"Of Frightened Horses and Autonomous Vehicles: Tort Law and its Assimilation of Innovations", Graham 2012:

The liability rules that come to surround an innovation do not spring immediately into existence, final and fully formed. Instead, sometimes there are false starts and lengthy delays in the development of these principles. These detours and stalls result from five recurring features of the interplay between tort law and new technologies, which also will provide the basic architecture for this Article. First, the initial batch of cases presented to courts may be atypical of later lawsuits that implicate the innovation, yet relate rules with surprising persistence. Second, these cases may be resolved by reference to analogies that rely on similarities in form, and which do not wear well over time. Third, it may be difficult to isolate the unreasonable risks generated by an innovation from the benefits it is perceived to offer. Fourth, claims by early adopters of the technology may be more difficult to recover upon than those that arise later, once the technology develops a mainstream audience. Fifth, and finally, with regard to any particular innovation, it may be impossible to predict whether, and for how long, the recurring themes within tort law and its application that tend to yield a “grace” period for an invention will prevail over those tendencies with the opposite effect. In short, it takes time for an innovation, such as autonomous vehicles, to become fully assimilated within everyday tort law, and one rarely can anticipate the precise timetable for this process, or its final results.

Consider, for example, the initial batch of negligence lawsuits that involved automobiles.5 These cases, dating from the very first years of the twentieth century, entailed none of the usual fodder for automobile litigation today: • They generally did not involve multi-car collisions— there were not enough automobiles on the road at the time for these accidents to occur very often.6 • Suits involving injured pedestrians also did not predominate, though they certainly were not unheard of; the reasons for this are less clear, but likely owed to the potency of the contributory negligence defense in that era.7
There also were few lawsuits brought by injured passengers against early automobile drivers. This deficit owed in part to interspousal immunity rules, which were still vibrant at that time.8 Unspoken social norms also may have discouraged such suits, especially among the elite who comprised most early automobile owners,9 and perhaps these passengers also worried that if they brought suit, the driver would have a viable assumption of the risk or contributory negligence defense.10
• Finally, there were few (if any) lawsuits brought against automobile manufacturers, in which plaintiffs complained of negligently constructed vehicles.11 Possibly, these suits were suppressed by a sense that a lack of contractual privity with the manufacturer would defeat a claim, except in rare circumstances.12 Yet many early automobile consumers purchased their vehicles directly from the manufacturer,13 such that any privity rule, on its own, would not have precluded lawsuits these buyers might have been inclined to file.
...So, if none of the modern automobile-lawsuit categories predominated back in 1900, what did? The answer: most early automobile cases involved claims that the sight or sound of a motor carriage caused a horse to take fright, resulting in injury either to the horse’s rider, the occupants of a carriage or wagon the horse had been towing, or the horse itself.15
...These frightened-horse cases would make for an interesting anecdote, and little more, except for the fact that they produced automobile-friendly rules with staying power. Turn-of-the-century judges and juries knew that any new contrivance on the highways could frighten a horse23—as the earlier caselaw (which had involved sources of highway terror as varied as bicycles,24 trolley cars,25 and elephants26) had demonstrated. It also was understood that as time passed and animals became more familiar with a device, the number of accidents attributable to frightened horses would decrease.27 Thus there was less need for stringent judicial regulation; if automobiles were more than a passing fad,28 then the problem would resolve on its own.29 Accordingly, courts swiftly and soundly rejected the view that operation of an automobile on a public highway might, by itself, amount to negligence simply because the vehicle was prone to frighten horses.30 Moreover, rulings in early frightened-horses cases also snuffed out any prospect that automobiles would be regarded as nuisances,31 or that their owners would be subject to strict liability for harms associated with their use.32
...Particularly given that many early automobile owners were among the nation’s ultra-rich, the possibility of a strict-liability regime to govern automobile use was not all that far-fetched. See KIMES, supra note 7, at 92 (“Who bought automobiles? Prior to the turn of the century, they were, most conspicuously, the same people who owned private railroad cars, had stables for at least twenty carriages at their estates on Long Island, and summered at their opulent ‘cottages’ in Newport.”). Id. at 152 (observing that as of 1900, only 4,000 Americans owned automobiles, with this number rising to 40,000 by 1904). Id. at 189 (quoting future President Woodrow Wilson as stating, in 1906, “Nothing has spread socialistic feeling in this country more than the automobile,” as “to the countryman they are a picture of the arrogance of wealth, with all its independence and carelessness.”).

More specifically, as Professor Gregory Mandel and others have observed, in resolving disputes that involve a new device, courts often focus on similarities in form between the innovation and existing technologies.35 Over time, however, these analogies tend to give way to analysis that incorporates a more thoughtful consideration of the risk profiles presented by the invention.
The early application of defamation law to broadcasts by radio and television offers one example of such a transition. The question of whether defamation communicated by way of radio or television should be regarded as libel, or as slander, vexed courts for decades. The distinction matters because it is easier to recover for libel than for slander.36 In trying to place these new technologies in the proper slot (the notion of a third category never having caught on37), precedent and policy were at odds. On the one hand, black-letter law provides that slander involves aural communications, whereas libel entails printed material.38 This rule supports the treatment of defamation by radio and television as slander, since both rely upon sound for their publication.39 At the same time, the more liberal standard for recovery for libel recognizes that words on the printed page tend to reach a broad audience, as radio and television broadcasts did.40 This distinction points in the direction of treating defamation by radio and television as libel.41
Courts that wrestled with this conflict in the early days of radio and television reached some awkward compromises. Among them, some courts held that if a radio broadcast had been read from a script, it was libel; otherwise, it was slander.42

