overruled in part by Dillon v. Legg, 441 P.2d 912 (Cal. 1968) (“[A] harm is ‘foreseeable’ only if, in the final analysis, a court or jury says that it is. . . . [T]he question is largely semantic, and any such debate tends to sterility.”); see also Kentucky v. King, 563 U.S. 452, 464–66 (2011) (rejecting a foreseeability test in Fourth Amendment exigent circumstances cases on the grounds that such a test would “introduce an unacceptable degree of unpredictability”); Migliori v. Airborne Freight Corp., 690 N.E.2d 413, 417–18 (Mass. 1998) (rejecting a foreseeability standard for negligent infliction of emotional distress claims brought by rescuers); Nycal Corp. v. KPMG Peat Marwick LLP, 688 N.E.2d 1368, 1370 (Mass. 1998) (rejecting a foreseeability standard lest accountants be exposed to “a liability in an indeterminate amount for an indeterminate time to an indeterminate class” (quoting Ultramares Corp. v. Touche, 174 N.E. 441, 444 (N.Y. 1931))). Others warn that foreseeability unduly expands liability, though (as we will see shortly) it is unclear whether this is actually so.66Tobin v. Grossman, 249 N.E.2d 419, 423 (N.Y. 1969) (“If foreseeability be the sole test, then once liability is extended the logic of the principle would not and could not remain confined.”). Still others, such as Justice Brett Kavanaugh in a recent case in the U.S. Supreme Court, express concern that undifferentiated foreseeability standards leave potential defendants with few guidelines by which to shape their behavior, or that such standards would produce undue behavioral modifications by potential defendants eager to evade liability.67Air & Liquid Sys. Corp. v. Devries, 139 S. Ct. 986, 993–95 (2019) (contending that a duty to warn of all foreseeable dangers arising out of the use of third-party parts with defendant’s product “would impose a difficult and costly burden on manufacturers, while simultaneously overwarning users”). On closer examination, the Air & Liquid Systems Corp. opinion offers not a rejection of foreseeability but a constrained version of it. The ruling institutes a duty to warn when a manufacturer’s product requires incorporation of a part that the manufacturer has reason to know makes the final product dangerous. Id. When will such a defendant manufacturer know or have reason to know its final integrated product is dangerous to users? When that dangerousness is foreseeable, obviously. Air & Liquid Systems Corp. no more escapes the logic of foreseeability than earlier efforts. Once again, it is not at all clear that these worries are well-founded. But such worries are widespread, and in light of them, it is perhaps no surprise that the New York Court of Appeals, as one authority observes, has “push[ed] against foreseeability since at least the mid-1970s.”68See Zipursky, supra note 49, at 1261–62 (citing Pulka v. Edelman, 358 N.E.2d 1019, 1022 (N.Y. 1976)).

A second group objects that the foreseeability concept fails to reflect the contours of the caselaw; such critics champion alternative tests that purport to better characterize the cases. Tort insiders may think here of the classic “directness” test offered in In re Polemis,69See In re Polemis (1921) 3 K.B. 560. the relational duty championed by the civil recourse school in tort theory,70Goldberg & Zipursky, supra note 49, at 1820–21. or the “abnormality” test recently advanced by Joshua Knobe and Scott Shapiro.71See Knobe & Shapiro, supra note 18, at 230–34.

Here, we focus on a third set of criticisms that offer a more thoroughgoing critique of the foreseeability concept. In this view, foreseeability’s problem is neither its supposed effects in the world nor the poor fit between the concept and the cases. Instead, this third critique objects that foreseeability is essentially and unsalvageably indeterminate. In hard cases, virtually any event caused by that actor can be characterized as foreseeable. Or not. It all depends. Foreseeability is an empty vessel, say the strongest critics—“a notoriously malleable ‘rule’” in the words of Jane Stapleton.72Jane Stapleton, Legal Cause: Cause-in-Fact and the Scope of Liability for Consequences, 54 Vand. L. Rev. 941, 980–81 (2001) (“[Foreseeability] is, of course, a notoriously malleable ‘rule’ that will rarely provide control because it crucially depends on how broadly or narrowly the kind/type/nature of the harm is described.”).

To see why, consider the criticisms first offered by Dean Leon Green, the legal realist who launched his career as the leading torts scholar by critiquing foreseeability standards.73See Leon Green, Are Negligence and Proximate Cause Determinable by the Same Test?—Texas Decisions Analyzed, 1 Tex. L. Rev. 423, 423 (1923). Green taught torts at the University of Texas School of Law and Yale Law School and served as dean at Northwestern University School of Law in the middle of the twentieth century. As Green saw it, foreseeability appealed to courts looking for a way to avoid complicated causation inquiries.74Id. at 435–36 (citing City of Dallas v. Maxwell, 248 S.W. 667, 670, 673 (Tex. Comm’n App. 1923)). Pushing further, Green contended that foreseeability did not decide cases at all. Courts made decisions based on the facts of the case and the intuitions those facts elicited. Courts “weighed the facts in the scales of common sense and experience after all the facts were before it.”75Id. at 437. The foreseeability standard, he suggested, was “a loose cover-all without integrity,”76Leon Green, Proximate Cause in Texas Negligence Law, 28 Tex. L. Rev. 755, 772 (1950); see also Frank J. Vandall, The Restatement (Third) of Torts, Products Liability, Section 2(b): Design Defect, 68 Temp. L. Rev. 167, 183 n.95 (1995). “a ‘word curtain’ behind which” courts concealed “professional paralysis of thought.”77Green, supra note 76, at 772. Courts that relied “wholly on the gossamer of foreseeability”—a mere “transparent fiction[]”—were “so far removed from the practical world of affairs as to suggest the Wizard of Oz or the creations of Disneyland.”78Green, supra note 18, at 1412, 1421. Drawing “the line between the foreseeable and the unforeseeable in the world of everyday affairs,” Green scoffed, “raises even more difficulties than the determination of where space leaves off and outerspace begins.”79Id. at 1413. Words like foreseeability were “capable of absorbing any meaning given them.”80Id. at 1421.

Clarence Morris, who was Green’s colleague at the University of Texas and then taught torts for over two decades at the University of Pennsylvania,81See Jefferson B. Fordham, Clarence Morris, 121 U. Pa. L. Rev. 419, 420 (1973). sharpened Green’s critique. In Morris’s assessment, there were three types of foreseeability cases.82Clarence Morris, Duty, Negligence and Causation, 101 U. Pa. L. Rev. 189, 196–98 (1952). Some cases were so typical in their structure that no jury could be persuaded to think of the injury as unforeseeable.83Id. at 196. Others were so freakish and unexpected that the opposite was true.84Id. at 196–97. The vast majority of cases lay in the middle, and, as to those, the outcome of the foreseeability inquiry turned entirely on how the court chose to characterize the facts of the case. “If the official description of facts adopted by the court is detailed,” Morris explained, “the accident is called unforeseeable; if it is general, the accident is called foreseeable.” The real question for the law was about the appropriate level of generality at which the facts of the case should be characterized. Everything turned on it. But, as Morris observed grimly, there was “no authoritative guide to the proper way to describe facts.” There was no law governing how to craft the relevant descriptions. And as a result, court decisions holding that a loss was foreseeable or not were “fluid,” in Morris’s gentler term.85Id. at 198. Morris’s critique soon became standard in the field. For mid-century giants Prosser and Keeton, it was common sense that the manner in which a court described a given event determined whether or not the event was found to be foreseeable or unforeseeable. Prosser and Keeton put the point this way: In one sense, almost nothing is entirely unforeseeable, since there is a very slight mathematical chance, recognizable in advance, that even the most freakish accident which is possible will occur, particularly if it has ever happened in history before. In another, no event whatever is entirely foreseeable, since the exact details of a sequence never can be predicted with complete omniscience and accuracy. If one takes a very broad, type-of-harm perspective in describing both the “foreseeable risk” and the “result” of which the plaintiff is complaining, very likely the result will appear to be within the foreseeable risk. Keeton, Dobbs, Keeton & Owen, supra note 56, § 43, at 297 (footnote omitted). Some of the most highly respected jurists in the Anglo-American world dissented. But their dissents were oddly nonsensical. Consider two leading theorists on causation and the law: [I]f the negligent act is firing a revolver, the harm is foreseeable if it is to things within the apparent range of the revolver, made of a material apparently permeable to bullets. On the other hand, if the shot punctures an inkpot which discharges ink on plaintiff’s linen, the harm to the linen is not foreseeable (unless the presence of the inkpot within range was itself known to defendant) because the generalization about ink staining neighbouring objects when it is able to flow freely would not have been cited in support of the judgment that the person firing the revolver had acted negligently. Hart & Honoré, supra note 17, at 258. Nothing in logic excludes inked linens from the foreseeable category of damaged property, which is the point critics like Morris made so powerfully. Others might have said arbitrary. The point was that the availability of multiple ways of telling any given story, from the most granular to the most abstract, seemed to leave a test like the foreseeability standard hopelessly indeterminate.86Critiques based on the levels of generality in framing problems reverberate through many areas of the law. Recent accounts in legislation include Anita S. Krishnakumar, Statutory History, 108 Va. L. Rev. 263 (2022), Anita S. Krishnakumar, Backdoor Purposivism, 69 Duke L.J. 1275, 1313 (2020), and William N. Eskridge, Jr. & Victoria F. Nourse, Textual Gerrymandering: The Eclipse of Republican Government in an Era of Statutory Populism, 96 N.Y.U. L. Rev. 1718 (2021). See generally Keith E. Whittington, Constitutional Interpretation: Textual Meaning, Original Intent, and Judicial Review 187 (1999). For classics, see Bruce Ackerman, Liberating Abstraction, 59 U. Chi. L. Rev. 317 (1992) (government powers and individual rights); Mark Kelman, Interpretive Construction in the Substantive Criminal Law, 33 Stan. L. Rev. 591 (1981) (criminal law); Daryl J. Levinson, Framing Transactions in Constitutional Law, 111 Yale L.J. 1311 (2002) (constitutional law); John F. Manning, Federalism and the Generality Problem in Constitutional Interpretation, 122 Harv. L. Rev. 2003, 2016–17 (2009) (federalism); Max Radin, Statutory Interpretation, 43 Harv. L. Rev. 863, 875 (1930) (legislation); Frederick Schauer, The Generality of Law, 107 W. Va. L. Rev. 217, 229–31 (2004) (legal reasoning); Frederick Schauer, Precedent, 39 Stan. L. Rev. 571, 596 (1987) (common law precedent); Peter J. Smith, Originalism and Level of Generality, 51 Ga. L. Rev. 485 (2017) (original public meaning); Laurence H. Tribe & Michael C. Dorf, Levels of Generality in the Definition of Rights, 57 U. Chi. L. Rev. 1057 (1990) (rights definition).

At much the same moment, philosopher Donald Davidson was developing a sophisticated version of Morris’s critique. Statements about causes and effects, Davidson observed, are highly sensitive to the many different true ways of describing a given event.87See Donald Davidson, Actions, Reasons, and Causes, 60 J. Philosophy 685, 686 (1963). A richly described causal statement, replete with detail and including surrounding conditions, is more likely to supply a sufficient causal explanation of some effect than a sparsely described causal statement. Why? Because the richly populated sentence encompasses a fuller array of the causes and conditions required to produce the relevant effect. Conversely, the same richly described causal statement is less likely to supply a necessary causal connection to the same effect (identically characterized) than a sparsely described causal statement because the former description excludes alternative paths to the same effect. Flipping the point around, fuller descriptions of effects make a given causal statement less likely to supply a sufficient causal explanation and more likely to supply a necessary causal connection. Sparse descriptions of effects, in turn, make causal statements more likely to supply sufficient causal explanations and less likely to supply necessary causal connections.88Donald Davidson, Causal Relations, 64 J. Philosophy 691, 698 (1967).

Because causal relations have moral significance, the upshot of Davidson’s insight was that deciding a person’s moral relationship to an event turns on which myriad of possible descriptions the decisionmaker selects. Lawyer-philosopher Michael Moore calls this the “multiple description problem.”89Moore, supra note 18, at 364. As Moore sees it, the key flaw in the foreseeability concept is the one Davidson identified. The problem is not that the term foreseeability is ambiguous (i.e., open to multiple interpretations), nor that it is vague (i.e., unclear at its edges).90Id. Foreseeability’s defect is rather an indeterminacy borne of the bewildering number of available ways to characterize the underlying facts to which the foreseeability test is then applied. Was it foreseeable that helping a passenger with an unobtrusive newspaper package to board a moving train would cause fireworks to explode inside the station and cause harm to a young mother at the other end of the platform many feet away? Probably not. Was it foreseeable that negligence by railroad personnel might injure passengers on the platform? Almost certainly yes. Both sentences accurately describe the facts underlying the famous Palsgraf case. Judge Cardozo chose the former description and ruled for the defendant railroad; Judge William Andrews selected the latter characterization and dissented.91Richard A. Posner, Cardozo: A Study in Reputation 42–46 (1990); see Palsgraf v. Long Island R.R., 162 N.E. 99 (N.Y. 1928).

Torts supply especially useful examples here because the canonical cases abound in examples of foreseeability’s indeterminacy. Surely it was not reasonably foreseeable that a vessel would break free of its mooring in an unlikely winter thaw and speed downriver, knocking another vessel free, and that the two ships would then slam into a carelessly managed drawbridge, producing a wedge and causing flooding for miles back upstream all the way to the initial mooring. On the other hand, it was just as surely foreseeable that negligently mooring a vessel could do damage to downstream property owners. Both sentences—one granular, the other abstract—accurately describe the facts of the great case of the 1959 Buffalo flood.92See In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708, 711–13 (2d Cir. 1964). Judge Friendly chose the latter description in favor of the plaintiff property owners.93Id. at 725 (“We see no reason why an actor engaging in conduct which entails a large risk of small damage and a small risk of other and greater damage, of the same general sort, from the same forces, and to the same class of persons, should be relieved of responsibility for the latter simply because the chance of its occurrence, if viewed alone, may not have been large enough to require the exercise of care.”). We will have more to say later about why Judge Friendly chose the description he chose. For now, the important point is that one can easily see why Moore and Heidi Hurd call foreseeability “incoherent”94Moore, supra note 18, at 398; Heidi M. Hurd & Michael S. Moore, Negligence in the Air, 3 Theoretical Inquiries Law 333, 333 (2002). and so “manipulable”95Michael S. Moore, The Metaphysics of Causal Intervention, 88 Calif. L. Rev. 827, 851 (2000). as to license “purely arbitrary judgement.”96Moore, supra note 18, at 392; see Moore, supra note 95; Hurd & Moore, supra note 94. Moore contends that the foreseeability concept retains its usefulness in the negligence inquiry, even if not in the proximate cause inquiry, because (as he contends) the Hand formula for negligence accounts for all possible harms arising out of some act or activity, no matter how unlikely, discounted for their relative probability. See Moore, supra note 18, at 397–98. Uncharacteristically, Moore is too optimistic. The Hand formula for negligence retains the agent-specific epistemological standpoint of the proximate causation inquiry and requires that the jurist decide what harms were reasonably foreseeable to a reasonable person in the actor’s position. See United States v. Carroll Towing Co., 159 F.2d 169, 173 (2d Cir. 1947). From this standpoint, the full train of conceptual horrors that Moore documents in the proximate causation inquiry inevitably follows. It is no wonder that the assault on foreseeability continues apace in the literature97See, e.g., Cardi, Purging Foreseeability, supra note 49, at 740; Eric A. Johnson, Dividing Risks: Toward a Determinate Test of Proximate Cause, 2021 U. Ill. L. Rev. 925 (2021); Kelley, supra note 18, at 1046 (noting that foreseeability is “so open-ended [that it] can be used to explain any decision, even decisions directly opposed to each other”). and that jurisdictions like New York have sharply reduced the role of foreseeability in determining the duty owed by a defendant to a plaintiff.98Zipursky, supra note 49, at 1260–62. The Restatement (Third) of Torts follows suit and objects to the use of foreseeability in determining duty, channeling Leon Green from beyond the grave.99Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 7 cmt. j (Am. L. Inst. 2010). Note that the Restatement (Second) of Torts championed the foreseeability test. See Restatement (Second) of Torts § 320 (Am. L. Inst. 1965); id. at cmt. d (detailing that “[o]ne who has taken custody of another” has a “[d]uty to anticipate danger”); id. § 319 (excluding conduct that is “unforeseeable” from the “duty of those in charge of persons having dangerous propensities”); id. § 413 cmt. d (including foreseeable circumstances in an employer’s duty to take special precautions).

C. Reinventions? (Harm Within the Risk)

And yet for all the criticisms of foreseeability, the concept bids to be the Banquo of American law.100See generally William Shakespeare, Macbeth (Kenneth Muir ed., 1957). There are, of course, many candidates for this role. See, e.g., Inga Markovits, Invitation to a Picnic, 60 Am. J. Compar. L. 856, 858 (2012) (reviewing Encyclopedia of Law & Society: American and Global Perspectives (David S. Clark ed., 2007)) (comparing the right to a jury trial to Lord Banquo in Macbeth). It remains a vital part of a wide array of doctrinal fields.101See supra notes 1–15 and accompanying text. Tort doctrine, as we have noted, has become more reliant (not less) on the foreseeability standard over the last three quarters of a century, thanks to cases like Dillon v. Legg, Rowland v. Christian, and Tarasoff v. University of California. By 1963, torts scholar Robert Keeton reported that foreseeability had become the dominant interpretation of proximate cause, quietly influencing even those courts that purported to apply contrary tests of directness and indirectness.102Robert E. Keeton, Legal Cause in the Law of Torts 41–45 (1963).

