Publication Title

Washington University Law Review


Copyright law lacks a coherent method to determine non-literal infringement. The core inquiry, “substantial similarity,” purports to assess whether two works are so alike that an accused work infringes the original. Substantial similarity is a fundamental limit on the scope of copyright, but it is plagued by confusion and governed by a series of arcane tests that differ in each circuit. Even more troubling, courts lack a consistent method to go about comparing two works and how the comparison between two works is framed. There is no consensus, for example, on whether the original work or the accused work should be used as the baseline when assessing similarity. Courts sometimes adopt the perspective of the original creator, and sometimes of the alleged infringer, in determining whether seemingly copyrightable expression has become an uncopyrightable idea or functional standard. Courts are even confused as to whether dissimilarities or new material added by the defendant have any relevance to the comparison.

This Article seeks to bring analytical clarity to copyright’s similarity analysis, with a focus on these often-implicit framing issues. It argues that how courts frame the comparison, more than the legal test applied, is strongly associated with case outcomes. It urges courts to take a consistent approach to framing issues in similarity analysis so as not to improperly bias the comparison in favor of either party. In particular, courts should adopt a flexible, contextual approach to framing. This method considers both the perspective of the original creator and of the alleged infringer, as relevant, in drawing the line between permissible and substantial copying. It rejects the rigid approach that predominates in the case law, and endeavors to consider all relevant information about what was copied, how it was used in context, and why. The result is a similarity analysis that is not only more consistent, but a robust and vital limitation on the scope of copyright.