An arguably similar story surrounds airplanes. Ground damage attributable to fallen aircraft long was considered one of the rare circumstances in which strict liability, as opposed to negligence, would provide the appropriate rule. This approach dates back to Guille v. Swan, an 1822 case that involved hot-air ballooning.49 After the Wright brothers’ fateful flight, commentators assumed that Guille’s strict-liability rule for ground damage would apply to motorized aircraft.50 Both balloons and airplanes rose to, and fell from, the sky, after all; and the plaintiffs in such cases presumably would be equally blameless. Suitably, the Restatement of Torts endorsed strict liability for ground damage caused by aircraft crashes.51 This view proved remarkably persistent, even as commercial aviation became commonplace and its After a pitched debate, the safety record improved.52
Restatement (Second) of Torts retained the strict-liability rule.53 Only relatively recently has a critical mass of courts and commentators determined that airline operators, given their relatively benign safety profile, should be subject to a liability rule for ground damage that is no more stringent than the one applied to, for example, automobile drivers.54 Restatement (Second) of Torts retained the strict-liability rule.53 Only relatively recently has a critical mass of courts and commentators determined that airline operators, given their relatively benign safety profile, should be subject to a liability rule for ground damage that is no more stringent than the one applied to, for example, automobile drivers.54

According to conventional wisdom, tort law routinely penalizes innovation, while rewarding manufacturers who adhere to the status quo. The leading article in this vein is Peter Huber’s Safety and the Second Best: The Hazards of Public Risk Management in the Courts.104 In this article, Huber posits that, as a general matter, “courts . . . greatly prefer natural, old, or established hazards to those deriving from new technologies.”105 According to Huber, more tort suits are directed against innovations than is warranted by the risks these technologies generate, as compared to the hazards posed by more- entrenched substitutes.106 Huber credits this disparity to both a sort of cognitive dissonance among courts and the general public, and to a surfeit of information regarding the risks posed by new technologies.
...This Article essay leaves a full critique of Huber’s thesis for another day. At a minimum, however, if one examines the historical record, one quickly discovers that Huber’s argument applies to some innovations far better than it does to others. Information regarding the hazards associated with new technologies is often quite limited, relative to the data that surrounds their more well-entrenched substitutes. For example, in 1900, almost every American experienced the dangers (and other disagreeable qualities) of horses on a daily basis.110 Meanwhile, the public at that time appreciated that automobiles might be dangerous, but many observers associated these perils with irresponsible use, instead of perceiving them as being inherent in the technology itself.
And as discussed elsewhere in this Article, other technologies also have benefitted, early on in their diffusion, from the perception that they were less risky than they ultimately proved to be, or that those who adopted these innovations were taking their chances in doing so. As to any specific invention, it may be impossible to anticipate whether these forces will subsidize the method or device in its incipiency, or whether the dynamics that Huber describes will prevail.
A related limitation of Huber’s thesis is that, insofar as it suggests that uncertainty regarding potential liability tends to stymie innovation, it overlooks the fact that uncertainty can cut two ways. Uncertainty as to the prospect, viability, and magnitude of tort claims regarding an invention may chill its development or diffusion. But uncertainty as to matters such as the existence of a cause of action and the likelihood of recovery also may stifle the filing of claims that attack the innovation as unreasonably dangerous.111

At the outset, I believe that the types of tort claims associated with autonomous vehicles will evolve over time. Early claims likely will resemble contemporary lawsuits that allege negligent vehicle use.112 These cases will probably also involve a relatively “passive” plaintiff class; that is to say, plaintiffs who did not knowingly expose themselves to any of the potential novel risks generated by these vehicles. As basic ground rules regarding the use of autonomous vehicles emerge, so too will new causes of action. The first such claims likely will continue to ascribe fault to the users of autonomous vehicles, drawing distinctions between “proper” and “improper” use premised on the slowly accumulating body of knowledge on this topic. Some of these claims may lack analogues in current torts practice. For example, perhaps plaintiffs will attack decisions to utilize autonomous vehicles in specific areas where experience has shown that they present relatively significant dangers.
When suits against the manufacturers of autonomous automobiles first appear, they likely will sound in a failure to warn of some danger associated with vehicle use, as opposed to a design defect. For a plaintiff to reach a jury on a design- defect claim, she may have to engage in a searching review of the computer code that directs the movement of these vehicles.
This project may be difficult, and expensive. Warning-defect claims would seem, on the whole, to be easier to grasp and prosecute, and represent more logical candidates for user defendants looking to foist blame onto another party. On the whole, I am more optimistic than Professor Marchant appears to be about the interplay between tort law and autonomous vehicles. It is possible that the development of these vehicles will succumb to an onslaught of tort suits, or that the prospect of ruinous liability will chill investment to the point that research on these devices stalls. But I doubt it. Assaults on this technology may not be intuitive, and the types of claims especially feared by Professor Marchant will emerge only gradually, if ever, such that the technology will have an opportunity to evolve and further reduce its risk profile prior to encountering a wave of tort litigation. It strikes me as at least as likely that autonomous vehicles will benefit from a “honeymoon period,” and even after that span expires, the innovations incorporated within autonomous vehicles may confer upon them a competitive advantage relative to conventional automobiles—which, after all, also may be attacked insofar as they fail to incorporate the safety-enhancing devices that appear within autonomous cars.