Learning by Negligence - Torts, Experimentation, and the Value of Information

How should tort law deal with agents that employ novel and imperfectly understood technologies that later turn out to involve harm? There is no agreement among different legal systems whether strict liability or negligence rules should govern these so-called 'development risks'. The law-and- economics literature, however, has predominantly favored strict liability. The present paper shows that the choice depends on the characterization of how society learns about technology risks. When experiential public data is an irreducible input into learning, theory justifies the use of specific negligence rules in order to govern development risks. We reconcile the existence of the negligence doctrine for development risks with the theoretical literature using a simple two-period unilateral care model. There, an optimally designed negligence rule can provide a better balancing of benefits, harm to third parties, costs of care effort, and the value of information from learning than strict liability. If feasible, the optimal negligence rule partitions the population of potential users into two groups. Only the high benefit group engages in the risky activity, subject to due care levels designed to deter the low benefit group.