One effort to blunt the critiques of the concept observes that the law sets outer bounds for how to describe the relevant events in a foreseeability inquiry. As early as 1875, courts held that the “precise form” of injury does not need to have been foreseeable.103Hill v. Winsor, 118 Mass. 251, 251 (1875). Jurists quickly picked up on the point. Sir Frederick Pollock included it in his 1887 torts treatise.104Pollock, supra note 51, at 34 (“Perhaps the real solution is that here, as in Hill v. New River Co., the kind of harm which in fact happened might have been expected, though the precise manner in which it happened was determined by an extraneous accident.”). More than a century later, the same point is hornbook tort law.105See 1 Fowler V. Harper, Fleming James, Jr. & Oscar S. Gray, The Law of Torts §§ 18.2, 20.5 (2d ed. 1986). Versions of it can be found in every circuit106See, e.g., Bobo v. Tenn. Valley Auth., 855 F.3d 1294, 1305 (11th Cir. 2017); Munn v. Hotchkiss Sch., 795 F.3d 324, 331 (2d Cir. 2015); In re Great Lakes Dredge & Dock Co., 624 F.3d 201, 211 (5th Cir. 2010); Kleinknecht v. Gettysburg Coll., 989 F.2d 1360, 1369 (3d Cir. 1993); Jorgenson v. Mass. Port Auth., 905 F.2d 515, 523 (1st Cir. 1990); Tropea v. Shell Oil Co., 307 F.2d 757, 766 (2d Cir. 1962). and all three editions of the Restatement of Torts.107Restatement (First) of Torts § 435 (Am. L. Inst. 1934); id. at ch. 16, tit. B, intro. note (“[C]ertain factors such as the fact that the actor had no reason to foresee that the other would be harmed in the precise manner in which the harm was sustained, are not sufficient to relieve the actor from liability . . . .”); Restatement (Second) of Torts § 435 (Am. L. Inst. 1965) (“[T]he fact that the actor neither foresaw nor should have foreseen the extent of the harm or the manner in which it occurred does not prevent him from being liable.”); Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 29 cmt. d (Am. L. Inst. 2010) (“If the harms risked by that tortious conduct include the general sort of harm suffered by the plaintiff, the defendant is subject to liability for the plaintiff’s harm.”); see also id. at cmt. i (discussing “the appropriate level of generality at which to characterize the harms”).

Yet, setting outer bounds for the description of contested events still leaves vast space for characterizing harms. Some courts and jurists have therefore tried to rescue the foreseeability standard by giving it more definition and structure. One such effort insists that the foreseeability inquiry examines the foreseeability of a “class of persons” in which the relevant party was a member, rather than the foreseeability of any one particular person.108Valentine v. On Target, Inc., 727 A.2d 947, 949 (Md. 1999); see also Cardi, The Hidden Legacy, supra note 49, at 1895 & n.56 (collecting cases). Yet, nothing in this kind of class-level analysis constrains the scope of the relevant class, which courts still seem free to frame in as narrow or broad a manner as they please. Courts adopting a variety of further strategies run into similar difficulties. Confining damages to harm of a foreseeable type, for example, merely produces the dilemma of how to characterize the relevant type. In the Buffalo flood case of 1964, was the foreseeable type of downstream harm property damage or collision damage?109See Hart & Honoré, supra note 17, at 269–74; cf. Overseas Tankship (U.K.) Ltd. v. Morts Dock & Eng’g Co. (Wagon Mound I) [1961] AC 388, 390. The flooding that occurred would be the same type as the former, but not as the latter. Another seemingly more constraining strategy is to confine foreseeable events to those for which there have been “prior similar incidents.” But this accomplishes little, since the analysis begs the question which events are similar and which are not, which, in turn, produces the same level of generality problems that beset the foreseeability inquiry in the first place.110Clohesy v. Food Circus Supermarkets, Inc., 694 A.2d 1017, 1024–26 (N.J. 1997); Laura DiCola Kulwicki, A Landowner’s Duty to Guard Against Criminal Attack: Foreseeability and the Prior Similar Incidents Rule, 48 Ohio St. L.J. 247 (1987).

The so-called harm-within-the-risk approach has gotten more traction than either the class-level strategy or the prior incidents analysis, but it, too, reproduces foreseeability’s underlying problem. The harm-within-the-risk can be traced at least back to 1874 in the English case of Gorris v. Scott,111See Gorris v. Scott (1874) 9 LR Exch. 125; Hurd & Moore, supra note 94, at 340–41. On the history of the harm-within-the-risk rule, see Richard W. Wright, The Grounds and Extent of Legal Responsibility, 40 San Diego L. Rev. 1425, 1494 (2003). where the plaintiff’s sheep were swept overboard and drowned when the defendant failed to comply with a Contagious Disease Act requirement that domestic animals be held in pens during shipping into British ports.112Gorris, 9 LR Exch. 125. The Gorris court rejected the plaintiff’s argument for liability under the Act, ruling that the risk against which the Act guarded was the communication of disease among animals, not the washing of animals overboard.113Id. The harm-within-the-risk standard asks not whether the injury in question flowed foreseeably from the negligent act at issue, but instead whether that injury was one of the risks that made the defendant’s conduct negligent in the first place. The New York Court of Appeals wrote a version of this test into law in Palsgraf,114See Palsgraf v. Long Island R.R., 162 N.E. 99 (N.Y. 1928). where Judge Cardozo determined that the railroad defendant’s conduct, even if negligent toward some passengers, was “not negligence at all” relative to the unforeseeable plaintiff, who stood far away from the conduct in question.115Id. at 99–100 (“The risk reasonably to be perceived defines the duty to be obeyed, and risk imports relation; it is risk to another or to others within the range of apprehension.”). Judge Cardozo’s test, in turn, was championed by scholars like Warren Seavey and Robert Keeton116Warren A. Seavey, Principles of Torts, 56 Harv. L. Rev. 72 (1942); Warren A. Seavey, Mr. Justice Cardozo and the Law of Torts, 52 Harv. L. Rev. 372 (1939); see also Keeton, supra note 102, at 10. Keeton’s example was the famous episode of a chef placing rat poison near the stove. If the poison explodes from the heat of the stove and injures a delivery man, this harm is not within the risk of the negligence. Id. at 1, 5–6. According to Justice Cardozo’s harm-within-the-risk standard, then, the chef is not liable for the delivery man’s injuries. If the poison accidentally got into the food, however, the chef would be liable, because this harm follows from the risk. See Larrimore v. Am. Nat’l Ins. Co., 89 P.2d 340 (Okla. 1939). and inscribed into the Restatement (Second) of Torts.117The Restatement (Second) of Torts carried the idea forward, asserting that successful plaintiffs must show that they are within the class of people foreseeably at risk from the defendant’s negligence. Restatement (Second) of Torts § 281(b) (Am. L. Inst. 1965); see also Restatement (First) of Torts § 281(b) (Am. L. Inst. 1934) (“[An] actor is liable for an invasion of an interest of another, if . . . the conduct of the actor is negligent with respect to such interest . . . .”); Hurd & Moore, supra note 94, at 342.

D. Crisis

And yet, the harm-within-the-risk test has problems of its own. For one thing, the test contemplates that a person may act wrongfully and thereby cause injury to an innocent third party without accountability. While some distinguished critics, like Professor Zipursky, conclude that the rule is justified, others, such as Judge Friendly, ask why a plaintiff should lack recourse against a wrongdoer merely because the wrongdoer’s wrong turned out to cause even more harm of more types than were expected.118Compare Zipursky, supra note 49, at 1268–70, with In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708, 725 (2d Cir. 1964). Cf. Wright, supra note 111, at 1493–94 (contending that the risk formulation seems “both too stringent and too lenient” at once).

A second criticism sets aside the moral objection to insist that the risk rule is not really an advance on the foreseeability test. In this view, the risk rule shares the latter rule’s incoherence because it is equally dependent on a decisionmaker’s description of the underlying events. Green’s, Morris’s, Davidson’s, and Moore’s critiques of foreseeability apply fully to harm-within-the-risk. General descriptions of the risks created by an actor’s negligent conduct will tend to produce determinations that a harm is within the scope of the relevant risk. Highly granular descriptions of the same risk will tend to produce an opposite conclusion.119Wright, supra note 111, at 1479.

Consider again Judge Friendly’s decision in the Buffalo River case. When the Kinsman Transit cargo ship broke free of its dock in the unexpected winter thaw, what risk did it produce? On whom did it impose that risk? Did it produce a risk of vessel collisions as it careered downstream? Or did it produce a risk of property damage? Of course, the uncontrolled vessel produced all these risks. Each description is available as a true characterization of the events underlying the case, even though the competing descriptions produce different results. The way one characterizes the relevant risk matters every bit as much as it does in the foreseeability analysis.120See generally Kinsman Transit I, 338 F.2d 708.

All of which is to say that at the beginning of the twenty-first century, after nearly two hundred years of relying on the doctrine of foreseeability and seeking to develop better alternatives, the law seems not to have come up with a doctrine that solves the problem of foreseeability’s multiple descriptions. At the heart of tort doctrine lies a seemingly uncontrolled, arbitrary choice. Decisionmakers appear to be without guidance or constraint.

But are they?

III. A Conventional Account

In this Part, we develop and defend the idea that in many instances, though not all, decisionmakers do, in fact, have guidance and are, in fact, constrained. They are not constrained, to be sure, by theory or doctrine, or at least not exactly. Instead, their decisions are shaped by conventions that resolve the crisis of multiple descriptions by delivering a rule of thumb or conventional answer to the foreseeability inquiry.

A. The Conventions Idea

Holmes glimpsed a piece of the solution when he asserted that the law measured reasonable foresight by reference to the “common experience” of the “ordinary man of reasonable prudence.”121Holmes, supra note 47, at 57. Common experience supplied a common stock of assessments from which guidelines for foreseeability might be developed.

Leon Green also found value in the concept, despite his critique of it. Green believed that foreseeability helped resolve cases by serving as a placeholder that allowed judges and juries the space to make decent judgments according to their intuitions about how to resolve the damages claim before them. The open-endedness of the foreseeability standard, as Green saw it, was a virtue; it allowed wise decisionmakers to make good judgments and to advance important social values in particular disputes. The concept’s value, he explained, was that it left “the evaluation of the factual data of the case for the intelligence of the advocate and of the jury” and for “the judgment scales of the triers.”122Green, supra note 18, at 1421. Green’s view turned foreseeability into a vehicle for delivering the right values to resolve disputes. We find much value in Green’s assessment, so far as it goes. But his theory offered no help in those cases in which we disagree about the relevant values. Green’s open-ended placeholder theory quickly and unnecessarily leads to a kind of doctrinal nihilism.

A better account starts with the multiple description problem and observes how the law has developed resolutions of that problem for particular social and doctrinal settings. What the law actually does is offer conventions for how to tell certain stories. To say it more precisely, the law draws on common experience to offer prescriptions for how to talk about foreseeability. Sometimes, the law ascribes foreseeability (or unforeseeability) to certain stereotyped social situations. At other times, the law selects the appropriate level of generality at which to characterize the relevant facts in a particular category of legal problems. Both strategies serve as foreseeability conventions to resolve knotty foreseeability conundrums according to the social values embedded in the law. Both strategies were readily apparent in the Buffalo flood case in 1964.123Compare Kinsman Transit I, 338 F.2d at 725 (holding defendant-city liable for injuries “of the same general sort” as those risked by its conduct), with id. (rejecting liability for injuries caused by delay of doctor). And both strategies pop up in doctrinal fields such as contract, tort, and crime.

B. Field-Specific Conventions

1. The Canonical Convention: Contract Versus Tort

The conventionality of foreseeability is clearest in one of the most famous doctrinal contrasts in the common law. Tort and contract actions, lawyers and judges often say, diverge sharply for damages purposes. Plaintiffs recover more extensive and remote damages in tort than in contract.124See E. Allan Farnsworth, Contracts 875 (1982); Keeton, Dobbs, Keeton & Owen, supra note 56, § 92, at 665; John C.P. Goldberg & Benjamin C. Zipursky, Vosburg v. Baxendale: Recourse in Tort and Contract, in Civil Wrongs and Justice in Private Law 463, 463–83 (Paul B. Miller & John Oberdiek eds., 2020). This is the ordinary view. And it is correct.125Among other things, the divergence between contract damages and tort damages is a central reason for the pervasive skirmishing over the tort-contract boundary in litigation. Vast energies are spent by parties trying to push their disputes into one box rather than the other for purposes of shaping the applicable damages rules.

But here is a striking thing about foreseeability in tort and contract. These two doctrinal domains employ nearly identical linguistic formulae for measuring the recoverability of damages. Damages in contract are recoverable under the canonical rule of Hadley126Hadley v. Baxendale (1854) 156 Eng. Rep. 145. if they were within the foreseeable contemplation of the parties at the time of contract formation.127Id.; Restatement (First) of Contracts § 330 (Am. L. Inst. 1932) (“[C]ompensation is given for only those injuries that the defendant had reason to foresee as a probable result of his breach when the contract was made.”); U.C.C. § 2-715(2)(a) (Am. L. Inst. & Unif. L. Comm’n 1977) (“Consequential damages resulting from seller’s breach include . . . any loss resulting from general or particular requirements and needs of which the seller at the time of contracting had reason to know . . . .”); William R. Anson, Principles of the Law of Contract: With a Chapter on the Law of Agency § 404, at 490 (Arthur L. Corbin ed., 16th ed. 1924); 3 Samuel Williston, The Law of Contracts § 1356, at 2420 (1920). Damages in tort, similarly, are often said to be recoverable according to the same foreseeability test—they, too, many courts have said, may be recovered only if they were reasonably foreseeable.128E.g., Mendelson v. Davis, 281 F. 18 (8th Cir. 1922); Nunan v. Bennett, 212 S.W. 570 (Ky. 1919); Gaupin v. Murphy, 145 A. 123 (Pa. 1928); Sears v. Tex. & New Orleans Ry. Co., 247 S.W. 602 (Tex. Civ. App. 1922); Stephens v. Mut. Lumber Co., 173 P. 1031 (Wash. 1918); Overseas Tankship (U.K.) Ltd. v. Morts Dock & Eng’g Co. (Wagon Mound I) [1961] AC 388, 390. There is contrary precedent, to be sure, but the point here is that even in those jurisdictions where one version or another of the Wagon Mound I foreseeability rule applies, no one doubts that there is, nonetheless, still a difference between contract and tort damages.

The formal similarity in language of the tests for contract and tort damages has led some observers to contend that the damages rules in the two domains are actually the same.129See Banks McDowell, Foreseeability in Contract and Tort: The Problems of Responsibility and Remoteness, 36 Case W. Rsrv. L. Rev. 286, 289 (1985); see also Stephen A. Smith, Duties, Liabilities, and Damages, 125 Harv. L. Rev. 1727, 1728 (2012). This is not quite right, but it offers an insight. The difference between contract damages and tort damages lies beneath the formal analysis of the doctrine. The difference exists principally as a narrative convention of foreseeability that instructs decisionmakers in contract and tort cases about the level of generality at which to resolve foreseeability controversies. In tort, the law instructs decisionmakers to tell the relevant stories at an abstract level of generality. In contract cases, by contrast, the law instructs decisionmakers to tell the relevant stories at a finely detailed level of specificity.130Compare supra text accompanying notes 102–07, with Restatement (Second) of Contracts § 351 cmt. a (Am. L. Inst. 1981) (“The mere circumstance that some loss was foreseeable, or even that some loss of the same general kind was foreseeable, will not suffice if the loss that actually occurred was not foreseeable.”).

Consider Hadley itself, in which (as every contracts student knows) a plaintiff grain-milling firm sued a common carrier for breach of contract after the defendant failed to deliver a crankshaft according to the contract.131Hadley v. Baxendale (1854) 156 Eng. Rep. 145. Was it foreseeable or within the contemplation of the parties that a miller might lose profits in the absence of a crankshaft? Every reader of this paper by now knows that the answer could well be yes, and could well be no. The question at issue is how one characterizes the risk at issue. The genius of Hadley is precisely that it issues an instruction for how to tell the story of the relevant facts. Baron Alderson explained that the plaintiff can recover lost profits as damages only if the specific risk in dispute was part of the contracting process in a particular and granular way:

[I]f the special circumstances under which the contract was actually made were communicated by the plaintiffs to the defendants, and thus known to both parties, the damages resulting from the breach of such a contract, which they would reasonably contemplate, would be the amount of injury which would ordinarily follow from a breach of contract under these special circumstances so known and communicated.132Id.

Seen our way, Hadley’s significance is not to set the standard of foreseeability as such, but rather to establish a narrative convention for how to carry out foreseeability inquiries. Contract foreseeability, Hadley instructs, is granular and specific rather than general and abstract.133See id.; see also Howard v. Stillwell & Bierce Mfg. Co., 139 U.S. 199, 207–08 (1891) (noting that the flour mill plaintiff in Hadley could not recover lost profits because these losses were not foreseeable, the “special circumstances” not having been “communicated by the plaintiffs to the defendants” at the time of the contract’s formation). Courts are to ask whether the particular loss at issue was foreseeable, not merely whether some loss was foreseeable, or even some loss of the same general type.

The foreseeability convention in contracts implements distinctive values specific to contract law. In tort, high levels of generality allow courts to insist that public values be protected, regardless of the will of particular parties. Such public values may include reducing accident costs,134Guido Calabresi, The Costs of Accidents: A Legal and Economic Analysis (1970). compensating injury victims,135Stephen D. Sugarman, Doing Away with Personal Injury Law (1989). promoting corrective justice,136Ernest J. Weinrib, The Idea of Private Law (2012). offering civil recourse,137John C.P. Goldberg & Benjamin C. Zipursky, Recognizing Wrongs (2020). sanctioning culpable conduct,138Ketan Ramakrishnan, The Problem with Palsgraf (July 2022) (on file with author). or supporting social justice.139Martha Chamallas, Social Justice Tort Theory, 14 J. Tort L. 309 (2021). In any of these formulations, however, tort’s public values depart from the promotion of private ordering embedded in contract law. The Hadley convention of specificity responds to the felt value of private ordering in contract adjudication.140Cf. Victor P. Goldberg, Rethinking Contract Law and Contract Design 117 (2015) (arguing that in cases utilizing the Hadley convention, the real concern is “the ability of the respective parties to control the outcome”). In the famous opinion in Globe Refining Co. v. Landa Cotton Oil Co.,141190 U.S. 540 (1903). for example, Justice Oliver Wendell Holmes, Jr. explained that contract damages are based on terms set by the parties, while tort damages are set “by force of the law.”142Id. at 543. “[U]nlike the case of torts,” Justice Holmes expanded, “as the contract is by mutual consent, the parties themselves, expressly or by implication, fix the rule by which the damages are to be measured.”143Id. Only a specific and granular convention for telling foreseeability stories protects the different mechanisms of the two domains—one public and general, the other private and tailored.144See also Vanderbeek v. Vernon Corp., 50 P.3d 866, 870–71 (Colo. 2002) (“Under either a tort or a contract standard, the foreseeability of the consequences is a factor. However, the test derived from Hadley imposes a more restrictive foreseeability limitation. To be recoverable under the Hadley test, consequential damages must be so likely that ‘it can fairly be said’ both parties contemplated these damages as the probable result of the wrong at the time the tort occurred. Under the tort standard, damages need only be reasonably foreseeable.”). As Ayres and Gertner pointed out nearly a century later, a granular specificity obligation in contract foreseeability goes a step further. Not only does it protect parties’ capacity to make enforceable private arrangements, but the specificity convention for contracts may produce better arrangements by promoting the sharing of information between the parties.145See Ian Ayres & Robert Gertner, Strategic Contractual Inefficiency and the Optimal Choice of Legal Rules, 101 Yale L.J. 729, 762–63 (1992); Ian Ayres & Robert Gertner, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rules, 99 Yale L.J. 87, 94 (1989).

