Controlling Tomorrow: Explaining Anticipatory Restrictions on Emerging Technologies

Abstract

Arms control predominantly concerns technologies that at least one side already has in-hand, not what either side could one day develop. The easy explanation is that states cannot anticipate all potential technological harms, but this explanation is incomplete. If anticipatory control is possible at all, why do states prefer to control today tomorrow instead of controlling tomorrow today? I argue that states rely on technology secrecy to prevent illicit spillover, but secrecy also hinders assessments about relative progress. By the time first-movers learn about fast-followers, most technologies are already emerging or deployed. I formalize this logic with an endogenous learning model, showing how first-movers become ensnared in technology competitions that they cannot win. Anticipatory control is politically attractive only for unusually prescient states who have reason to suspect their initial lead will be short-lived. The theory is validated with new research on two little-known cases—weather weapons and deep sea launch systems—plus original data on anticipatory agreements since 1850. By highlighting conditions under which tomorrow’s technologies are subjected to anticipatory control, the findings contribute to debates on the governance of emerging technologies, and may offer insights into the ongoing race for artificial super intelligence (ASI).

Keywords: technology, bargaining, decisionmaking, arms races, anticipatory arms control, technological surprise, technology competition, risk, regulation