ELSC Seminar Series

Effective storytelling relies on a delicate balance between meeting the reader’s prior expectations and introducing unexpected developments. In the domain of detective fiction, this tension is known as fair play, which includes the implicit agreement between the writer and the reader as to the possibility and challenge of identifying the culprit through the story’s clues. I will present a probabilistic framework that aims to quantify this elusive notion. Our definitions lead to an inherent tension between the coherence of the story, which measures how much the resolution “makes sense” in explaining the clues, and the surprise it induces. Due to this tension, balancing these qualities is challenging. We operationalize the proposed framework, defining an approximate computable metric of fair play. We conduct simulations to validate this approach. Our simulations include both real stories and LLM-generated ones (the latter allowing maximal experimental control). Our results present similar trends to the ones predicted by the framework. I will conclude by discussing some potential implications to the broader tension between coherence and surprise.

This is joint work with Eitan Wagner and Renana Keydar.