An analysis of the Word Adjacency Network method—Part 1—The evidence of its unsoundness.

This article critically examines the Word Adjacency Network (WAN) method, a mathematical technique for authorship attribution that gained attention for claiming Christopher Marlowe co-authored scenes in Shakespeare’s Henry VI trilogy. While the method shows high accuracy (around 89–94%) in attributing complete sole-authored early modern plays to their known authors, the author’s replication tests reveal it performs poorly on attributing individual scenes, especially when distinguishing between Marlowe and Shakespeare, with success rates only slightly better than chance. The article identifies a fundamental flaw in the WAN method’s use of relative entropy calculations, which exclude large amounts of relevant evidence due to mathematical constraints, leading to unreliable and often accidental attributions at the scene level. Consequently, the article concludes that the WAN method’s scene-level attributions, including those implicating Marlowe in Henry VI, lack sufficient evidential basis and should be treated with skepticism.