2. Foreseeability Conventions in the Criminal Law

Foreseeability is a crucial dimension of criminal liability, too. How could it not be? In criminal negligence cases, courts assert that the relevant injury must have been reasonably foreseeable in order for the defendant to have the right kind of mental state for moral culpability.146See United States v. Guillette, 547 F.2d 743, 749 (2d Cir. 1976); People v. Warner-Lambert Co., 414 N.E.2d 660, 664–65 (N.Y. 1980). In felony-murder rule cases, many states require that the felony in question was foreseeably dangerous to human life.147Jenkins v. State, 230 A.2d 262, 269 (Del. 1967); Ford v. State, 423 S.E.2d 255, 256 (Ga. 1992); Fisher v. State, 786 A.2d 706, 728–29 (Md. 2001); State v. Nunn, 297 N.W.2d 752, 754 (Minn. 1980); State v. Thompson, 185 S.E.2d 666, 672 (N.C. 1972). States use the terms “inherently” and “foreseeably” interchangeably. See, e.g., People v. Washington, 402 P.2d 130, 133 (Cal. 1965) (referring to an “inherently dangerous felony” and murder as “a risk reasonably to be foreseen”). See generally Erwin S. Barbre, Annotation, What Felonies Are Inherently or Foreseeably Dangerous to Human Life for Purposes of Felony-Murder Doctrine, 50 A.L.R.3d 397 §§ 1–2 (1973). In attenuated causation cases, courts hold that criminal liability attaches when “the ultimate harm is something which should have been foreseen,”148People v. Kibbe, 321 N.E.2d 773, 776 (N.Y. 1974). even when those consequences come about in unusual or unexpected ways.149See Sanford H. Kadish, Complicity, Cause and Blame: A Study in the Interpretation of Doctrine, 73 Calif. L. Rev. 323, 361–62 (1985); Moore, supra note 95, at 847–48; see also People v. Scott, 185 N.W.2d 576, 580 (Mich. Ct. App. 1971); People v. Reagan, 723 N.E.2d 55 (N.Y. 1999); People v. Roth, 604 N.E.2d 92, 93–94 (N.Y. 1992); Warner-Lambert Co., 414 N.E.2d at 666; Kibbe, 321 N.E.2d at 776; People v. Kane, 616 N.Y.S.2d 554, 554 (App. Div. 1994); Commonwealth v. Root, 170 A.2d 310, 313–14 (Pa. 1961). Foreseeability can bolster defenses, such as when a victim consents to a foreseeably dangerous activity.150Model Penal Code § 2.11(2)(b) (Am. L. Inst. 1962). And foreseeability can limit the defense of duress, when defendants place themselves in a situation where it is foreseeable that they would be coerced to commit a crime, though only where negligence is sufficient to establish culpability.151Id. § 2.09(2).

The doctrinal language of foreseeability is familiar, and distinguished commentators such as Moore have contended that the foreseeability tests are formally identical in tort and crime.152Michael Moore, Causation in the Law, Stanford Encyc. of Phil. Archive § 5.3 (Oct. 3, 2019), https://plato.stanford.edu/‌archives/‌win2019/‌entries/‌causation-law [https://perma.cc/‌3HW7-P3DK] (noting how foreseeability as a proximate cause standard is a “rule universally applicable to all criminal and tort cases”). A number of courts say that they see matters the same way,153State v. McFadden, 320 N.W.2d 608, 612–13 (Iowa 1982) (holding that “[p]roximate cause is based on the concept of foreseeability” and equating the foreseeability “instructions used in civil trials regarding proximate cause” with those appropriate “for criminal trials”). affirming that “[t]he principles of causation normally associated with civil tort litigation” apply in criminal cases as well.154State v. Garcia, 616 N.W.2d 594, 596 (Iowa 2000) (citing State v. Murray, 512 N.W.2d 547, 550 (Iowa 1994)); see State v. Melcher, 487 P.2d 3, 7–8 (Ariz. Ct. App. 1971); People v. Kemp, 310 P.2d 680, 682–83 (Cal. Dist. Ct. App. 1957); Campbell v. State, 285 So. 2d 891, 893–95 (Miss. 1973); State v. Fennewald, 339 S.W.2d 769, 772–73 (Mo. 1960); see also State v. Tyler, 873 N.W.2d 741, 750 (Iowa 2016); State v. Adams, 810 N.W.2d 365, 372 (Iowa 2012); State v. Tribble, 790 N.W.2d 121, 127 (Iowa 2010); State v. Dalton, 674 N.W.2d 111, 118–19 (Iowa 2004).

But the truth of the matter is that, as with contract and tort, foreseeability in crime is distinctive.155People v. Kibbe, 321 N.E.2d 773, 776 (N.Y. 1974) (“[T]his standard is greater than that required to serve as a basis for tort liability.”). Special foreseeability conventions arise out of the particular considerations and goals that underlie criminal law doctrine. In criminal law, the function of foreseeability requirements is to prevent unduly harsh punishment. As the Pennsylvania Supreme Court put it, a highly general foreseeability standard that “makes guilt or innocence of criminal homicide depend upon such accidental and fortuitous circumstances” may be “too harsh to be just.”156Commonwealth v. Root, 170 A.2d 310, 312 (Pa. 1961). Concern for the grave power of states’ criminal law thus powerfully shapes the foreseeability conventions of the law of crime. The Iowa Supreme Court, for example, explains that foreseeability must be “coupled with the requirement of recklessness [to] prevent the possibility of harsh or unjust results.”157McFadden, 320 N.W.2d at 613 (citation omitted).

As a result, courts have developed a narrative convention for foreseeability in criminal law that cuts off foreseeability considerably short of its comparably distant horizons in tort. A set of leading New York cases illustrate the point. In People v. Warner-Lambert Co., the defendants owned a chewing gum factory in which an explosion killed six employees when a dangerous chemical-laden dust built up in the building.158People v. Warner-Lambert Co., 414 N.E.2d 660, 661 (N.Y. 1980). Though defendants knew about the risk of explosion, they did not take immediate steps to make the machinery safer, and instead chose to await the “eventual elimination” of the dust.159Id. at 662–63. In tort, such a fatal error would lead to substantial civil liability.160See, e.g., Tropea v. Shell Oil Co., 307 F.2d 757 (2d Cir. 1962); Atl. Mut. Ins. Co. v. Clearview Club, Inc., 264 F. Supp. 608 (E.D.N.Y. 1967); Esteves v. Somco Fuel, Inc., 390 N.E.2d 1171 (N.Y. 1979); Donohue v. Walter, 548 N.Y.S.2d 435 (App. Div. 1989); Condomanolis v. Boiler Repair Maint. Co., 355 N.Y.S.2d 135 (App. Div. 1974). In the explosion case involving Warner-Lambert Company, workers’ compensation arrangements precluded at least one prominent tort suit from moving forward. See Orzechowski v. Warner-Lambert Co., 460 N.Y.S.2d 64 (App. Div. 1983). But in the criminal prosecution, the court ruled that the particular explosion that actually occurred was not reasonably foreseeable, applying a foreseeability convention for crime that was granular and discriminating where its tort analogue would have been abstract and general.161Warner-Lambert Co., 414 N.E.2d at 666.

A few years later, People v. Roth involved a worker cleaning an oil tank trailer who was killed when his trailer exploded. A stream of water from the high-pressure washer he was using struck an unprotected electric light bulb and caused a spark and consequent explosion.162People v. Roth, 604 N.E.2d 92, 93 (N.Y. 1992). In tort, a cause of action for negligence against the party providing the light for such work would have faced no substantial foreseeability obstacle. But the New York Court of Appeals ruled that the evidence was not sufficient to support criminal charges for manslaughter or criminally negligent homicide. The particular manner in which the explosion came about, the court explained, was not reasonably foreseeable.163Id. at 94.

In tort, of course, the convention is different and more general; a party need not have foreseen every specific detail to be responsible for a bad consequence.164Hill v. Winsor, 118 Mass. 251, 251 (1875). But the New York courts self-consciously adopt a convention for telling stories about foreseeability in criminal cases that narrates the relevant facts in a more granular and demanding fashion. In the criminal law, the courts assert, the injury at issue must be “the directly foreseeable consequence[]” of the defendant’s actions.165People v. Kibbe, 321 N.E.2d 773, 776 (N.Y. 1974). The relevant foreseeability, the courts note, “is greater than that required to serve as a basis for tort liability.”166Id.; see also Roth, 604 N.E.2d at 94 (“[T]he standard for criminal liability is higher than that required for civil liability . . . .”); People v. Reagan, 723 N.E.2d 55, 56 (N.Y. 1999) (affirming dismissal of an indictment of a general contracting company and its president for manslaughter and criminal negligence for deaths arising out of a trench collapse on grounds that the deaths were “an unforeseeable consequence of defendants’ conduct”).

Same linguistic formulation. Different outcome. And different narrative convention.

C. Situational Conventions (Mostly Tort)

The comparison of domains like contract, crime, and tort offers a clue to the structure and sources of foreseeability conventions. Such conventions respond to the felt moral and legal considerations that legal professionals think important for the respective domains of the law. Contract conventions protect private ordering. Crime conventions purport to guard against the overreach of the state.

Looking more closely, we can find a substantial number of such conventions tailored to recurring factual scenarios in the law. Sometimes, these are conventions for the way to tell foreseeability stories—narrative conventions. Other times, these conventions simply resolve foreseeability inquiries in the relevant domain by adopting a one-size-fits-all resolution to the foreseeability question for a given situation—per se conventions. Some of the tort principles most familiar to tort students and scholars turn out, on closer examination, to be foreseeability conventions.

1. Eggshell Skulls

One famous recurring situation is the problem of the especially vulnerable plaintiff—that grim character known as the eggshell skull plaintiff. Hornbook law states that a defendant takes the risk of the plaintiff’s vulnerabilities.167See Dulieu v. White & Sons (1901) 2 KB 669, 679 (“If a man is negligently run over or otherwise negligently injured in his body, it is no answer to the sufferer’s claim for damages that he would have suffered less injury, or no injury at all, if he had not had an unusually thin skull or an unusually weak heart.”). But why?

Jurists often contend that the eggshell plaintiff rule is a rejection of a foreseeability standard. Consider, for example, the heart-attack death of a man with a history of coronary disease to whom defendant truck driver and his employer had caused apparently minor injuries in a highway accident some six days earlier. In Benn v. Thomas, the Iowa Supreme Court allowed the decedent’s estate to recover damages for the death on the ground that the “eggshell plaintiff rule rejects the limit of foreseeability.”168Benn v. Thomas, 512 N.W.2d 537, 539 (Iowa 1994); see also Goldberg & Zipursky, supra note 124, at 464–65. So conceived, the eggshell plaintiff rule is an unattractive anomaly to the general moral proposition that liability requires foreseeability.

But there is a better way to think of the eggshell plaintiff principle. Properly understood, it is no exception. Instead, what the eggshell plaintiff rule really does is substitute a per se convention for the open-ended inquiry into foreseeability. The rule creates an affirmative finding of foreseeability for those circumstances in which the disputed injury is of the same type as an injury for which the defendant is also liable. The eggshell plaintiff rule is thus a categorical resolution of the foreseeability question, not an exception to it. The problem of unusually vulnerable victims, after all, is merely a variation of the general foreseeability problem. Is it foreseeable that the defendant’s negligent truck driving could cause bodily injury to the occupants of the car ahead of him? Yes. Is death a bodily injury? Yes. What the eggshell plaintiff rule does is instruct decisionmakers that this level of generality (rather than some more granular or specific level specifying a minor accident and a six-day delay) is the appropriate level of generality at which to tell the story of the collision between truck and car,169See Benn, 512 N.W.2d at 539–40. As the Dobbs treatise points out, the same issue arises in fire cases in certain jurisdictions. See Dobbs, Hayden & Bublick, supra note 55, § 206 & n.11. Some courts are inclined to rule that fires extending beyond a first plaintiff’s property are recoverable whether or not foreseeable. See, e.g., Smith v. London & S.W. Ry. Co. (1870) 6 LRCP 14. Others insist that such fires are best described in such a way as to deem them foreseeable. See, e.g., Milwaukee & Saint Paul Ry. Co. v. Kellogg, 94 U.S. 469 (1876). And, as we noted in the text above, still other courts insist on a convention of unforeseeability. See, e.g., Ryan v. N.Y. Cent. R.R., 35 N.Y. 210 (1866). Given the rich multifariousness of human descriptive capacities, such varying descriptions will usually be available. Our view is that all three approaches to the fire cases adopt different versions of a foreseeability convention of one sort or another. providing a ready-made conventional answer for an otherwise complex and open-ended inquiry. Drivers are thus liable when a car accident exacerbates an underlying condition.170See, e.g., Flood v. Smith, 13 A.2d 677, 678–79 (Conn. 1940). Students are liable for unusual injuries to a classmate’s leg.171See Vosburg v. Putney, 50 N.W. 403 (Wis. 1891). Grocery stores are liable when a slip and fall aggravates an underlying condition.172Jones v. Super One Foods, 774 So. 2d 200, 207–08 (La. Ct. App. 2000); Lawson v. Safeway, Inc., 878 P.2d 127, 131 (Colo. App. 1994). Rather than analyzing the facts of each case to determine if the plaintiff’s prior condition was reasonably foreseeable, the eggshell plaintiff convention provides a universally applicable rule to replace case-by-case assessments.173See generally Harper, James & Gray, supra note 105, § 20.3, at 1123–24, 1128 & n. 25; Keeton, Dobbs, Keeton & Owen, supra note 56, § 43, at 292.

Indeed, the eggshell plaintiff rule reveals a further nested convention about what counts as a similar type of injury such that the rule is implicated at all. A granular and specific account might distinguish Mr. Benn’s bruising from his heart attack as different types of injury (contusions versus internal tissue death). A more general version of the story would describe the two injuries as sharing a common type (bodily injury). An even more general typology might describe the relevant injuries as harms to Mr. Benn’s interests. All three descriptions would be true. Yet they yield different conclusions as to whether the injuries are of the same type or not. The eggshell plaintiff rule intervenes to supply a conventional answer. Damages to the body of a person are of the same type, full stop, says the rule.174See Smith v. Leech Brain & Co. (1962) 2 QB 405; Brice v. Brown [1984] 1 All ER 997; Robinson v. Post Office [1974] 1 WLR 1176; Wieland v. Cyril Lord Carpets, Ltd. [1969] 3 All ER 1006. Damages of varyingly different kinds to real property, by contrast, may or may not be of the same type.175This is why the Wagon Mound I decision could plausibly go the other way, distinguishing mucking damage to the plaintiff’s dock (recoverable) from fire damage to the same dock (unrecoverable). See Overseas Tankship (U.K.) Ltd. v. Morts Dock & Eng’g Co. [1961] AC 388, 390 (Wagon Mound I); see also Vacwell Eng’g Co. v. B.D.H. Chems., Ltd. [1969] 3 All ER 1681, 1696 (“A minor explosion involving minor damage to property and to persons was reasonably foreseeable. A violent explosion which killed one man and did over £74,000 worth of damage to a building was not.”).

Like all conventions, the eggshell plaintiff principle sweeps some dubious injuries into the ambit of foreseeability. Should a defendant really be liable for unusual or otherwise out-of-the-ordinary injuries to a plaintiff who had better reason to know about a particular vulnerability?176See Kenneth S. Abraham, Strict Liability in Negligence, 61 DePaul L. Rev. 271, 292–95 (2012). Contributory and comparative fault principles may mitigate such damages, to be sure.177See id. at 295 (“[T]he thin-skull rule operates . . . except when the plaintiff is considered contributorily negligent in failing to exercise reasonable care to protect herself against her special vulnerability.”). Should ordinary negligent drivers be insurers of a great baseball pitcher’s pitching arm? The law’s answer is yes.178Restatement (Second) of Torts § 461 cmt. b, illus. 3 (Am. L. Inst. 1965); see Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 31 cmt. b, illus. 3 (Am. L. Inst. 2010); see also Vaughn v. Nissan Motor Corp., 77 F.3d 736, 738–39 (4th Cir. 1996). The eggshell plaintiff rule thus presses at the edges of foreseeability. But the law relies on it nonetheless to resolve an entire category of preexisting condition cases that threaten otherwise to prove intractable.179See Keeton, Dobbs, Keeton & Owen, supra note 56, § 43, at 293 (“If nothing more than ‘common sense’ or a ‘rough sense of justice’ is to be relied on, the law becomes to that extent unpredictable, and at the mercy of whatever the court, or even the jury, may decide to do with it.” (footnote omitted) (quoting Palsgraf v. Long Island R.R., 162 N.E. 99, 103–04 (N.Y. 1928) (Andrews, J., dissenting))).

Strikingly, the eggshell plaintiff convention is so strong that it applies in the criminal law as well, where there are cases suggesting that “in criminal law as in tort law, the injurer takes his victim as he finds him.”180Brackett v. Peters, 11 F.3d 78, 81 (7th Cir. 1993); see Cunningham v. People, 63 N.E. 517, 525 (Ill. 1902). In such cases, courts apply a “legal presumption” of foreseeability.181Cunningham, 63 N.E. at 525. Defendants are liable for murder when assaults lead to death because of an underlying illness,182Id. when assaults lead to poor mental health and eventual death,183State v. Govan, 744 P.2d 712, 717 (Ariz. Ct. App. 1987) (explaining that defendants face criminal liability when they cause “the victim to commit suicide or lose the will to live because of extreme pain from wounds inflicted”). and when they sell toxic substances to known alcoholics.184See Commonwealth v. Feinberg, 253 A.2d 636 (Pa. 1969).

