What Makes a Creepypasta Stick? On the Recipe for Canonicity of Online Literary Productions

Abstract

The determinants of canon formation are numerous (gatekeeping, cognitive appeal, recombinability, etc.) and hard to disentangle. We therefore work on a simple case for study: online narratives called "creepypastas". The access we have to the work of a vast number of authors, and the relatively scarce intermediation compared to printed literature, make this a quasi-experimental case for testing canon theories. Drawing on a corpus of ca. 25k short-form, user-generated horror ("creepypasta") from Reddit and Fandom (2007-2024), we evaluate three hypotheses: (H1) readable yet stylistically elevated prose raises canonical odds; (H2) uncanny situations embedded in ordinary settings fare better than spectacular monstrosity; (H3) high-arousal affect boosts the virality of a narrative. Stepwise logistic modeling supports H1 and H2 but not H3: canonicity aligns with readable, noun-heavy, descriptively framed narration situated in mundane spaces, while overt high-arousal signals are not decisive once style is accounted for. The results generalize print-era findings on formal economy to platform ecosystems and suggest an online "recipe" for endurance grounded in cognitive ease and affective latency rather than peak arousal. </div>

Affiliated Institutions

• École Nationale des Chartes FR
• École Pratique des Hautes Études FR

Related Publications