When does a difference in degree become a difference in kind? Answers to this perennial techlaw question inevitably include mentions of “scale”—but what exactly does “scale” mean?
In Taking Scale Seriously in Technology Law, Mark McKenna and Woodrow Hartzog clarify and complicate the concept of “scale.” Certainly, “scale” can simply mean more of something (what the authors refer to as “scale as more”). Social media content moderation is difficult both because of the complexity of the analysis and because of the sheer number of posts to evaluate. But more of something might not just result in more of the associated benefits or harms. One of the article’s main takeaways is that, in some situations, more of something can change the nature of the benefits or harms (“scale as different”). If we are sensitive to these distinctions—let’s call them different kinds of differences in kind—we can better identify harms that might otherwise be ignored and design more effective regulations.
“Scale as more” is a familiar concept in law. Minor harms are unlikely to spur regulation; minor harms at scale, however, are more likely to be regulated. Scale can show up in considering what precautions were reasonable in a negligence analysis, in certifying class actions, in determining when punitive damages are warranted in tort or when stricter sentences are warranted for crimes, as well as more broadly in evaluating whether recognizing certain rights would “open the floodgates” and overload courts.
Shifting to “scale as different,” the authors note that sometimes more of an activity can result in “entirely new effects along different dimensions.” (P. 397.) For example, when a high-enough percentage of a population is vaccinated, it achieves herd immunity. A disease is effectively prevented from spreading, creating a benefit—protection for those who cannot be vaccinated for health reasons—that isn’t achievable without scale. On the flip side, if herd immunity is not attained, a disease may continue to circulate and mutate, creating problems that wouldn’t otherwise exist.
The authors identify four ways in which scale being different could affect a legal analysis or decision. First, the affected population(s) could change. At certain magnitudes, data about individuals permits population-level analysis, enabling population-level insights that could be used to affect individuals outside of the dataset. Second, scale can create problems that didn’t exist at lesser levels of activity. One sidewalk robot may be a nuisance; fifty makes a sidewalk unusable. (As a professional contrarian, I am obliged to note the flip side: As highlighted twice in the vaccine example, scale can also create benefits that didn’t exist at lesser levels of activity, which may provide support for regulatory action that supports scaling an activity.) Third, scale can alter assumptions about the costs and benefits of an activity. One person writing down online data isn’t a threat to privacy by obscurity; multiple companies deploying scraping tools that can scour and aggregate data across the internet are. Fourth, scale can affect the potential set of effective policy solutions: it can make “certain institutional designs more effective and take[] some legal, social, design, and market-based remedies and strategies entirely off the table.” (P. 400.) For example, “at scale, we might see that certain remedies that sound good in isolation, like data access and deletion rights in private law . . . are not nearly as effective . . . as infrastructural or institutional design remedies.” (P. 422.)
The authors identify three ways in which misunderstandings of scale’s effects result in bad regulation: (1) a failure to recognize effects that occur only at scale, (2) resulting inappropriate framings of the problems that need to be addressed; and (3) resulting misidentifications of the appropriate regulators and appropriate timing of regulatory action.
As a correction, they recommend that policymakers and scholars “start with scale”: assume a technology will proliferate, then attempt to imagine both the immediate and broader social implications. There are some normative effects of this reframing: “starting with scale” will generally favor a more precautionary approach and building in “technological sunrises” and regulatory re-evaluations.
In addition to the benefits the authors note, their analysis also impacts “difference in degree” versus “difference in kind” debates. When a “difference in degree” becomes a “difference in kind” is a fundamental techlaw question, not least because the answer results in a powerful rhetorical argument for or against legal change. As BJ Ard and I have observed, “[c]haracterizing an innovation or its impacts as being only a difference in degree suggests that existing law can handle, or at least muddle through, the unanticipated scenario. Meanwhile, a difference in kind suggests a momentous and unforeseen development, with unanticipated social impacts, which implies the need for legal action, possibly even the creation of new law.” But a technology’s social impacts rarely obviously result in a difference in degree or difference in kind—instead, these claims usually reflect the evaluator’s concerns or desired regulatory stance. Those promoting a more precautionary or proactive approach will favor difference in kind assessments; those preferring a more permissive, laissez-faire approach will argue the social shift is merely one of degree.
McKenna and Hartzog’s four scale-as-different categories provide four ways of arguing that a technological change is a difference in kind. While it’s tempting to link “scale as more” to “difference in degree” and “scale as different” to “difference in kind,” the relationship is more nuanced. The harm of most import to content moderators, for example, is arguably just the increase in number of posts, suggesting this is a scale-as-more issue and thus a difference in degree. But if the sheer volume of posts makes human moderation impossible, perhaps it is better understood as a scale-as-different situation, supporting an argument that there is a difference in kind that justifies new regulation.
Overall, McKenna and Hartzog’s contribution is a welcome exploration of “scale” that will enable both more comprehensive and appropriate regulatory actions and more persuasive supporting rhetorical arguments.
