Nicholas Polson posts 258 SSRN papers in 2026, apparently AI-assisted

A blog post on a statistics site relays a count by Jeremy Horpedahl: by his tally from Nicholas Polson's SSRN page, Polson has already written 258 working papers in 2026, including six posted on August 26, 2026 alone. Horpedahl says most are of normal academic length (he cites 32, 27 and 58 pages) and that the productivity of Polson and his co-author Sokolov is obviously aided by AI. He adds that any single paper feels like a normal academic paper, the kind an academic might work on for a few months.

The blog author wanted to test that judgment and opened one paper that looked interesting: "Theories of Human Connection", which runs 80 pages. