Claude Code 27: Research and Publishing Are Now Two Different Things

  • Author/Source: Scott Cunningham (Baylor University), via Substack ("Causal Inference")

  • Original: https://causalinf.substack.com/p/claude-code-27-research-and-publishing

  • Key Ideas

  • AI collapses the marginal cost of producing a submission-quality manuscript to near zero, triggering a massive supply-side shock to academic publishing.
  • With roughly 3,800 publication slots across 87 economics journals and a fixed referee pool, a 5x increase in submissions mechanically crashes acceptance rates and overwhelms editors.
  • The resulting arms race is a prisoner's dilemma: individually rational to scale volume, collectively destructive for nearly everyone.
  • Project APE (University of Zurich) demonstrates fully automated economics papers that already win 4.7-7.6% of head-to-head matchups against AER-quality articles.

  • Journal revenue from submission fees could rise dramatically (from ~$6.2M to ~$31M across 87 journals), but the editorial and refereeing system will break long before journals can adapt.

  • Summary

Cunningham presents what he calls "Claude Code fan fiction" -- a supply-and-demand analysis of academic publishing when AI agents radically reduce the cost of producing manuscripts. He draws an analogy to romance novelist Coral Hart, who went from 10-20 novels per year to 200+ using ChatGPT, and to the Reimers and Waldfogel finding that new book titles on Amazon tripled post-ChatGPT while average quality fell. He argues the same dynamic is coming to economics, where roughly 12,000 research-active economists currently generate about 39,000 submissions per year for 3,800 slots.

The core economic argument is that when each economist can go from 3 to 10 submissions per year at a total incremental cost of roughly $3,200 (submission fees plus a Claude subscription), the expected value calculation favors mass production. But since publication slots are fixed in the short run, acceptance rates must fall -- from 5% to 1% or lower at the top 5. The referee pool cannot scale with submissions, forcing desk rejection rates to rise from 50% toward 90%, making editorial decisions noisier and more dependent on heuristics like pedigree and institutional affiliation rather than manuscript quality.

Cunningham argues the binding constraint on science is shifting from production to evaluation. He predicts noticeable disruptions within months, not years, and notes that the "paper mill signature" -- dozens of working papers on a researcher's website -- may itself become a negative signal that markets will price accordingly.

  • Relevance to Economics Research

This article directly addresses how AI agents will reshape the economics publishing pipeline. It is essential reading for anyone thinking about the future of peer review, journal economics, and the incentive structures around academic careers. The supply-demand framework and back-of-the-envelope calculations provide a concrete foundation for understanding the institutional disruption ahead.