Trade Secrets, Good Fences, and Big Data

The "fencing costs" theory of trade secrecy posits that in the absence of legal protection, holders of confidential information will overinvest in secrecy precautions relative to the social optimum. Trade secret law enhances economic efficiency--or so the theory goes--by providing a lower-cost substitute for the excessively high "fences" that firms would otherwise construct around their secrets. The goal of trade secret law, according to the theory, is to assure firms that they can rely on legal remedies rather than erecting expensive barriers to protect their confidential information.&nbsp;<br><br>While the rise of big data and the advent of artificial intelligence magnify the economic significance of trade secrecy, these same phenomena also challenge the underlying premise of the fencing costs theory: that the sole function of fences is to enable firms to capture the value of the information that those firms produce. In an algorithmic age, trade secret misappropriation significantly increases the risk that malicious actors will be able to steal identities, manipulate consumer preferences, interfere in elections, attack critical infrastructure, and inflict a range of other harms on third parties. Standard economic theory suggests that secret holders are likely to underinvest in fencing relative to the social optimum because they internalize only a small portion of these harms. Insofar as legal protection displaces private fences that might have prevented these third-party harms, it potentially exacerbates the problem of underinvestment in secrecy precautions.&nbsp;<br><br>The idea of "good fences"--secrecy precautions that protect third parties from the harms of trade secret misappropriation--turns the fencing costs theory upside down. The prevalence of third-party harms potentially undermines one important normative justification for trade secret law, but it also potentially rationalizes one otherwise-puzzling feature of trade secret law: the requirement that secret holders invest in "reasonable precautions" as a precondition for legal protection. The "reasonable precautions" requirement incentivizes secret holders to invest in fences that protect third parties from harm even when policymakers lack the ability to specify--ex ante--which secrets require protection and what forms those protections should take. Our analysis of fencing costs and reasonable precautions in the shadow of big data and AI thus shows how new technological developments-in trade secret law, as in other areas-recast and invert decades-old theoretical and doctrinal debates.