The Promises of Algorithmic Copyright Enforcement

How to Cite

Husovec, M. (2018). The Promises of Algorithmic Copyright Enforcement: Takedown or Staydown? Which Is Superior? And Why?. The Columbia Journal of Law & The Arts, 42(1), 53–84.


Under the prevailing model of copyright liability for user-generated content, right holders and intermediaries are both involved in the enforcement of exclusive rights on the Internet. While right holders are expected to identify and notify the infringing content that they wish to remove, the intermediaries have to react by assessing the received notices and taking appropriate action, including taking the information “down” from the service in case it is infringing. This “notice and takedown” system, championed by the Digital Millennium Copyright Act, became a model for many countries around the world. However, in the last few years, the right holders have begun to advocate for a fundamental re-design of the system. According to the number of initiatives, some of the right holders would prefer that intermediaries not only take down the notified content but also prevent its re-appearance in the future. This alternative model, often dubbed “notice and staydown,” is currently proposed by the European Commission as part of its upcoming copyright reform. If successful, it will constitute a huge change for the existing global online environment.

This article scrutinizes the potential switch from notice and takedown policy (“NTD”) to notice and staydown policy (“NSD”) in order to answer two important questions: (1) What are the (economic) costs and benefits of two policy options and how do they compare? (2) Is NSD really superior in delivering better tools for automation? The overall goal of the paper is to offer general policy guidance for national or regional policymakers currently considering such policy change.

This article concludes that algorithmic enforcement is inevitable and, under some conditions, socially desirable. First, high-quality automation of copyright enforcement that produces negligible enforcement errors offers many opportunities for improvement of the status quo and therefore should be embraced and incentivized. Second, to make such automation a reality, we need to push innovation in the right direction by conditioning acceptance of algorithmically generated notices upon their quality. Third, an enhanced notice and takedown framework can promote such automation better than notice and staydown. It provides for stronger market incentives for the development of new filtering technologies and allows area-by-area deployment as the technologies improve. Last, as a consequence, enhanced NTD can become a superior policy option from a social perspective. However, in order to realize these benefits, some changes to the NTD framework are required, too. These could take the form of standardized submission formats or interfaces for robo-notices that come with quality conditions and effective sanctions to enforce them.