2. The Palsgraf Convention

A very different kind of foreseeability convention arises out of perhaps the most famous case in all of tort. The Palsgraf case, which we have already had occasion to note, featured two railroad guards jostling a passenger boarding a passenger train, which caused the passenger’s small newspaper package to fall to the tracks.185Palsgraf v. Long Island R.R., 162 N.E. 99, 99 (N.Y. 1928). The package contained fireworks, which exploded and caused a weighing scale some distance down the platform to tip over, striking the plaintiff and injuring her.186Id. Judge Cardozo, as is well known, ruled that the guards’ negligence toward the passenger was not sufficient to establish the railroad’s liability to the plaintiff. “[N]egligence in the air, so to speak, will not do,” he wrote, quoting Frederick Pollock’s dictum on torts.187Id. (quoting Frederick Pollock, The Law of Torts: A Treatise on the Principles of Obligations Arising from Civil Wrongs in the Common Law 455 (11th ed. 1920)). The key question, Judge Cardozo insisted, was whether the defendants’ conduct put the plaintiff in particular at foreseeable risk of harm. Ever since, jurists have focused on Judge Cardozo’s dispute with dissenting Judge Andrews, who contended that the question of whether the plaintiff was foreseeable was properly for the jury.188Cardi, The Hidden Legacy, supra note 49, at 1898. Others focus on whether Palsgraf is properly categorized as a case of breach (as Judge Friendly later hinted),189See In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708, 721 n.5 (2d Cir. 1964) (“How much ink would have been saved over the years if the Court of Appeals had reversed Mrs. Palsgraf’s judgment on the basis that there was no evidence of negligence at all.”). proximate causation (as Judge Andrews insisted), duty (as Judge Cardozo held), or perhaps a case about the nexus between breach and duty (as Professor Zipursky has suggested).190Benjamin C. Zipursky, Palsgraf, Punitive Damages, and Preemption, 125 Harv. L. Rev. 1757, 1761–67 (2012).

Looking at Palsgraf through the lens of foreseeability conventions, however, yields a new perspective on the case. What Judge Cardozo accomplished was to lay down a blueprint or instruction manual for how to describe the events in question when asking about foreseeability. Recall that the problem with foreseeability is the bewildering variety of available ways to characterize the events at issue. Judge Andrews, in dissent, does nothing to help narrow the inquiry; all he asks is whether, “by the exercise of prudent foresight, . . . the result [could have been] foreseen.”191Palsgraf, 162 N.E. at 104 (Andrews, J., dissenting) (“[W]hat the prudent would foresee, may have a bearing—may have some bearing, for the problem of proximate cause is not to be solved by any one consideration. It is all a question of expediency.”). The answer to that question, asked that way, will be whatever the decisionmaker wants it to be. Judge Cardozo’s Palsgraf opinion, by contrast, instructs the decisionmaker to ask very specifically whether the defendant’s tortious conduct gave rise to a foreseeable risk to the particular plaintiff in the case, not to a class of people of whom the plaintiff is one (passengers, for example), but to the plaintiff themself (a person many feet down the platform, far from the guard and the package).192See Michael D. Green & Ashley DiMuzio, Cardozo and the Civil Jury, 34 Touro L. Rev. 183, 195 (2018). In the Palsgraf case, this left the plaintiff, Mrs. Palsgraf, outside the scope of the opinion’s carefully framed foreseeability test. In the Kinsman Transit case, by contrast, Judge Cardozo’s Palsgraf convention brought the downstream property owners within the ambit of plaintiffs with a protected interest.193In re Kinsman Transit Co. (Kinsman Transit I), 338 F.2d 708, 721–22 (2d Cir. 1964).

The caselaw is split over whether the particular manner of a plaintiff’s harm need also be foreseeable. Judge Cardozo did not exactly say one way or the other,194Compare Palsgraf, 162 N.E. at 100 (“The risk reasonably to be perceived defines the duty to be obeyed . . . .”), with id. at 101 (“We may assume, without deciding, that negligence . . . in relation to the plaintiff[] would entail liability for any and all consequences, however novel or extraordinary.”). though Judge Friendly was determined to preserve the value of the Palsgraf convention by not also asking whether the manner of their injury was foreseeable.195See Kinsman Transit I, 338 F.2d at 723. Either way, what Palsgraf does is offer decisionmakers at least a partial map of the myriad possible paths of the foreseeability analysis.

3. Subsequent Negligence (Medical Variety)

A very different category of cases yields a second per se convention, though one that may be less familiar than the eggshell skull cases. The second per se rule of foreseeability arises in situations of subsequent negligence during medical care or transportation following a prior tortious injury. The problem of liability for injuries occasioned by subsequent negligence is a more general one, of course. Myriad forms of follow-on injuries can cause harm to victims of an initial wrong. Should initial tortfeasors be liable in damages for such harms? For the most part, the law has a predictable and open-ended response. Such damages may be recoverable from the first tortfeasor where they are foreseeable.196See Horowitz v. Fitch, 30 Cal. Rptr. 882 (Dist. Ct. App. 1963) (holding initial defendant liable for aggravating harm caused by negligent third party when such harm was a proximate result of initial defendant’s actions); Bolin v. Hartford Accident & Indem. Co., 204 So. 2d 49 (La. Ct. App. 1967) (treating subsequent harm question as a general question of proximate causation); Restatement (Second) of Torts §§ 440–442 (Am. L. Inst. 1965); Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 35 cmt. c, illus. 3 (Am. L. Inst. 2010); 22 Am. Jur. 2d Damages § 255 (2022) (“If there has been an intervention of a separate independent agency which could not be anticipated, the original negligence is not considered the proximate cause of the injury . . . .”); id. § 335 (“A plaintiff who establishes actionable negligence on the defendant’s part may not hold the defendant liable for any additional or increased damages based on an intervening act of the plaintiff or a third party that is independent of the defendant’s act or omission and results in damages distinct from the damages resulting from the defendant’s act or omission if that intervening act was neither foreseeable nor the normal consequence of the defendant’s negligence.”).

But where the subsequent negligence arises in the course of medical care or transportation, the law replaces the all-things-considered foreseeability inquiry with a per se convention. Classic cases here involve negligent ambulance driving,197See Pridham v. Cash & Carry Bldg. Ctr., Inc., 359 A.2d 193, 197–98 (N.H. 1976); Atherton v. Devine, 602 P.2d 634 (Okla. 1979). negligent operation of a medivac helicopter,198See Anaya v. Superior Court, 93 Cal. Rptr. 2d 228, 230 (Ct. App. 2000). or medical malpractice.199See Stoleson v. United States, 708 F.2d 1217, 1221 (7th Cir. 1983); Wagner v. Mittendorf, 134 N.E. 539, 540–41 (N.Y. 1922). Should initial wrongdoers be on the hook for additional damages arising out of such downstream injuries? Are such injuries foreseeable?

Tort law’s conventional answer is yes. Such damages, the law asserts categorically, are foreseeable.200See, e.g., Gallagher v. Mercy Med. Ctr., Inc., 207 A.3d 634, 640 (Md. 2019). Yet when a court asserts foreseeability in such a case, it is not typically asserting that the subsequent harms were actually foreseeable in the case in question. Jurists in such cases do not even proceed with the foreseeability analysis. Instead, the law establishes and maintains a convention for resolving subsequent negligent medical care cases. Why do the courts award damages for such cases? “[U]nskillful treatment is a result which reasonably ought to have been anticipated by the wrongdoer,” says the Oregon Supreme Court.201McDonough v. Nat’l Hosp. Ass’n, 294 P. 351, 354 (Or. 1930). There is no need for a case-by-case inquiry. “[I]n law,” subsequent malpractice injuries are simply “regarded as one of the immediate and direct” forms of recoverable damage.202Id. Or consider a recent case in the same state upholding a trial court’s jury instructions that authorized damages for foreseeable subsequent medical malpractice. An “original tortfeasor’s liability,” held the Oregon Supreme Court, is limited to “reasonably foreseeable subsequent conduct and injuries.”203Sloan ex rel. Sloan v. Providence Health Sys., 437 P.3d 1097, 1106 (Or. 2019). The category of foreseeable injuries, the court further elaborated, includes as a matter of law any injury resulting from professional medical procedures addressing the initial injuries.204Id. at 1105. As to those injuries, no foreseeability inquiry need be made, and no decision need be reached about the level of generality at which to make the inquiry. The law supplies a per se convention of foreseeability.205See Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 35 (Am. L. Inst. 2010); Restatement (Second) of Torts § 457 (Am. L. Inst. 1965); see also Martin v. Bohrer, 733 P.2d 68, 71 (Or. Ct. App. 1987) (applying “a per se rule”). Some of the cases assert a foreseeability inquiry. See, e.g., Trail v. Green, 206 F. Supp. 896, 898 (D.N.J. 1962) (“This result would occur if a physician’s negligent treatment were a foreseeable intervening cause of plaintiff’s ultimate injury.”). But such cases are a clear minority, and in any event, the open-ended foreseeability approach more often gives way to stronger per se rules. See, e.g., St. Louis & S.F. Ry. v. Doyle, 25 S.W. 461, 461 (Tex. Civ. App. 1894) (holding that the “amputation of appellee’s foot was the result of the negligence of appellant, and whether or not the surgeon made a mistake is immaterial, as appellee was not guilty of any negligence in producing such result”). Missouri, the courts are clear that “subsequent medical malpractice” is reasonably foreseeable as a matter of law. See, e.g., Mackey v. Smith, 438 S.W.3d 465, 475 (Mo. Ct. App. 2014). In New Mexico, the courts call injuries from subsequent negligent medical care “foreseeable as a matter of law.” See, e.g., Payne v. Hall, 137 P.3d 599, 604 (N.M. 2006); Bustos v. City of Clovis, 365 P.3d 67, 73 (N.M. Ct. App. 2015).

Significantly, the law of subsequent negligence in medical care and transport cases is different from the law of subsequent negligence more generally. When tortious injuries produce discrete and additional losses from the subsequent negligence of a third party, the law does not typically supply a convention. In these unusual and idiosyncratic cases, the law throws decisionmakers on the mercy of the open-ended foreseeability standard206See sources cited supra note 196.—or even adopts a per se rule of unforeseeability.207See, e.g., Armstrong v. Bergeron, 178 A.2d 293, 294 (N.H. 1962). Why? Well, for one thing, subsequent medical care cases present a special and distinctive situation for the basic goals of the law of torts. In cases where the relationship between a defendant’s tortious conduct and a plaintiff’s injuries are attenuated, courts worry about whether the plaintiff’s conduct might have contributed to their injury.208See, e.g., Lyons v. Erie Ry. Co., 57 N.Y. 489, 490–91 (1874). Insurance lawyers call this the problem of moral hazard.209See, e.g., Kenneth S. Abraham, Four Conceptions of Insurance, 161 U. Pa. L. Rev. 653, 659 (2013). See generally Tom Baker, On the Genealogy of Moral Hazard, 75 Tex. L. Rev. 237 (1996). But in subsequent medical care cases, where medical professionals are the principal decisionmakers, the plaintiff’s conduct is rarely at issue. The moral hazard problem of extending damages to plaintiffs is substantially lessened, making categorical treatment of the cases more appropriate.

Fittingly, in subsequent negligent medical care cases, an opposite and paired convention closes off liability for especially extenuated injuries. If negligent care for the victim of a tortious wrong leads to a second round of negligent medical care, causing a third round of injuries, there is some authority for a per se rule against liability for the first tortfeasor. Damages for such injuries may not be recovered from the initial wrongdoer because these damages, at this stage of third-remove from the relevant wrongdoing, are not foreseeable.210See generally Sears v. Atl. Coast Line R.R., 86 S.E. 176 (N.C. 1915). Why? Presumably because the social value of recovery against an initial tortfeasor in such rare cases is not sufficient to justify the considerable cost of such litigation. Convention strikes again.

4. Emotional Distress

The tort of negligent infliction of emotional distress features another example of foreseeability conventions. Here, the conventions have famously changed over time. A century ago, courts held that emotional distress was not foreseeable absent a physical impact and could not have been “reasonably anticipated.”211See, e.g., Mitchell v. Rochester Ry. Co., 45 N.E. 354, 355 (N.Y. 1896), overruled by Battalla v. State, 176 N.E.2d 729 (N.Y. 1961). The classic case involved a woman who was startled and nearly struck by a pair of horses and suffered a miscarriage soon thereafter.212Id. Of course, described at a certain level of generality, the woman’s injuries were entirely foreseeable. The standard problem of multiple descriptions applies. But the courts established a one-size-fits-all convention of per se unforeseeability. There could be “no recovery for fright, terror, alarm, anxiety, or distress of mind.”213Spade v. Lynn & Bos. R.R., 47 N.E. 88, 89 (Mass. 1897) (“[I]t is unreasonable to hold persons who are merely negligent bound to anticipate and guard against fright . . . .”); Comstock v. Wilson, 177 N.E. 431, 433 (N.Y. 1931) (“Serious consequences from mere mental disturbance unaccompanied by physical shock cannot be anticipated, and no person is bound to be alert to avert a danger that foresight does not disclose.”).

But as every torts student knows, the early convention of unforeseeability gave way. Courts observed that the per se convention of unforeseeability was overbroad since, after all, few doubted that negligence could, in many cases, cause severe emotional distress even absent a physical impact.214See, e.g., Orlo v. Conn. Co., 21 A.2d 402, 405 (Conn. 1941) (“There may well be situations where . . . harm from fright was not a reasonably foreseeable result of the negligence claimed. But it is equally true that in other situations it would be well within the realm of reasonable foreseeability . . . .”); Robb v. Pa. R.R., 210 A.2d 709, 712 (Del. 1965) (“[E]arly difficulty in tracing a resulting injury back through fright or nervous shock has been minimized by the advance of medical science . . . .”); Falzone v. Busch, 214 A.2d 12, 14 (N.J. 1965). Some courts tried to rescue the convention of unforeseeability by taking a narrow but important slice of the emotional distress cases—those where the plaintiff was within immediate physical danger, though not ultimately physically injured—and throwing them back into the scrap heap of the undifferentiated case-by-case foreseeability inquiry.215See Robb, 210 A.2d at 715 (“[T]he injured party is entitled to recover under an application of the prevailing principles of law as to negligence and proximate causation.”). Emotional distress plaintiffs, such courts held, could recover if they were able to show in any given case that their emotional injuries were foreseeable in a particular way.216Such was the rule in England and Australia for some time. See, e.g., Bourhill v. Young [1943] AC 92 (HL); King v. Phillips (1953) 1 QB 429; Mount Isa Mines Ltd. v Pusey (1970) 125 CLR 383, 401–02 (Austl.); Jaensch v Coffey (1984) 155 CLR 549, 595 (Austl.). But the law soon adopted new conventions for the foreseeability question in these cases of immediate danger. The most common such rule is that damages from fright in such cases are simply foreseeable as a matter of law; this is the “zone of danger” test, laid down in the aptly titled Falzone case from New Jersey.217Falzone, 214 A.2d at 17; see also Consol. Rail Corp. v. Gottshall, 512 U.S. 532, 534 (1994) (holding that negligent-infliction-of-emotional-distress defendants can be liable “to those plaintiffs who . . . are placed in immediate risk of physical impact by that negligence”); Chiuchiolo v. New England Wholesale Tailors, 150 A. 540 (N.H. 1930); Battalla v. State, 176 N.E.2d 729 (N.Y. 1961); Niederman v. Brodsky, 261 A.2d 84 (Pa. 1970). This rule was also codified in Restatement (Second) of Torts § 436 (Am. L. Inst. 1965). In the United Kingdom, the House of Lords came to a different formulation amounting to the same convention: where a plaintiff could show that the defendant could reasonably foresee that his conduct exposed the plaintiff to the risk of physical harm, the plaintiff need not also show that injury by nervous shock was reasonably foreseeable by the defendant. See Page v. Smith (1996) 1 AC 155 (HL). The Restatement (Second) of Torts, published in 1965, proposed a different and slightly narrower convention limited to “shock or fright at harm or peril to” members of the plaintiff’s immediate family, but the point was the same: certain emotional distress damages arising out of fright would be recoverable because they would be deemed foreseeable as a matter of law.218Restatement (Second) of Torts § 436(3) (Am. L. Inst. 1965) (deeming emotional distress damages foreseeable and thus recoverable when shock or fright arose out of peril to a member of the plaintiff’s immediate family in the plaintiff’s presence).

In 1968, California struck out in a new path by trying to abolish all conventions in cases of negligent infliction of emotional distress. In Dillon v. Legg, the court held that recovery for negligently inflicted emotional distress turns on what is “reasonably foreseeable,” full stop.219Dillon v. Legg, 441 P.2d 912, 919–20 (Cal. 1968). Here was the promise of adopting a completely undifferentiated, all-circumstances-considered foreseeability standard for emotional distress. Courts, Dillon announced, would proceed “on a case-to-case basis, analyzing all the circumstances” to “decide what the ordinary man under such circumstances should reasonably have foreseen.”220Id. at 921.

But even before Dillon had been announced, the open-ended foreseeability standard began to give way in the face of ordering conventions designed for decisionmakers looking for guidance in framing the appropriate level of generality in the foreseeability inquiry. The Dillon court identified three factors that bore especially closely on foreseeability: proximity in space, in time, and in familial relationship.221The three factors are as follows: (1) Whether plaintiff was located near the scene of the accident as contrasted with one who was a distance away from it. (2) Whether the shock resulted from a direct emotional impact upon plaintiff from the sensory and contemporaneous observance of the accident, as contrasted with learning of the accident from others after its occurrence. (3) Whether plaintiff and the victim were closely related, as contrasted with an absence of any relationship or the presence of only a distant relationship. Id. at 920. Thus was born a new conventional solution to foreseeability’s intractable open-endedness: the “‘relative bystander’ test,” as the U.S. Supreme Court labels it,222Gottshall, 512 U.S. at 548 (citing Dillon, 441 P.2d 912). which, in turn, has been adopted in the Restatement (Third) of Torts.223Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 48 (Am. L. Inst. 2012) (“Negligent Infliction of Emotional Harm Resulting from Bodily Harm to a Third Person”).

