Using experimental evidence to improve delegated enforcement

Lenka Fiala, Martin Husovec

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

Digital content today is governed by online providers like Facebook or YouTube. Increasingly, these providers are expected to enforce the law by removing illegal content, such as copyright infringement or hate speech. Typically, once they are notified of its existence, they have to assess it and, if infringing, remove it. Otherwise, they face liability. This system of content moderation is a form of delegation of the state's tasks to private parties. In literature, it is empirically established that some schemes of delegated enforcement can trigger substantial false positives, mostly due to over-compliance by providers and under-assertion of rights by affected content creators. This results in a phenomenon known as over-blocking: collateral removal of lawful content. We conduct a laboratory experiment to test a possible solution to this issue, as proposed by Husovec (2016). Our results show that an external dispute resolution mechanism subject to a particular fee structure can significantly reduce over-compliance by providers and improve the accuracy of their decisions, largely thanks to the content creators taking initiative. It does so by re-calibrating the typical asymmetry of incentives under the delegated enforcement schemes. The principles behind the solution have the potential to improve also other schemes of delegated enforcement where providers have weak incentives to properly execute delegated tasks in the public interest.

Original languageEnglish
Article number106079
JournalInternational Review Of Law And Economics
Volume71
DOIs
Publication statusPublished - Sept 2022

Keywords

  • Alternative dispute resolution
  • Content moderation
  • Copyright
  • Lab experiment
  • Notice and takedown
  • Online enforcement

Fingerprint

Dive into the research topics of 'Using experimental evidence to improve delegated enforcement'. Together they form a unique fingerprint.

Cite this