The automated motor vehicles that I have discussed this week are just one example of the remarkable technologies coming to our roads, skies, homes, and even bodies. A decade from now, we’ll marvel at how advanced these new products are. But a decade after that, we’ll marvel at how anachronistic they have become.
Rapid technological change means that obsolescence is inevitable, and planning for it is as much a safety strategy as a business strategy. Responsible developers and regulators will need to consider the full lifecycle of products long before those products ever reach the market.
Cars of the early 20th Century (JSTOR) were essentially beta products. In 1901, Horseless Age magazine noted that “[i]f a manufacturer finds that the axles of his machine are” breaking, then “the next lot of vehicles are provided with axles of a slightly larger diameter and so on until they begin to stand up pretty well.” In 1910, a GM engineer testifying in MacPherson v. Buick Motor Co. explained that “the only means” for a designer to get information about a vehicle’s performance “is to use the customers, that is to go over the complaint correspondence.”
As I noted yesterday, it is at least conceivable that a similar approach to modern design could counterintuitively end up saving lives by accelerating safety-critical innovation. But even a more cautious approach to product design and deployment is necessarily iterative.
The general bent of incremental innovation is toward greater safety. The electronic stability control now required in new cars, for example, could save thousands of lives a year if deployed fleetwide. But given the slow turnover in cars–the average age of today’s fleet exceeds ten years–reaching saturation could take years.
At the same time, new products can present new dangers. Most of these dangers are marginal (though certainly not for the victims), but some may attain the social significance of asbestos or thalidomide. Responding–through recall, remediation, and restitution–can be costly and difficult.
The coming years may see the commercialization of automated vehicles, aerial drones, embedded medical devices, service robots, wearable computers, smart houses, and other highly advanced and safety-sensitive technologies. I am optimistic that early generations of these products will for the most part be reasonably safe by today’s standards.
However, I am concerned that, first, isolated incidents involving these products will create feelings of helplessness and panic that unjustifiably stymie their wider adoption and that, second, these early products will still be around years later when they are much less safe than whatever has become state of the art.
Actively managing both sudden and creeping obsolescence in these systems compels the integration of key legal and technological tools into their design and marketing. Foremost among these is the technical ability and the legal authority to “virtually recall” a product–to automatically update it or, in an extreme case, simply brick it.
My new working paper, Proximity-Driven Liability, considers over-the-air updates as part of a broader expansion in commercial sellers’ information about, access to, and control over their products, product users, and product uses. Tesla, for example, already tracks and updates its electric cars remotely, and it remains in privity with their owners.
The paper’s key point is that this growing proximity could significantly expand commercial sellers’ point-of-sale and post-sale obligations in tort to people endangered by their products. However, it also argues that the uncertainty caused in part by changing liability could actually drive companies to further embrace proximity. This observation, while largely predictive in the paper, also has prescriptive merit.
When the continued use of a legacy product imposes safety or environmental costs that far outweigh the benefits, regulators should have the ability to effectively remedy the dangerous condition or else remove the product from use. In other words, a robotic equivalent of the refrigerator deathtrap should not take decades to solve. This is particularly important in rapidly changing fields where startup companies and even their comfortably established competitors may no longer exist in a few years.
(At least in the context of motor vehicles, prospective and retrospective conceptions of risk differ. When the National Highway Traffic Safety Administration mandated electronic stability control in new vehicles, for example, it did not simultaneously consider whether the lack of ESC in older vehicles constituted a defect that posed an unreasonable risk.)
Similarly, commercial sellers should have recourse to what my paper calls private repose–precluding, through technological and contractual tools, the continued use of products that have become functionally obsolete. In contrast to a statute of repose like the General Aviation Revitalization Act of 1994, private repose eliminates liability by actually eliminating the hazard.
Arguing that regulators and developers should be able to permanently retire obsolete products raises two questions related to cost. The first asks what costs and benefits matter in this determination. Replacing physical products can have broad environmental and specific individual costs. What, for example, is the cost of depriving a rural low-income family of a personal car, however substandard?
The second question asks who should bear the costs of product retirement. A vehicle recall under NHTSA’s evolving standard would belatedly impose these costs on manufacturers and, by extension, future customers. The refusal by a state department of motor vehicles to relicense older vehicles would impose these costs most immediately on those vehicles’ end users. Declining to retire these vehicles, however, imposes costs on the people injured in or by them.
In contrast, leases, subscriptions, and service-based models could, by extending manufacturer control and responsibility, facilitate the close and continuing management of risks and recovery of costs. Under this regime, commercial sellers would retain the legal and technological ability to update or even disable their products. As my paper argues, this control could lead to greater obligations in tort. It could also be complemented by public requirements for insurance, bonding, and takeback.
While this contract-, tort-, and regulation-based approach to product retirement would still raise concerns about equity, privacy, and market entry (as well as security), it could reduce the uncertainty involved in the initial deployment of these new products. In short, rather than worry about the next forty years, perhaps regulators and developers can deal with some risks a year or two at a time.