Today, nearly every state has recognized negligent infliction of emotional distress as a recoverable tort.224Alabama and Arkansas are apparent exceptions. See, e.g., Allen v. Walker, 569 So. 2d 350 (Ala. 1990) (finding no cause of action for negligent infliction of emotional distress”). Arkansas has a murkier stance. Compare Mechs. Lumber Co. v. Smith, 752 S.W.2d 763, 765 (Ark. 1988) (“[A] claim of negligent infliction of emotional distress is not recognized in Arkansas.”), with M.B.M Co. v. Counce, 596 S.W.2d 681, 684 (Ark. 1980) (holding that “there can be no recovery for fright or mental anguish caused by mere negligence” but allowing recovery for emotional distress without physical injury when “fright or mental anguish is caused by wilful conduct,” “committed with the intention of causing mental distress” (quoting Wilson v. Wilkins, 25 S.W.2d 428, 428 (Ark. 1930))). Despite any number of leading scholarly accounts, however, the story of the doctrinal evolution has not been a trajectory from the per se rule of unforeseeability to a case-by-case foreseeability test. Instead, the real story has been one of swapped conventions: a convention of unforeseeability has given way to a convention of foreseeability for certain recurring categories of emotional distress cases. Most states adopt the “relative bystander” convention.225See, e.g., Croft v. Wicker, 737 P.2d 789 (Alaska 1987); Thing v. La Chusa, 771 P.2d 814 (Cal. 1989); Clohessy v. Bachelor, 675 A.2d 852 (Conn. 1996); Zell v. Meek, 665 So. 2d 1048 (Fla. 1995); Leong v. Takasaki, 520 P.2d 758 (Haw. 1974); Groves v. Taylor, 729 N.E.2d 569 (Ind. 2000); Fineran v. Pickett, 465 N.W.2d 662 (Iowa 1991); Lejeune v. Rayne Branch Hosp., 556 So. 2d 559 (La. 1990); Cameron v. Pepin, 610 A.2d 279 (Me. 1992); Stockdale v. Bird & Son, Inc., 503 N.E.2d 951 (Mass. 1987); Nugent v. Bauermeister, 489 N.W.2d 148 (Mich. Ct. App. 1992); Entex, Inc. v. McGuire, 414 So. 2d 437 (Miss. 1982); Maguire v. State, 835 P.2d 755 (Mont. 1992); James v. Lieb, 375 N.W.2d 109 (Neb. 1985); Buck v. Greyhound Lines, Inc., 783 P.2d 437 (Nev. 1989); Wilder v. City of Keene, 557 A.2d 636 (N.H. 1989); Frame v. Kothari, 560 A.2d 675 (N.J. 1989); Folz v. State, 797 P.2d 246 (N.M. 1990); Johnson v. Ruark Obstetrics & Gynecology Assocs., 395 S.E.2d 85 (N.C. 1990); Paugh v. Hanks, 451 N.E.2d 759 (Ohio 1983); Sinn v. Burd, 404 A.2d 672 (Pa. 1979); Reilly v. United States, 547 A.2d 894 (R.I. 1988); Kinard v. Augusta Sash & Door Co., 336 S.E.2d 465 (S.C. 1985); Ramsey v. Beavers, 931 S.W.2d 527 (Tenn. 1996); Boyles v. Kerr, 855 S.W.2d 593 (Tex. 1993); Gain v. Carroll Mill Co., 787 P.2d 553 (Wash. 1990); Heldreth v. Marrs, 425 S.E.2d 157 (W. Va. 1992); Bowen v. Lumbermens Mut. Cas. Co., 517 N.W.2d 432 (Wis. 1994); Contreras v. Carbon Cnty. Sch. Dist. No. 1, 843 P.2d 589 (Wyo. 1992); see also Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 48 cmt. a (Am. L. Inst. 2012). At least fourteen states employ the “zone of danger” test.226See, e.g., Keck v. Jackson, 593 P.2d 668, 670 (Ariz. 1979); Towns v. Anderson, 579 P.2d 1163, 1165 (Colo. 1978); Robb v. Pa. R.R., 210 A.2d 709, 714–15 (Del. 1965); Williams v. Baker, 572 A.2d 1062, 1073 (D.C. 1990); Rickey v. Chi. Transit Auth., 457 N.E.2d 1, 5 (Ill. 1983); Resavage v. Davies, 86 A.2d 879 (Md. 1952); Engler v. Ill. Farmers Ins. Co., 706 N.W.2d 764, 770–71 (Minn. 2005); Asaro v. Cardinal Glennon Mem’l Hosp., 799 S.W.2d 595, 596 (Mo. 1990); Bovsun v. Sanperi, 461 N.E.2d 843, 848–49 (N.Y. 1984); Whetham v. Bismarck Hosp., 197 N.W.2d 678, 684 (N.D. 1972); Shelton v. Russell Pipe & Foundry Co., 570 S.W.2d 861, 864 (Tenn. 1978); Boucher v. Dixie Med. Ctr., 850 P.2d 1179, 1182 (Utah 1992); Jobin v. McQuillen, 609 A.2d 990, 993 (Vt. 1992); Garrett ex rel Kravit v. City of New Berlin, 362 N.W.2d 137, 141–42 (Wis. 1985); see also Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 48 cmt. a (Am. L. Inst. 2012). Variations on the zone of danger test appear as well. See, e.g., Nielson v. AT&T Corp., 597 N.W.2d 434, 442 (S.D. 1999) (adopting a zone of danger standard that accommodates fear for injury to a third person).

5. Economic Losses

The common law typically treats pure economic losses, unaccompanied by physical damages, as unrecoverable in negligence cases.227Dobbs, Hayden & Bublick, supra note 55, § 41.3, at 1061. Why? Because, say the canonical cases, pure economic losses are unforeseeable in law as a categorical matter. The law, in other words, adopts a convention of unforeseeability for pure economic losses.228Even where courts try to base the doctrine of pure economic loss in other concepts, the roots of the doctrine in a per se rule of unforeseeability are visible. See 532 Madison Ave. Gourmet Foods, Inc. v. Finlandia Ctr., Inc., 750 N.E.2d 1097, 1101 (N.Y. 2001) (rejecting foreseeability as a measure of duty but explaining further that courts make duty determinations “by balancing factors, including the reasonable expectations of parties and society generally” (quoting Hamilton v. Beretta U.S.A. Corp., 750 N.E.2d 1055, 1060 (N.Y. 2001))).

In recent decades a number of jurisdictions have departed from the traditional convention of per se unforeseeability. As one leading case put it, the economic loss rule’s hard-and-fast convention “capriciously showers compensation along the path of physical destruction, regardless of the status or circumstances of individual claimants.”229People Express Airlines, Inc. v. Consol. Rail Corp., 495 A.2d 107, 111 (N.J. 1985). But as in the emotional distress cases, abandonment of one convention has meant embrace of others. In some jurisdictions, this has meant a narrative foreseeability convention: namely, a foreseeability test accompanied by a script instructing courts to carry out the inquiry with a fine-grained level of specificity. New Jersey courts, for example, prescribe that the foreseeability inquiry in pure economic loss rule cases specify “particular plaintiffs or an identifiable class of plaintiffs” and “ascertainable economic damages.”230Id. at 115. The “more particular . . . the foreseeability,” on this view, the “more just” the liability for economic losses.231Id. at 116; see also Kalitta Air, L.L.C. v. Cent. Tex. Airborne Sys. Inc., 315 F. App’x 603, 605 (9th Cir. 2008); Mattingly v. Sheldon Jackson Coll., 743 P.2d 356, 358–60 (Alaska 1987); Hawthorne v. Kober Constr. Co., 640 P.2d 467, 469–70 (Mont. 1982); Aikens v. Debow, 541 S.E.2d 576, 589 (W. Va. 2000) (“[S]ome other special relationship [is required] between the alleged tortfeasor and the individual who sustains purely economic damages sufficient to compel the conclusion that the tortfeasor had a duty to the particular plaintiff and that the injury complained of was clearly foreseeable to the tortfeasor. The existence of a special relationship will be determined largely by the extent to which the particular plaintiff is affected differently from society in general. It may be evident from the defendant’s knowledge or specific reason to know of the potential consequences of the wrongdoing, the persons likely to be injured, and the damages likely to be suffered.”).

A second approach for cases departing from the traditional view is to adopt per se foreseeability conventions that are more nuanced and nested than the traditional blunderbuss rule of unforeseeability for purely economic damages. The loss of future wages for sailors and seamen is a foreseeable result of damaging the vessel on which they work.232See Carbone v. Ursich, 209 F.2d 178, 181–82 (9th Cir. 1953). This is true even when the sailors and seamen have no ownership interest in the vessel. Id. Economic damages to those whose livelihoods depend on the use of a natural resource are foreseeable (and thus recoverable) when polluters destroy it.233See Exxon Shipping Co. v. Baker, 554 U.S. 471 (2008); see also In re Exxon Valdez, 104 F.3d 1196, 1197–98 (9th Cir. 1997); Union Oil Co. v. Oppen, 501 F.2d 558, 570 (9th Cir. 1974); Leo v. Gen. Elec. Co., 145 A.D.2d 291, 294 (N.Y. App. Div. 1989); Louisiana ex rel. Guste v. M/V Testbank, 752 F.2d 1019, 1021 (5th Cir. 1985). Purely economic losses to third parties are foreseeable (and thus recoverable) when financial auditors fail to live up to their professional obligations.234See Billings Clinic v. Peat Marwick Main & Co., 797 P.2d 899, 901 (Mont. 1990); Rino v. Mead, 55 P.3d 13, 19 (Wyo. 2002). Likewise, professionals such as engineers,235See City of Eveleth v. Ruble, 225 N.W.2d 521, 527–28 (Minn. 1974); Bales for Food, Inc. v. Poole, 424 P.2d 892, 893 (Or. 1967). lawyers,236See Reilly v. Cavanaugh, 29 Ind. 435, 436 (1868); French v. Armstrong, 76 A. 336, 337–38 (N.J. 1910); Glenn v. Haynes, 66 S.E.2d 509, 512–13 (Va. 1951); Stephens v. White, 2 Va. 203, 211 (1796); Shoemake ex rel. Guardian v. Ferrer, 225 P.3d 990, 994 (Wash. 2010). public notaries,237See Smith v. Maginnis, 89 S.W. 91 (Ark. 1905); McDonald v. Plumb, 90 Cal. Rptr. 822, 824–26 (Ct. App. 1970); Peters v. Hyatt Legal Servs., 440 S.E.2d 222, 227 (Ga. Ct. App. 1993); Hope v. Victor, 162 N.W.2d 918, 920 (Mich. Ct. App. 1968). weighers,238Glanzer v. Shepard, 135 N.E. 275, 276 (N.Y. 1922). architects,239Paver & Wildfoerster v. Cath. High Sch. Ass’n, 345 N.E.2d 565, 567–68 (N.Y. 1976); County of Milwaukee v. Schmidt, Garden & Erikson, 168 N.W.2d 559, 563 (Wis. 1969). and telegraph operators240Russ v. W. Union Tel. Co., 23 S.E.2d 681, 682 (N.C. 1943). are liable in negligence to third parties for economic losses, at least in ordinary cases, because such losses are, in law, sufficiently foreseeable. Suicide by insanity is also a foreseeable result of injurious negligent conduct.241Knobe & Shapiro, supra note 18, at 220.

Whence do such relational conventions of foreseeability arise? As the New Jersey high court observes, categorical rules of foreseeability in the area of economic losses express courts’ confidence that plaintiffs so described are “particularly foreseeable” and thus eligible to recover.242See People Express Airlines, Inc. v. Consol. Rail Corp., 495 A.2d 107, 115–16 (N.J. 1985). When courts invoke one or another of these categorical exceptions to the rule of unforeseeability, they go one step further than the script of the granular foreseeability inquiry. They adopt a per se rule of foreseeability for the category as a whole.

6. Planes and Trains, but Not Automobiles

Common carriers such as trains, buses, airplanes, and hotels have a duty “to exercise the highest degree of care that human judgment and foresight are capable of, to make [their] passenger’s journey safe.”243William Eggleston, Eggleston on Damages: A Treatise on the Law of Damages § 53, at 46 (1880). One of the things the rule has meant in practice is that courts in common carrier cases are to describe the relevant events at a high level of abstraction such that even unusual or otherwise relatively remote events are more likely to appear as foreseeable. Thus, commentators routinely assume that common carriers’ foresight obligations for tort liability purposes are especially extensive.244See Loren Page Ambinder, Note, Dispelling the Myth of Rationality: Racial Discrimination in Taxicab Service and the Efficacy of Litigation Under 42 U.S.C. § 1981, 64 Geo. Wash. L. Rev. 342, 362–63 (1996); Janet E. Assimotos, Comment, To Warn or Not to Warn?: The Airlines’ Duty to Disclose Terrorist Threats to Passengers, 56 J. Air L. & Com. 1095, 1105–06 (1991); Catherine Stone Bowe, Comment, “May I Offer You Something to Drink from the Beverage Cart?”: A Close Look at the Potential Liability for Airlines Serving Alcohol, 54 J. Air L. & Com. 1013, 1035 (1989); Carolyn Ritchie, Comment, Potential Liability from Electromagnetic Interference with Aircraft Systems Caused by Passengers’ On-Board Use of Portable Electronic Devices, 61 J. Air L. & Com. 683, 720 (1996). Common carriers, for example, typically have more extensive obligations with respect to care and foresight in hiring than do other employers.245Burch v. A & G Assocs., Inc., 333 N.W.2d 140, 144 (Mich. Ct. App. 1983); see Johnson v. Detroit, Y. & A.A. Ry., 90 N.W. 274, 274 (Mich. 1902). The law has also long held that the foresight required of common carrier defendants extends to injuries arising out of assaults and criminal attacks on passengers by third parties.246United Rys. & Elec. Co. of Balt. v. State ex rel. Deane, 49 A. 923 (Md. 1901); see Banks v. Hyatt Corp., 722 F.2d 214 (5th Cir. 1984); Lopez v. S. Cal. Rapid Transit Dist., 710 P.2d 907, 909–10 (Cal. 1985); Todd v. Mass Transit Admin., 816 A.2d 930, 935 (Md. 2003); Pittsburgh, Fort Wayne & Chi. Ry. Co. v. Hinds, 53 Pa. 512, 516–17 (1866). But see R.M. ex rel. M.W. v. Am. Airlines, Inc., 338 F. Supp. 3d 1203, 1215–16 (D. Or. 2018).

The logic of the common carriers convention implicitly conveys the difficulty the law would have trying to craft similar conventions for, say, automobiles. The common carrier cases present a recurring institutional relationship with a recurring set of facts. Common carriers are typically in positions to effectively manage risks arising in conjunction with their operations, especially compared to their customers. Treating such risks as foreseeable risks of being a common carrier, rather than as the foreseeable risks of being a customer or passenger, makes sense in a way that treating vehicle risks as a risk of driving green cars rather than silver ones would not—i.e., nothing in the relationship between cars of different colors offers a toehold for a sensible foreseeability convention. In the law, such instances can only be left to case-by-case analysis, at least if foreseeability is the goal. Of course, such inquiries often will not be worthwhile to any of the parties involved. It should hardly be surprising, then, that the ad hoc automobile cases are the area most notorious for a settlement system that relies on a set of private conventions. When the public system fails to produce per se conventions, private parties—insurers’ and plaintiffs’ lawyers—fill the gap.247See Nora Freeman Engstrom, An Alternative Explanation for No-Fault’s “Demise”, 61 DePaul L. Rev. 303 (2012); see also Nathaniel Donahue & John Fabian Witt, Tort as Private Administration, 105 Cornell L. Rev. 1093 (2020).

7. Special Relationships

In addition to having an obligation to exercise all possible care and foresight, common carriers have a further duty to exercise reasonable care in protecting customers from the actions of third parties. The law calls this a “special relationship,”248Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 40 & cmt. g (Am. L. Inst. 2012). and the special relationship is a window into the way in which courts have taken important recurring categories of social relations and pulled them out of the foreseeability inquiry, deeming them foreseeable as a matter of law. The result is an entire suite of categorical special relations with duties to protect.

The logic of the doctrine proceeds as follows: One person generally has no duty to protect another by exercising control over the conduct of a third party.249Restatement (Second) of Torts § 315 (Am. L. Inst. 1965); Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 37 (Am. L. Inst. 2012). But such a duty arises in certain situations, courts say. The test for a duty to control rests on a handful of factors, of which the foreseeability of injury in question is arguably the most prominent.250Burton v. E.I. du Pont de Nemours & Co., 994 F.3d 791, 820 (7th Cir. 2021) (“In Wisconsin, ‘one has a duty to exercise ordinary care under the circumstances.’ ‘Ordinary care involves the concept of foreseeability, in that a reasonable person exercising ordinary care would have foreseen injury as a consequence of his act.’” (citation omitted) (quoting Hoida, Inc. v. M & I Midstate Bank, 717 N.W.2d 17, 28–29 (Wis. 2006))); Glascock v. City Nat’l Bank of W. Va., 576 S.E.2d 540, 544 (W. Va. 2002) (“The ultimate test of the existence of a duty to use care is found in the foreseeability that harm may result if it is not exercised.”); Sinclair Wyo. Refin. Co. v. A & B Builders, Ltd., 989 F.3d 747, 793 (10th Cir. 2021) (“Under Wyoming law, one owes a duty of care to those in the zone of foreseeable risk.”). The cases come out in different and often contradictory ways,251For examples denying a special relationship between parties, see Brown v. Balt. & Ohio R.R., 805 F.2d 1133 (4th Cir. 1986) (discussing railroad employee injured by construction materials left on tracks by unknown third party); Levrie v. Dep’t of Army, 810 F.2d 1311 (5th Cir. 1987) (discussing army and contractors injured by hazardous chemicals on property); Glade ex rel. Lundskow v. United States, 692 F.3d 718 (7th Cir. 2012) (discussing Veterans Administration and patient battered by employee during outpatient visits); Figueroa v. Evangelical Covenant Church, 879 F.2d 1427 (7th Cir. 1989) (discussing owner of parking lot and individual sexually assaulted in parking lot); Freeman v. Busch, 349 F.3d 582 (8th Cir. 2003) (discussing resident advisor and student sexually assaulted in dorm); Turbe v. Gov’t of V.I., 938 F.2d 427 (3d Cir. 1991) (discussing power authority and pedestrian assaulted near broken street lights); First Nat’l Bank v. United States, 829 F.2d 697 (8th Cir. 1987) (discussing probation officer and crime victim). For examples affirming a special relationship between parties, see Miles v. Melrose, 882 F.2d 976 (5th Cir. 1989) (discussing crew members and union aware of worker’s violent nature); Novak v. Cap. Mgmt. & Dev. Corp., 452 F.3d 902 (D.C. Cir. 2006) (discussing night club and criminal attack on patrons); Panion v. United States, 385 F. Supp. 2d 1071 (D. Haw. 2005) (discussing United States and patient assaulted while sedated at Army hospital); Nova Se. Univ., Inc. v. Gross, 758 So. 2d 86 (Fla. 2000) (discussing university and student assaulted at mandatory off-campus internship); Shurben v. Dollar Rent-a-Car, 676 So. 2d 467 (Fla. Dist. Ct. App. 1996) (discussing car rental agency and attack on tourist); Mulloy v. United States, 937 F. Supp. 1001 (D. Mass. 1996) (discussing Army and death of servicemember’s wife). but there is nevertheless a striking conventionality to them. Once a court identifies a particular social relation as characterized by the appropriate foreseeability (or not), that relation becomes a categorical special relation (or not) for which future cases need not repeat the foreseeability inquiry. Poof! Mirabile dictu! A foreseeability convention is made.252The Tennessee courts offer an especially vivid sequence. Tennessee law generally states that individual A has a duty to control the conduct of individual B when the risk of B’s conduct is foreseeable. In practice, each case in the area has come to stand for a new standard superseding the general foreseeability inquiry. Compare Bradshaw v. Daniel, 854 S.W.2d 865, 871–73 (Tenn. 1993) (holding categorically that physicians owe a duty of reasonable care to nonpatients to warn of foreseeable risks of exposure to the source of a patient’s noncontagious disease), with Turner v. Jordan, 957 S.W.2d 815, 819–21 (Tenn. 1997) (holding categorically that psychiatrists owe a duty of reasonable care to protect nurses when a psychiatrist knows or ought to know that a patient poses an unreasonable risk of harm to a nurse), and Biscan v. Brown, 160 S.W.3d 462, 479–82 (Tenn. 2005) (holding that adult social hosts owe a duty of reasonable care to protect third parties from the dangerous conduct of minors under a host’s care).

