Satirical arXiv paper 'Supercazzola' skewers LLM-assisted academic writing

preslav_nakov · x · 2026-11-16

NLP researcher Preslav Nakov posted a satirical arXiv paper, "Supercazzola: On the Use of LLMs in Research" (cs.CL, comments: "Satire. Results pending; significance confirmed."). The abstract mocks LLM-era academic writing with faux-technical claims: models "pre-validate the discursive plausibility of results"; "recursive citation infolding" achieves "a statistically significant increase in significance, independently of the underlying statistics." Hallucinations are reclassified as "anticipatory references," originality becomes "controlled redistribution of previously published novelty," and the human stays in the loop "principally to click regenerate until the limitations become contributions" — establishing "evidence-adjacent scholarship" where the paper is publication-ready before the research is. Kicker: "Further work is needed, preferably by someone else."

Related event: NLP Researcher Mocks LLM Academic Prose as "Supercazzola"(4 posts)→

Original post →

More from Research

Research channel →