Consider the case of Otis Engineering Corp. v. Clark, in which the Supreme Court of Texas asked whether an employer had a duty to exercise reasonable care in controlling a drunk employee who was sent home in his own car and shortly thereafter struck and killed two people.253Otis Eng’g Corp. v. Clark, 668 S.W.2d 307, 309 (Tex. 1984). The court’s analysis was not one-off or case-specific; instead, it was categorical. Deciding that the employer’s conduct “created a foreseeable risk,” the court ruled that “when, because of an employee’s incapacity, an employer exercises control over the employee, the employer has a duty to take such action as a reasonably prudent employer under the same or similar circumstances would take to prevent the employee from causing an unreasonable risk of harm to others.”254Id. at 311.

The foreseeability inquiry in Otis Engineering could have yielded a one-off determination for this particular drunk employee case to be repeated in a similarly fact-specific manner the next time such a case arose. Instead, Otis Engineering occasioned a new precedent for like cases in its class. To seal the point, the court explained that its affirmative-duty conclusion was the rule “that we now adopt for this and all other cases currently in the judicial process.”255Id. The Otis Engineering precedent thus established a convention of foreseeability for a class of cases—a convention that, in turn, has produced further categorical conventions in its wake.256See Peavy v. Tex. Home Mgmt., Inc., 7 S.W.3d 795 (Tex. Ct. App. 1999) (holding defendant mental health management company has duty to control patient); Tex. Dep’t of Mental Health & Mental Retardation v. McClain, 947 S.W.2d 694 (Tex. Ct. App. 1997) (holding state mental hospital has duty to control patient); Golden Spread Council, Inc. v. Akins, 926 S.W.2d 287 (Tex. 1996) (holding local Boy Scout council has duty to control scoutmaster).

The Texas caselaw is hardly an outlier. Time and again, state high courts pull particular social relations out of the undifferentiated foreseeability inquiry and deem them risks sufficiently foreseeable (or not) to make such relations special as a matter of law for purposes of a duty to protect.257See, e.g., Janis v. Pratt & Whitney Can., Inc., 370 F. Supp. 2d 1226, 1230 (M.D. Fla. 2005) (“The recognized special relationships continue to evolve. In fact, the Third District Court of Appeals extended the special relationship doctrine to rental agency-customer . . . .” (citation omitted)).

8. Rescuers

Arguably, the pattern of trying to rescue cases from the bottomless pit of the open-ended foreseeability standard began particularly fittingly—with a famous case of an actual rescue attempt. In Wagner v. International Railway Co.,258133 N.E. 437 (N.Y. 1921). Judge Cardozo upheld a cause of action by a rescuer on the grounds that, as the Dobbs treatise puts it, the rescuer in the case “was within the scope of the foreseeable risk.”259Dobbs, Hayden & Bublick, supra note 55, § 202; see Wagner, 133 N.E. 437. But Judge Cardozo went further. He announced a rule for the class of cases involving rescuers. “Danger,” he wrote, “invites rescue.”260Wagner, 133 N.E. at 437.

As courts have received the rule, “the rescue doctrine embodies a policy choice by courts to deem rescue attempts to be foreseeable for purposes of tort recovery.”261Gray v. Russell, 853 S.W.2d 928, 931 (Mo. 1993). “[I]t is foreseeable,” asserts the Washington Supreme Court flatly, that “a rescuer will come to the aid of the person imperiled by the tortfeasor’s actions.”262McCoy v. Am. Suzuki Motor Corp., 961 P.2d 952, 956 (Wash. 1998). This is so even in improbable cases. As one tort jurist put it, “rescuers of life are always treated as foreseeable, however unlikely they may be, [even though] rescuers of property are treated as intervening actors, however common they may be.”263Arthur Ripstein, In Extremis, 2 Ohio St. J. Crim. L. 415, 431 (2005) (emphasis added).

The same rule holds in the United Kingdom, where the High Court of Justice in 1967 in the Lewisham railway disaster case held that helpers at the scene of an accident are foreseeable plaintiffs for purposes of emotional distress damages.264Chadwick v. British Transp. Comm’n [1967] 1 WLR 912 (QB) at 952 (“It could also be foreseen that somebody might try and rescue passengers and suffer injury in the process . . . .”). In 1991, in his opinion on the great Hillsborough football tragedy in which ninety-five spectators were killed by crushing, Lord Hoffmann objected to the categorical treatment of rescuers and advocated for the case-by-case application of foreseeability principles. See White v. Chief Constable of S. Yorkshire Police (1999) 2 AC (HL) 455 (Lord Hoffmann) (appeal taken from Eng.). “Cardozo,” wrote one American judge, “believed that rescuers always should be regarded as foreseeable plaintiffs.”265Herman v. Welland Chem., Ltd., 580 F. Supp. 823, 826 (M.D. Pa. 1984).

D. Situational Conventions (Contract)

So far, the situational foreseeability conventions we describe here have arisen mostly out of tort. But tort has no monopoly on foreseeability conventions. Contract deploys them, too, and though we do not list many here, it may be useful to note a few.

Recall that Hadley266Hadley v. Baxendale (1854) 156 Eng. Rep. 145. is, on our account, a case establishing a storytelling convention.267See supra text accompanying notes 131–32. It instructs that foreseeability inquiries in a contract are to be carried out at a granular level of specificity. The storytelling convention, as we explained above, embodies a principle of private ordering: it aims to induce the parties to contracts to flush out the relevant information in the contracting process.268See supra text accompanying notes 134–45. The cases in contract nonetheless also yield per se conventions that provide hard-and-fast answers to foreseeability questions in certain special cases.

1. Distress Redux

Consider, for example, that emotional distress damages are generally deemed unforeseeable in a contract and thus not recoverable in an action for breach.269See, e.g., Ruiz de Molina v. Merritt & Furman Ins. Agency, 207 F.3d 1351, 1361 (11th Cir. 2000) (“The rule in Alabama remains, however, that recovery of mental anguish damages is permitted for breach of contract only in a narrow range of cases involving contracts which create especially sensitive duties, the breach of which cause highly foreseeable and significant mental anguish.”); Wilcox v. Richmond & D.R. Co., 52 F. 264, 266 (4th Cir. 1892); Brossia v. Rick Constr., L.T.D., 81 P.3d 1126, 1131 (Colo. App. 2003); Gregory & Swapp, PLLC v. Kranendonk, 424 P.3d 897, 906 (Utah 2018); see also Restatement (Second) of Contracts § 353 (Am. L. Inst. 1981) (“Recovery for emotional disturbance will be excluded unless the breach also caused bodily harm or the contract or the breach is of such a kind that serious emotional disturbance was a particularly likely result.”); 11 Arthur Linton Corbin & Joseph M. Perillo, Corbin on Contracts § 59.1 (1979) (explaining that emotional distress is unforeseeable in contract cases). The emotional distress rule is a convention of unforeseeability, one that announces a default rule of per se unforeseeability for contract breaches.270See Young v. U.S. Dep’t of Just., 882 F.2d 633, 641 (2d Cir. 1989). The apparent aim is to advance the same project of private ordering that the underlying specificity convention for contracts advances.

2. Wedding Dresses and Marriage Promises

Yet contract has conventions of foreseeability, too. Courts resolving contracts arising out of certain highly personal arrangements, in particular, have removed whole classes of cases from the usual foreseeability convention in contracts and instead generated a set of hard-and-fast rules of foreseeability. In such cases, where breach “is particularly likely to cause serious emotional disturbance,”271Restatement (Second) of Contracts § 353 cmt. a (Am. L. Inst. 1981). the courts deem such damages foreseeable. Thus, when a bride-to-be contracts for a wedding dress, her emotional distress is foreseeable when the tailor fails to produce the finery in time.272Lewis v. Holmes, 34 So. 66, 68 (La. 1903); see Mitchell v. Shreveport Laundries, Inc., 61 So. 2d 539 (La. Ct. App. 1952) (allowing mental distress damages against laundry that failed to deliver wedding suit as agreed to groom-to-be); see also Griffin-Amiel v. Frank Terris Orchestras, 677 N.Y.S.2d 908, 912 (City Ct. 1998) (allowing disappointment and humiliation damages against a defendant who breached contract by failing to provide wedding singer); Deitsch v. Music Co., 453 N.E.2d 1302, 1304 (Ohio Mun. Ct. 1983) (allowing mental anguish damages against wedding band that failed to appear on the contracted wedding day). But see Seidenbach’s, Inc. v. Williams, 361 P.2d 185, 187–88 (Okla. 1961) (rejecting emotional distress damages for contract breach where defendant failed to deliver wedding dress in time for plaintiff’s wedding). Breach of contract to marry also produces foreseeable emotional distress.273Vanderpool v. Richardson, 17 N.W. 936 (Mich. 1883); see 1 J.G. Sutherland, A Treatise on the Law of Damages 156–67 (1882) (“Contracts are not often made for a purpose, the defeating or impairing of which can, in a legal sense, inflict a direct and natural injury to the feelings of the injured party. A breach of promise of marriage is an instance of such a contract, and such considerations enter into the estimate of the damages.”).

3. Dead Bodies

So do breaches of contracts for burial vaults,274Lamm v. Shingleton, 55 S.E.2d 810, 812–13 (N.C. 1949). funeral services,275Flores v. Baca, 871 P.2d 962, 970 (N.M. 1994). or failure to perform on a contract to deliver a death notice.276Russ v. W. Union Tel. Co., 23 S.E.2d 681, 682–83 (N.C. 1943); Lamm, 55 S.E.2d at 813–14. Just as in tort, the law of contracts extracts such recurring situations out of the morass of the undifferentiated foreseeability inquiry and supplies not only guidelines for the level of generality but a kind of per se convention of foreseeability, offering a certain predictability of administration and attending to the asymmetries between naïve and distracted consumers and repeat-play participants in the trades.

E. Origins, Sources, and Functions

Foreseeability conventions are the kudzu of the law; once one thinks to look for them, they appear virtually everywhere. In New York, it is unforeseeable as a matter of law that fires will spread,277Ryan v. N.Y. Cent. R.R., 35 N.Y. 210 (1866); Dobbs, Hayden & Bublick, supra note 55, § 206 (“New York courts developed a unique rule that permitted recovery only by the first person to whose property the fire spread.”); 59 N.Y. Jur. 2d Explosives and Fires § 70 (2022) (“[T]he present rule appears to be that . . . if the fire spreads from defendant’s land to plaintiff’s building across intervening land, and if the fire is fed by buildings, materials, or other objects upon the intervening land, there can be no recovery.”); see Homac Corp. v. Sun Oil Co., 180 N.E. 172 (N.Y. 1932). though elsewhere, it is foreseeable as a matter of law that they will.278See, e.g., Milwaukee & Saint Paul Ry. Co. v. Kellogg, 94 U.S. 469 (1876). Both positions are conventional in our sense. Damage to cargo moved onto a vessel after the vessel had already been damaged by the defendant’s negligent act is not foreseeable to the defendant.279Sinram v. Pa. R.R., 61 F.2d 767 (2d Cir. 1932). So-called “brand tarnishment,” when a manufacturer’s negligence damages the value of customers’ products is not foreseeable in law,280See In re Gen. Motors LLC Ignition Switch Litig., No. 14-MD-2543, 2016 WL 874778 (S.D.N.Y Mar. 3, 2016). nor is fear of cancer arising out of negligent exposure to carcinogens, at least absent some carcinogen-triggered illness.281Metro-N. Commuter R.R. v. Buckley, 521 U.S. 424 (1997); Norfolk & W. Ry. Co. v. Ayers, 538 U.S. 135 (2003). A person with venereal disease may be liable to the spouse of someone with whom they engage in sexual contact because courts have established as a matter of law that it is foreseeable that a sexual partner will have sex with a spouse.282Carsanaro v. Colvin, 716 S.E.2d 40, 48 (N.C. Ct. App. 2011); Mussivand v. David, 544 N.E.2d 265, 266 (Ohio 1989). The rule is not surprising or wrong; the point here is only that it is categorical, not contextual, to the nearly infinite range of marriage arrangements. Injuries arising out of the criminal acts of third parties were once treated as unforeseeable under the “superseding cause” doctrine—a per se convention of unforeseeability.283See, e.g., Loftus v. Dehail, 65 P. 379 (Cal. 1901); Miller v. Bahmmuller, 108 N.Y.S. 924 (App. Div. 1908). In principle, of course, such injuries could be characterized as foreseeable, especially when the defendant’s conduct was negligent precisely because it risked such third-party criminal acts. The doctrine of superseding criminal acts has thus largely given way to case-by-case inquiries. In certain contexts, such as apartment building owners and common carriers, injuries arising out of the criminal acts of third parties have even begun to be treated under a per se convention of foreseeability.284See, e.g., Nallan v. Helmsley-Spear, Inc., 407 N.E.2d 451 (N.Y. 1980); Kline v. 1500 Mass. Ave. Apartment Corp., 439 F.2d 477 (D.C. Cir. 1970).

In police chases where a fleeing suspect damages or injures a third party, some states apply a convention that law enforcement pursuit is not a proximate cause of the injury absent “reckless disregard” by the relevant officer or officers.285Ga. Code Ann. § 40-6-6(d)(2) (West 2022); Strength v. Lovett, 714 S.E.2d 723, 728–29 (Ga. Ct. App. 2011); see Stanley v. City of Independence, 995 S.W.2d 485, 488 (Mo. 1999). Other jurisdictions follow the pattern of the emotional distress tort and adopt what one court calls the “modern trend” toward leaving such judgments to case-by-case evaluation.286Montgomery v. Saleh, 466 P.3d 902, 911 (Kan. 2020); see, e.g., Argabrite v. Neer, 75 N.E.3d 161, 164–65 (Ohio 2016). In products liability cases, New York adopts a convention effectively deeming injuries arising out of substantial post-sale modifications to the product to be unforeseeable as a matter of law,287Robinson v. Reed-Prentice Div. of Package Mach. Co., 403 N.E.2d 440, 443–44 (N.Y. 1980). though the convention applies only to product design cases and not duty to warn cases.288Liriano v. Hobart Corp., 700 N.E.2d 303, 307–08 (N.Y. 1998).

Even the iconic rules of landowner-occupier liability come into new light once one thinks in conventional terms. Why are unlicensed entrants onto land typically unable to recover for injuries caused by the occupier of the relevant property?289See, e.g., United Zinc & Chem. Co. v. Britt, 258 U.S. 268 (1922). Courts treat the trespasser rule as a substitute for a foreseeability inquiry.290Mayer ex rel. Mayer v. Willowbrook Plaza Ltd. P’ship, 278 S.W.3d 901, 915 (Tex. Ct. App. 2009) (“[F]oreseeability of entry is not the standard for determining whether an individual is classified as an invitee.”). Other courts hold that trespassers are categorically unforeseeable because the common law classifications serve as embodiments of foreseeability and unforeseeability.291See Mellon Mortg. Co. v. Holder, 5 S.W.3d 654, 666 (Tex. 1999) (O’Neill, J., dissenting) (“[T]he concept of foreseeability in the context of premises liability is embodied in the classifications that have defined a landowner’s duty for over one hundred years.”). Either way, the point is the same: trespasser status stands in for foreseeability. When courts cut back on landowners’ immunity to trespassers; moreover, they typically do so in cases in which the convention of unforeseeablity is stretched to the breaking point.292See, e.g., Choate v. Ind. Harbor Belt R.R., 980 N.E.2d 58, 65 (Ill. 2012) (recognizing “the foreseeability of harm to children as the cornerstone of liability”); see also Kahn v. James Burton Co., 126 N.E.2d 836, 842 (Ill. 1955) (“[T]he true basis of liability [is] the foreseeability of harm to the child.”).

Foreseeability conventions of the per se variety function much like the hard-and-fast categorical rules that the law deploys in a variety of other knotty doctrinal spaces. Consider Justice Holmes’s effort to assert a stop-look-and-listen rule for breach of ordinary care at railroad grade crossings—a rule he hoped might substitute for the “featureless generality” of the reasonableness inquiry on which the law of torts otherwise relied.293See Holmes, supra note 47, at 111; see also Balt. & Ohio R.R. v. Goodman, 275 U.S. 66 (1927). Justice Cardozo famously resisted, insisting that breach determinations turn on too many circumstances and considerations to suit one-size-fits-all rules.294Pokora v. Wabash Ry. Co., 292 U.S. 98 (1934). Foreseeability inquiries pose the same problem Justice Holmes famously identified in the railroad grade crossing cases. Recurring foreseeability cases may be left to case-by-case evaluation: either with or without a narrative convention as a guideline or decided categorically by reference to a per se convention. Interestingly, on the foreseeability side, then-Judge Cardozo did not resist the same move. “Danger invites rescue,” he ruled.295Wagner v. Int’l Ry. Co., 133 N.E. 437, 437 (N.Y. 1921). And sometimes it surely does.

Per se foreseeability conventions in Justice Holmes’s sense are the congealed judicial labor-time of previous inquiries into foreseeability in particular settings.296Cf. 1 Karl Marx, Capital: A Critique of Political Economy 130 (Ben Fowkes trans., 1990) (“[C]ommodities are merely definite quantities of congealed labour-time.”). Cases on vulnerable plaintiffs, emotional distress, economic loss, special relationships, common carriers, rescuers, wedding dresses, and more—even cases on railroad grade crossings, as in Justice Holmes’s day—serve as off-the-shelf modules for otherwise dumbfounding foreseeability questions. The work of previous inquiries supplies ready-made solutions for classes of cases moving forward. They are tort’s answer to what Carol Rose famously called the “crystals and mud” of the law of property, or what Duncan Kennedy styled as the “form and substance” of contract law: the sharp rules that compete with the law’s all-things-considered standards.297See Carol M. Rose, Crystals and Mud in Property Law, 40 Stan. L. Rev. 577 (1988); Duncan Kennedy, Form and Substance in Private Law Adjudication, 89 Harv. L. Rev. 1685 (1976). Judges and juries rely on their ease of administration (no more intractable inquiries with undefined levels of generality). Actors in the world will sometimes prefer conventions, too, favoring the guidance provided by their definite contours to the open-ended generality of foreseeability standards. For both legal decisionmakers and for people and institutions trying to shape their own affairs, foreseeability conventions economize on the information costs that would otherwise arise out of needing to study the relevant situations in detail. In this sense, they are like the real property rules that do the same for estates in land.298See Thomas W. Merrill & Henry E. Smith, Optimal Standardization in the Law of Property: The Numerus Clausus Principle, 110 Yale L.J. 1 (2000). And they offer an alternative to the ineradicable uncertainty of unconstrained foreseeability standards.

Conventions of foreseeability arise for some of the same reasons that underwrite categorical treatments in all sorts of areas. Sometimes, as Sandra Sperino urges, statutes adopt concrete limits on liability in place of “statutory proximate cause” inquiries.299Sandra F. Sperino, Statutory Proximate Cause, 88 Notre Dame L. Rev. 1199, 1245–47 (2013) (“[I]t makes more sense to develop a limit on liability that comports with the underlying statutory regime.”). At other times, some feature of a set of cases allows for a shortcut or proxy that gets a high enough percentage of the cases right—or right enough, according to the values the law cares about most. Either way, in a statute or sequence of common law cases, bespoke inquiries into an intractable question like foreseeability are often too costly and difficult.300See Frederick Schauer, Profiles, Probabilities, and Stereotypes (2003); Issacharoff & Witt, supra note 59. In this respect, foreseeability conventions function as what Daryl Levinson calls “decisionmaking institutions,” which “bundle” outcomes in a legal technology that accomplishes the law’s goals—or at least come close enough.301Daryl J. Levinson, Parchment and Politics: The Positive Puzzle of Constitutional Commitment, 124 Harv. L. Rev. 657, 693–94 (2011). Storytelling conventions, in particular, resemble what Parchomovsky and Stein call the “catalog” approach302Gideon Parchomovsky & Alex Stein, Catalogs, 115 Colum. L. Rev. 165 (2015).—and what Chief Justice John Roberts and Justice Clarence Thomas (following tort jurist Thomas Atkins Street) have dubbed the “illustration” approach—to legal norms: resolutions that offer decisionmakers more guidance than the unbounded foreseeability inquiry provides, but also more flexibility than the rigidity that the per se rules allow.303CSX Transp., Inc. v. McBride, 564 U.S. 685 (2011) (Roberts, C.J., dissenting) (“The cases do not provide a mechanical or uniform test and have been criticized for that. But they do ‘furnish illustrations of situations which judicious men upon careful consideration have adjudged to be on one side of the line or the other.’” (quoting Exxon Co., U.S.A. v. Sofec, Inc., 517 U.S. 830, 839 (1996))); 1 Thomas Atkins Street, A Presentation of the Theory and Development of the Common Law: Theory and Principles of Tort 110 (1906); see also Keeton, Dobbs, Keeton & Owen, supra note 56, § 67, at 474–75.

Strikingly, the judicial administration of tort and other doctrinal domains through foreseeability conventions parallels the basic approach adopted in the private administration of tort through settlement practices and insurance arrangements.304See Donahue & Witt, supra note 247; Nora Freeman Engstrom, Sunlight and Settlement Mills, 86 N.Y.U. L. Rev. 805 (2011); Dana A. Remus & Adam S. Zimmerman, The Corporate Settlement Mill, 101 Va. L. Rev. 129 (2015); Dana A. Remus & Adam S. Zimmerman, Complex Litigation and Disaggregative Mechanisms, 63 Emory L.J. 1317 (2014); Tom Baker, Blood Money, New Money, and the Moral Economy of Tort Law in Action, 35 Law & Soc’y Rev. 275 (2001); Kathryn Zeiler, Charles Silver, Bernard Black, David Hyman & William Sage, Physicians’ Insurance Limits and Malpractice Payments: Evidence from Texas Closed Claims, 1990–2003, 36 J. Legal Stud. 59 (2007). The literature on the private administration of tort via settlement and insurance emphasizes the substitution of proxies and privately established, rough-hewn rules for the fine and laborious inquiries of the doctrine itself.305H. Laurence Ross, Settled Out of Court: The Social Process of Insurance Claims Adjustment (2d ed. 1980); Donahue & Witt, supra note 247. Consider, for example, that insurers typically reproduce the economic loss rule of tort—one of the foreseeability conventions in the doctrine—by requiring some kind of physical damage to a business as a condition of recovering on the policy. See, e.g., Peerless Dyeing Co. v. Indus. Risk Insurers, 573 A.2d 541 (Pa. Super. Ct. 1990). The profusion of foreseeability conventions is one way in which the doctrine adopts some of the strategies characteristic of private administration.

Foreseeability conventions come in more varieties than we have listed here. For example, we have focused on conventions made visible in the doctrine, largely because they are easier to document. But there is every reason to think that cultural conventions about foreseeability and appropriate levels of generality also powerfully shape the analysis in open-ended foreseeability inquiries. Such norms of foreseeability might vary from place to place and jurisdiction to jurisdiction. They undoubtedly change over time. They may reflect underlying assessments of the culpability of the relevant actors, as in a recent account offered by Ketan Ramakrishnan.306Ramakrishnan, supra note 138. They may track cultures of risk, as in the influential account of Mary Douglas and Aaron Wildavsky,307Mary Douglas & Aaron Wildavsky, Risk and Culture: An Essay on the Selection of Technological and Environmental Dangers (1983). or they may follow cultures of cognition, as in Dan Kahan’s formulation.308Dan M. Kahan, Cultural Cognition as a Conception of the Cultural Theory of Risk, in Handbook of Risk Theory: Epistemology, Decision Theory, Ethics, and Social Implications of Risk 725 (Sabine Roeser, Rafaela Hillerbrand, Per Sandin & Martin Peterson eds., 2012). Norms about foreseeability may reflect the biases and heuristics of the psychology of risk perception.309Amos Tversky & Daniel Kahneman, Judgment Under Uncertainty: Heuristics and Biases, 185 Science 1124 (1974). They will often be a function of powerful interests working to shape the way a community perceives and processes risk.310Cf. John Fabian Witt, Narrating Bankruptcy/Narrating Risk, 98 Nw. U. L. Rev. 303 (2003) (reviewing Edward J. Balleisen, Navigating Failure: Bankruptcy and Commercial Society in Antebellum America (2001); Bruce H. Mann, Republic of Debtors Bankruptcy in the Age of American Independence (2003); David A. Skeel, Jr., Debt’s Dominion: A History of Bankruptcy Law in America (2001)) (arguing that patterns of risk cognition arise, at least in part, out of institutionally constructed interests). Crucially, they reshape institutional allocations of power, turning open-ended jury questions into questions of law for judges.311Leon Green, Judge and Jury 248–56 (1930); see also Kenneth S. Abraham, The Forms and Functions of Tort Law 103–08 (5th ed. 2017); Fowler Vincent Harper, Judge and Jury, by Leon Green, 6 Ind. L.J. 285, 285 (1931) (book review). And in all events, foreseeability conventions, as Calabresi once crucially observed about formal doctrinal language, may advance tacit and even barely understood, but nonetheless valuable, social functions.312See Guido Calabresi, Concerning Cause and the Law of Torts: An Essay for Harry Kalven, Jr., 43 U. Chi. L. Rev. 69, 107 (1975).

We suspect that foreseeability conventions arise for all these reasons and more. But crucially, foreseeability conventions do not appear everywhere. They do not arise—or at least they are not especially useful—in social or doctrinal domains where the cases do not cluster or otherwise exhibit sufficient regularity to effectively yield conventional rules in place of ad hoc inquiry. And wherever they arise, and for whatever reasons they arise, the very idea of foreseeability conventions is fiercely contested. For us to successfully make the case that foreseeability conventions do the work we say they do, we need to dispel the daunting counterarguments of three distinguished philosophers.

IV. The Critics of Conventionality

A. Moore’s Objection and the Legal Process Response

Michael Moore anticipates a version of the conventional solution to foreseeability and rejects it. Foreseeability’s defenders, he observes, may try to rescue the concept by recourse to one or another variation of the idea that the multiple description problem is resolved by reference to cultural or cognitive conventions that shape or even dictate the legally appropriate description. Perhaps, he conjectures, “society ‘amalgamates accidents into a relatively small number of categories’” that then form the basis of foreseeability determinations.313Moore, supra note 18, at 392 (quoting Steven Shavell, An Analysis of Causation and the Scope of Liability in the Law of Torts, 9 J. Legal Stud. 463, 491 (1980)). Descriptive conventions might, for example, arise as typologies we use in our daily lives to deal with “our normal descriptive, explanatory, and predictive activities.” Or events may just cluster into clumps and categories for some unexplained but nonetheless regular reason, perhaps a reason arising out of cultural agreement, or perhaps a reason rooted in some common cognitive trait, maybe an evolutionary one. Or perhaps, Moore speculates, people establish conventions of foreseeability specific to particular “rights and interests.”314Id. at 392–95.

Moore contends that these descriptive conventions fail to rescue foreseeability from the open-endedness of descriptions. The variety of competing true descriptions for any one event, he says, is too bewilderingly vast to settle on any one conventional category. A conflagration is a fire. Or a yellow fire, or a hot fire, or a hot yellow fire.315See id. at 387–88, 392. Myriad further descriptors might be added to the list, or not, in any combination. To the extent we have categories that entrench particular ways of describing events, Moore observes, such conventions will often be cross-cutting: two available conventions will produce different foreseeability conclusions.316Id. A standard property damage description in the Buffalo River flooding case will yield a foreseeability finding, where a standard vessel collision story will not. Even when there is an available convention that might offer a solution, many of the cultural and cognitive categorizations most readily at hand seem to arise out of clumpings that are morally irrelevant. Human beings categorize according to morally arbitrary but heuristically convenient features, such as word length or mnemonic characteristics.317See id. at 393.

In the face of such a critique, the prospects for conventionality might seem bleak. Moore’s critique may pose a substantial problem for scholarly projects in other fields that place substantial weight on conventions for explaining legal practice.318See, e.g., Ashraf Ahmed, A Theory of Constitutional Norms, 120 Mich. L. Rev. 1361 (2022); Samuel Issacharoff & Trevor Morrison, Constitution by Convention, 108 Calif. L. Rev. 1913 (2020); Daphna Renan, Presidential Norms and Article II, 131 Harv. L. Rev. 2187 (2018); Katherine Shaw, Conventions in the Trenches, 108 Calif. L. Rev. 1955 (2020); Adrian Vermeule, Conventions of Agency Independence, 113 Colum. L. Rev. 1163 (2013); Adrian Vermeule, Conventions in Court (Harvard Pub. L. Working Paper No. 13-46, 2013).

But in truth, Moore’s critique lands wide of both types of foreseeability conventions we have described here. Per se foreseeability conventions, in particular, share none of the particular difficulties Moore identifies. Per se foreseeability rules require neither the selection of unspecified levels of specificity, nor culturally constructed clusters of ideas about foreseeability or event description. Once established, per se conventions of foreseeability fix in law a legally salient description of the kinds of events at issue.

A reader at this stage might be thinking: but doesn’t the per se foreseeability convention miss the point entirely? Sure, once established, a per se foreseeability rule may be determinate and workable, but that begs the questions: How are such rules established? Where do they come from? In Sinram, Judge Learned Hand suggested that any such determination “may seem merely a fiat.”319Sinram v. Pa. R.R., 61 F.2d 767, 771 (2d Cir. 1932). Isn’t that precisely Moore’s point?

The answer is no. Conventions do not arise out of arbitrary fiat but for reasons rooted in the legal process and its materials rather than metaphysics. Decisionmakers do not choose in the abstract among the vertigo-inducing array of possible descriptions. They choose among a small, curated sample of stories nominated by the parties. Plaintiffs set the agenda. Defendants respond. Moreover, the descriptions offered by the parties must map onto values embedded in the existing legal materials: values like deterrence or risk spreading in tort, private ordering in contract, or the prevention of undue punishment in the criminal law. Decisionmakers are thereby able to produce constrained (though not determined) answers to the foreseeability mystery. Judges and juries do not face a legal void of endless descriptors. They choose among a manageable, often binary, set of possibilities. And once they choose, at least when they choose in regularly recurring social situations, those choices are laid down as guideposts for the next time around.

The same kind of process gives definition to narrative foreseeability conventions, too, in areas like the liability of common carriers, cases of remote plaintiffs, or in the law of contracts. Each such doctrinal area instructs decisionmakers to describe the relevant events in a foreseeability inquiry at a particular level of generality. Moore objects that such conventions, which aim to narrow the field of available descriptions, nonetheless leave too many degrees of freedom to decisionmakers to tell the stories as they see fit.320Moore, supra note 18, at 393. But once again, the actually existing legal process and materials sharply reduce any such freedom. Judge and jury do not sit as uninhibited short story writers, crafting narratives on a blank slate. They adopt situated descriptions that draw on the material supplied to them by the parties and that aim to comport with values embedded in the law. Typically, they choose between the candidates supplied by the contending sides. The critic of the conventions thesis, who insists that such a decision smells of fiat, holds an impossibly demanding conception of legal decision-making. In hard cases, a constraining process that focuses the parties and the decisionmaker on a shared inquiry of obvious moral significance ought to count as success. That was Judge Hand’s view: sure, there were irreducible dimensions of judgment involved for the legal decisionmaker, he conceded, “but that is always true, whatever the disguise.”321Sinram, 61 F.2d at 771. All the better if tacit unspoken conventions from culture or social values or cognitive regularities shape the answers decisionmakers reach, and all the better still if those answers sometimes congeal into fixed and useful rules.322Cf. Ronald Dworkin, Law’s Empire (1986); Lon L. Fuller, Positivism and Fidelity to Law—A Reply to Professor Hart, 71 Harv. L. Rev. 630 (1958).

Most recently, a provocative collaboration between philosopher Joshua Knobe and lawyer-philosopher Scott Shapiro has offered a different critique of the foreseeability standard.323See generally Knobe & Shapiro, supra note 18. In the Knobe-Shapiro view, the tort law of proximate causation purports to be about foreseeability but is actually about abnormality. Drawing on experiments in folk intuitions about causation, they contend that foreseeability is relevant insofar as it informs judgments about what they call statistical normality, or brute probability.324Id. at 216 (“[F]oreseeability matters because foreseeable events are statistically normal.”). But foreseeability fails to explain the outcomes of a range of decided cases, they assert, because folk intuitions about proximate causation are not only about the statistical abnormality of the relevant harm but also about the moral abnormality of the harm.325Id. at 186–88. The foreseeability of harms in intervening causation cases, for example, will often not reflect the wrongfulness of an intervening actor’s conduct.326Id. at 214–29. Of course, if foreseeability is not the right standard, it goes without saying that foreseeability conventions will not be the right approach either.

The difficulty for Knobe and Shapiro is that abnormality (whether statistical or moral) has the same multiple description problem that jurists from Morris to Moore identified in the foreseeability standard. Whether a given sequence of cause and effect is abnormal or normal will depend substantially on the level of generality at which the decisionmaker characterizes the sequence. Think again of the Buffalo flooding case. Was it a spectacular flood inundating the city of Buffalo that was abnormal? Or was it the simple and also true fact of downstream property damage? The latter, of course, would be perfectly ordinary. Was the bridge attendant’s negligence normal? Or was it an unusual work-time trip for shots and chasers at a nearby tavern? Consider one of Knobe and Shapiro’s examples, in which the defendant drove negligently and caused a car accident, which, in turn, led a bystander to be injured while she was assisting a person injured in the accident when she was struck by a car involved in a second and causally unrelated accident.327Id. at 215. The events can be characterized as entirely ordinary. A bystander was injured by a car on a highway because she was put at risk by the defendant’s negligent driving. Or the episode can be characterized as altogether abnormal given the improbability of a second independent negligent accident at the same time and place. Or take one of our favorite cases, Brauer v. New York Central & Hudson Railroad Co., in which an accident at a railroad crossing caused barrels of hard cider to be thrown to the ground. Onlookers at the scene carried the barrels away and consumed them.328Brauer v. N.Y. Cent. & Hudson R.R., 103 A. 166, 166–67 (N.J. 1918). Is the right description that negligent railroading caused damage to the wagon and a loss of its cargo? Or that negligent railroading caused damage to the wagon, which happened to be loaded with delicious alcoholic beverages, with which third parties, in turn, ran off? Both are true stories. In one, the sequence is normal and foreseeable, and in the other, it is neither.

Description problems are pervasive in the abnormality argument. But there are solutions. In fact, there are conventional solutions. Knobe and Shapiro recite a series of cases in which abnormality judgments cluster into what are essentially per se rules of abnormality: rescuers appear in the normal course of cause-and-event sequences as a matter of law, for example, and so does subsequent negligent medical care. What Knobe and Shapiro are observing is the way in which the law takes under-specified analyses of the normality (or foreseeability) of a causal sequence and reduces them into more readily applied practical rules. We have called these foreseeability conventions, but one might also call them normality conventions without too much alteration, and in any event, the cases in this area (as Knobe and Shapiro observe) are famously imprecise in the terminology they use and the concepts they invoke.329For the series of cases cited by Knobe and Shapiro, see Knobe & Shapiro, supra note 18. The point is not really crucial for this Article, but the biggest puzzle we see for the Knobe-Shapiro account (and the best reason to stick with some version of foreseeability conventions instead of abnormality conventions) is that Knobe and Shapiro understate the prevalence of cases holding initial wrongdoers liable for damages caused by intervening criminal or otherwise intentional actors. In their view, such cases are more typically decided in favor of the lesser, background wrongdoer. The exceptions, they assert, are anomalies. Our reading of the caselaw—especially, though not exclusively, the more recent caselaw—is different. See, e.g., Nallan v. Helmsley-Spear, Inc., 407 N.E.2d 451 (N.Y. 1980). See generally Robert L. Rabin, Enabling Torts, 49 DePaul L. Rev. 435 (1999). Indeed, folk intuitions about normality may be shaping judgments about foreseeability, and vice versa. Intuitions about normality, for example, may help construct the tacit conventions that could (if sufficiently entrenched) give shape even to otherwise ill-specified normality or foreseeability standards, thereby rescuing them from the indeterminacy of the description problem. Of course, relying on such tacit storytelling conventions will run headlong into Moore’s objections that there are too many competing conventions—too many available ways of telling the relevant stories and characterizing the relevant events.330Note that the experimental method may itself have a multiple description problem—one that should be tested by experiment. On the other hand, the same responses that supported foreseeability conventions apply here as well. The legal process and its bilateral structure delimit the number of conventions on offer. The structured presentation of litigation excretes conventions for particular social settings. Once established, those conventions replace the open-ended inquiry in future cases and supply a convention with which to decide them.

B. Shortcuts or Categorical Imperatives?

One last objection warrants attention. On this view, the rules we identify and reinterpret here are not conventions at all, at least not if by “convention” we mean proxies, heuristics, or short-cuts designed to circumvent otherwise intractable moral inquiries and to advance, in rough-and-ready fashion, the law’s basic goals. Instead, the objection insists, the categories embody the right moral obligations attaching to the social relation at issue. The rules we describe as foreseeability hacks or foreseeability fixes reappear in this view as efforts to state the appropriate moral obligations in the relevant situation.

To be clear, nothing in this Article need dismiss the moral structure of the inquiry or the importance of properly identifying the moral relations of the parties. As we see it, foreseeability conventions aim to help legal decisionmakers analyze precisely such questions. But at least two reasons lead us to think that the rules we describe here are better thought of as foreseeability hacks rather than moral categories.

The first has to do with the fashion in which foreseeability conventions emerge. The pattern is well established. Courts ask about the foreseeability of some harm. They analyze foreseeability in the case in question. And then they set down a rule for the cases matching that category in future disputes. And so, we get a convention such as the zone-of-danger convention in emotional distress cases.331See Falzone v. Busch, 214 A.2d 12, 14 (N.J. 1965). Such a convention does not stand for the special moral wrongfulness of negligently causing emotional distress to someone who was also at risk of a physical impact. People outside that zone will regularly exert more pull on our intuitions. Consider parents or other family members of someone negligently injured or killed. But the law adopts a useful fix, nonetheless. Especially vulnerable plaintiffs are not always especially sympathetic once we look at all the facts, but the eggshell plaintiff rule generally treats them that way because it is useful to have a rough-and-ready answer in such cases. In certain settings, rescuers will be unforeseeable for sure, and their harms as a moral matter will be their own responsibility. The law treats them as a category because the proxy convention is useful. Each such rule or convention evolved out of the miasma of the foreseeability inquiry and then stuck, not because it would always be correct, but because it was correct often enough.

The second problem with the categorical view of the conventions we describe here is deeper. The categorical view is quintessentially circular. Do wrongdoers not owe third parties for economic losses? The view we have outlined here holds that the law has established a convention of unforeseeability to govern such cases. Does the rule instead reflect the law’s conclusion that the moral relationship between a wrongdoer and its victim is better reflected by a rule of no duty to prevent pure economic losses? Surely not, or at least not so long as moral duties are not reverse engineered from legal ones. Absent such circularity, conventions like the economic loss rule are too arbitrary to be sustained by strictly moral considerations. Are contracts for wedding dresses or the treatment of a deceased loved one’s body categorically more deserving of certain kinds of emotional distress damages? Well, sometimes. But as a rule across the category, the special rules of foreseeability for emotional distress in such cases is a rule of approximation, not a moral truth of the category as such.

Conclusion: The Firefighter’s Rule and the Logic of Doe v. Mckesson

Conventions of the kind we describe here do not resolve all of the law’s hard foreseeability questions, to be sure. When events in the world refuse to conform to experience—when they take no regular or predictable shape—conventions are of less use. In our recurring illustration from the great Buffalo River flood, for example, flooding damages to upstream plaintiffs would have presented an open-ended foreseeability question for which Judge Friendly had no readily available Palsgraf convention. Similarly, no convention we know of could have supplied a frame for resolving whether it was foreseeable that a 1964 Second Circuit decision might return baseball’s Yankees franchise to dominance. World-historic trajectories, like idiosyncratic sequences, do not cluster; they are insusceptible to per se conventions, and their unusual features evade even the looser narrative conventions.

The problem seems all the greater given that conventions resting on tacit agreement seem to be quickly breaking down in the age of political and cultural fracture.332Josh Chafetz & David E. Pozen, How Constitutional Norms Break Down, 65 UCLA L. Rev. 1430 (2018); Renan, supra note 318, at 2242–43. See generally Daniel T. Rodgers, Age of Fracture (2011). Will American tort doctrine be capable of producing a convention, for example, to govern foreseeability inquiries in cases of negligent exposure to novel viruses like COVID-19? What about third-party transmission cases?333Cf. Doe v. Cochran, 210 A.3d 469, 479 (Conn. 2019) (“[A] health care provider who negligently misinforms a patient that he does not have an STD [(sexually transmitted disease)] owes a duty of care to an identifiable third party who foreseeably contracts the STD as a result of the provider’s negligence.” (footnote omitted)).

Yet conventions continue to offer hope for unexpected solutions to some of today’s most contested foreseeability inquiries. Consider the much-discussed tort claims brought by Louisiana law enforcement officers against DeRay Mckesson, a BLM leader, on the theory that he negligently encouraged BLM protests in which the plaintiffs were injured. Commentators worry (with good reason) that the use of open-ended foreseeability standards in a “negligent protest” cause of action risks suppressing valuable social protests.334Tasnim Motala, “Foreseeable Violence” & Black Lives Matter: How Mckesson Can Stifle a Movement, 73 Stan. L. Rev. Online 61, 70 (2020); Ryan S. Park, Doe v. McKesson: The “Duty” of Certification and the “Risks” of a Negligent Protester Standard, 95 Tul. L. Rev. 1009, 1010–11 (2021); Nadine Strossen, The Interdependence of Racial Justice and Free Speech for Racists, 1 J. Free Speech L. 51, 55–56 (2021); Emerson J. Sykes, In Defense of Brandenburg: The ACLU and Incitement Doctrine in 1919, 1969, and 2019, 85 Brook. L. Rev. 15, 29–31 (2019); Timothy Zick, The Costs of Dissent: Protest and Civil Liabilities, 89 Geo. Wash. L. Rev. 233, 269 (2021). And the case has hardly come to a close. But as the process unfolds, the Louisiana Supreme Court has (at least for now) missed an opportunity to rely on the hidden juridical structure of foreseeability conventions to bring closure to an otherwise perilous set of legal claims.

Plaintiff John Doe alleges that Mckesson authorized, directed, and ratified BLM protests that then became violent.335Doe v. Mckesson (Mckesson III), 945 F.3d 818, 823 (5th Cir. 2019). Doe further alleges that during those protests, an unidentified person threw “a piece of concrete or similar rock-like object” that struck Doe in the head, causing “loss of teeth, a jaw injury, a brain injury, a head injury, lost wages,” and other losses.336Id. Initially, in 2017, the Federal District Court for the Middle District of Louisiana dismissed Doe’s complaint, relying on a 1982 decision of the U.S. Supreme Court holding that the First Amendment protects protest leaders from state tort claims absent the incitement of immediate violence.337Doe v. Mckesson (Mckesson I), 272 F. Supp. 3d 841, 846–48 (M.D. La. 2017) (citing NAACP v. Claiborne Hardware Co., 458 U.S. 886 (1982)). But in an arduous, drawn-out sequence over the subsequent years, the courts have reinstated Doe’s claims, giving foreseeability a starring (though ultimately star-crossed) role.

A panel of the Fifth Circuit Court of Appeals was the first to come out in favor of the plaintiff. Foreseeability, for the panel majority, was the heart of the matter: “By ignoring the foreseeable risk of violence that his actions created,” Judge E. Grady Jolly explained in Mckesson III that Mckesson had “failed to exercise reasonable care in conducting his demonstration.”338Mckesson III, 945 F.3d at 827. Negligent protest, it seemed to Judge Jolly, was a quintessential tort under the Louisiana Civil Code’s bedrock provision that “[e]very act whatever of man that causes damage to another obliges him by whose fault it happened to repair it.”339Id. at 826 (alteration in original) (quoting La. Civ. Code Ann. art. 2315 (2022)). If Mckesson could be said to have wrongly caused the foreseeable damages complained of, went the court’s reasoning, he was obliged to pay damages. Several months later, in an unusual move, the Fifth Circuit withdrew its decision and substituted a new one that redoubled its support for the foreseeability standard, authorizing actions for injuries “that were foreseeable as a result” of the defendant’s “negligent conduct,” even when the injuries were caused by the defendant’s “own negligent conduct and the violent actions of another.”340Id. at 829 (emphasis added). A month after that, the same panel withdrew and reissued its opinion again, now accompanied by a dissent that made moves now familiar to readers of this Article. Louisiana law, the dissenter contended, has adopted a convention under which injuries caused by third parties are deemed foreseeable only in cases of a “special relationship”341See supra Section III.C.7. between the defendant and either the plaintiff or the relevant third party.342Mckesson III, 945 F.3d at 836–38 (Willett, J., concurring in part and dissenting in part). In any event, according to the dissent, the plaintiff-officer’s injuries were unforeseeable when the foreseeability inquiry was framed at the appropriate level of generality.343Id. at 844–45.

Denial of Mckesson’s petition for rehearing en banc by the full Fifth Circuit produced a flurry of further dissenting opinions that both favored a conventions approach designed to avoid quagmires about levels of abstraction344See Doe v. Mckesson (Mckesson IV), 947 F.3d 874, 878 (5th Cir. 2020) (Dennis, J., dissenting). and reframed the level of generality of the foreseeability inquiry.345See id. at 879 (Higginson, J., dissenting). The panel painted in broad and general strokes to frame the foreseeability question: generic injuries followed foreseeably from unreasonable protests. The Fifth Circuit dissenters, by contrast, worked at a more granular level, filling in the details. “An assault on a police officer by a third-party,” Judge Stephen A. Higginson reasoned, was not a particular foreseeable result of the acts for which Mckesson was responsible.346Id.

In the midst of these foreseeability battles, an unexpected opinion by Judge James Ho pointed the way to a fix—and did so by exhuming a crucial per se foreseeability convention. Judge Ho concurred in the court’s refusal to rehear the case en banc. But faced with the panel’s capacious view of foreseeability on the one hand, and Judge Higginson’s narrow view on the other, Judge Ho proposed that a work-around to the intractable foreseeability debate lay already within the common law doctrine—a work-around that obviated the supposed need to resolve the open-ended foreseeability question, let alone any attendant constitutional questions about free speech. “The professional rescuer doctrine,” Judge Ho explained (quoting a two-decade-old Fifth Circuit opinion), “bars recovery by a professional rescuer injured in responding to an emergency.”347Id. at 875 (Ho, J., concurring) (quoting Gallup v. Exxon Corp., 70 F. App’x 737, 738 (5th Cir. 2003)).

And indeed, it does. Like other per se conventions, including the “danger invites rescue” rule, the eggshell plaintiff rule, the rule for subsequent negligence in medical care, the pure economic loss rule, and more, the “firefighter rule,” as the Restatement of Torts authors call it, cuts off the question of foreseeability and replaces it with a categorical, one-size-fits-all determination. “Some jurisdictions,” explains the Restatement (Third), “have a rule that firefighters—or, more broadly, professional rescuers—cannot recover for harm they suffer as the result of negligence that requires their services.” A homeowner, for example, is not liable to a firefighter injured when the homeowner’s “failure to exercise reasonable care wen smoking results in a fire.”348Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 51 cmt. m (Am. L. Inst. 2012). As of this writing, reporters Nora Freeman Engstrom and Michael D. Green for the American Law Institute are developing a new, as-yet unpublished restatement of the firefighter’s rule. See Nora Freeman Engstrom & Michael D. Green, Am. L. Inst., Restatement of the Law Third, Torts: Miscellaneous Provisions, https://www.ali.org/​projects/​show/​torts-miscellaneous-provisions/​#​_​status [https://perma.cc/​25FH-KCR5]. The firefighter’s convention is rooted in what the Restatement calls “a mélange of public-policy considerations”349Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 32 cmt. c (Am. L. Inst. 2012). that ultimately crystalize into a categorical foreseeability determination. As the Connecticut Supreme Court puts it, “[t]he most compelling argument for the continuing validity of the rule is the recognition that firefighters and police officers often enter property at unforeseeable times and may enter unusual parts of the premises under emergency circumstances.”350Furstein v. Hill, 590 A.2d 939, 943 (Conn. 1991). The convention, in short, is a per se resolution of difficult foreseeability questions, preventing “firefighters and police officers,” in the words of one federal district court, from recovering for risks “inherent in, and foreseeable as a part of their duties.”351Collins v. Flash Lube Oil, Inc., No. 09-cv-626-HTW-LRA, 2012 WL 4605562, at *2 (S.D. Miss. Sept. 30, 2012) (quoting Farmer v. B & G Food Enters., Inc., 818 So. 2d 1154, 1156 (Miss. 2002)).

The Restatement of Torts authors observe that the rationales behind the resolution have shifted over time.352See Restatement (Third) of Torts: Liab. for Physical & Emotional Harm § 51 cmt. m (2012). Historically, the rule was connected to the hard-and-fast, one-size-fits-all rules for parties entering onto land without the permission of a landowner or land occupier. Such rules, as we have seen, themselves function as conventional resolutions to the foreseeability inquiry in real property torts.353See supra notes 289–92 and accompanying text. More recently, courts explain the firefighter’s convention by reference to the double-recovery for state employees already covered by systems of workers’ compensation and often additional pension or accident programs,354E.g., Syracuse Rural Fire Dist. v. Pletan, 577 N.W.2d 527, 533 (Neb. 1998). the related notion that professional rescuers are already compensated for the risks they take by a mix of wage compensation and benefits,355E.g., Moody v. Delta W., Inc., 38 P.3d 1139, 1142 (Alaska 2002); Furstein v. Hill, 590 A.2d 939, 944 (Conn. 1991). and the interest in encouraging the public to call for help without fear of liability.356E.g., Pottebaum v. Hinds, 347 N.W.2d 642, 645 (Iowa 1984).

For a brief moment, the firefighter’s convention seemed likely to give Mckesson much-needed direction. Influential commentators championed the application of the Firefighter’s Rule on Mckesson’s behalf.357See Eugene Volokh, Negligence Claims Brought Against Protest Organizers: More on the Tort Law Side of Doe v. Mckesson, Reason (Dec. 19, 2019, 11:29 AM), https://reason.com/​volokh/​2019/​12/​19/​negligence-claims-brought-against-protest-organizers-more-on-the-tort-law-side-of-doe-v-mckesson [https://perma.cc/​QNJ3-KXTR]; Eugene Volokh, Supreme Court Sends DeRay Mckesson Black Lives Matter Protest Case Down to the Louisiana Supreme Court, Reason (Nov. 2, 2020, 1:10 PM), https://reason.com/​volokh/​2020/​11/​02/​supreme-court-sends-deray-mckesson-black-lives-matter-protest-case-down-to-the-louisiana-supreme-court [https://perma.cc/​CT6L-V5TD]. The U.S. Supreme Court remanded the case to the Fifth Circuit with instructions to certify the key state law tort questions to the Louisiana Supreme Court.358Mckesson v. Doe (Mckesson V), 141 S. Ct. 48, 51 (2020). The Fifth Circuit, in turn, subtly recommended the firefighter’s convention in its certification opinion.359See Doe v. Mckesson (Mckesson VI), 2 F.4th 502, 503–04 (5th Cir. 2021) (“In the meantime our attention has been drawn to a separate aspect of Louisiana law, the Professional Rescuer’s Doctrine, that could be dispositive.” (footnote omitted)), certifying questions to 320 So. 3d 416 (La. 2021). The state courts, or so it seemed, stood poised with a foreseeability convention at the ready to solve the knotty problem of protester liability with a single swift move.

Sadly, the Louisiana Supreme Court missed its opportunity, adopting instead the Fifth Circuit panel’s open-ended and abstract foreseeability inquiry.360See Doe v. Mckesson (Mckesson VII), 339 So. 3d 524 (La. 2022). Misconstruing the firefighter’s convention as a species of assumption of the risk, the Louisiana court ruled that the state had quietly abolished the firefighter’s rule when it abolished the old rule of contributory negligence.361Id. at 534–35 (ruling that under the Louisiana comparative negligence statute, assumption of the risk “no longer ha[s] a place in Louisiana tort law” (citing Murray v. Ramada Inns, Inc., 521 So. 2d 1123, 1132 (La. 1988))). But the firefighter’s convention is self-evidently not an assumption of the risk rule at all. It has nothing to do with negligent behavior on the part of the law enforcement officers in question; in any such case, after all, a simple application of the comparative negligence defense would suffice to allocate liability between plaintiff and defendant. Set alongside the panoply of doctrinal conventions we have described in this Article, the firefighter’s rule appears more properly as a foreseeability convention designed to meet the law’s most basic goals and to do so in a fashion more easily administered than the general foreseeability inquiry otherwise allows.

***

The law’s strategies for managing the open-endedness of foreseeability inquiries will only go so far in resolving such novel challenges as in the still-unfolding Mckesson litigation saga—and not just because the courts will sometimes miss the opportunity afforded by conventions as the Louisiana court did in Mckesson.

Conventional resolutions are backward-looking: they adopt past agreements to deliver certainty in the here-and-now. In this sense, critics of foreseeability like Green, Morris, and Moore are right. For the most contested legal challenges, foreseeability is indeed a circular and indeterminate standard.

Yet, even if new social and professional settlements are required to resolve pressing new legal problems, foreseeability conventions serve as an illustration of the process by which jurists resolve such settlements into new legal rules. As new problems arise, lawyers and judges quickly get down to the business of reducing foreseeability questions to a next generation of conventions. New conventions are added to the law’s growing catalog. Third parties will be foreseeable victims of virus-related negligence. Or not. Protest leaders will face negligent protest claims. Or not. Resolutions will congeal into rules that embody rough-and-ready value judgments on the goals of the law in the relevant area. In some instances, such value judgments will be clear, or at least uncontroversial. In others, the underlying questions about how society arranges its affairs will be suppressed beneath the smooth surface of the convention itself, buried out of view. And in others still, those underlying questions will remain controversial, and no conventional solution will prevail. In any event, foreseeability conventions will be (and have been) principal vehicles for the delivery of those resolutions the law is able to achieve.

Conventional solutions to the law’s foreseeability crisis in areas like contract, crime, and tort cry out for more research in the myriad legal domains in which foreseeability plays a crucial doctrinal role. Do conventions populate other areas, too? How, after all, does doctrine in other areas solve multiple description problems? Are there conventions for foreseeability in the law of bankruptcy, constitutional law, corporations, defamation, environmental law, antitrust, immigration, or takings? How about personal jurisdiction and patents? Do such doctrinal domains host narrative conventions or per se conventions? Or perhaps some further kind of convention altogether? The foreseeability convention, it seems, is a research agenda.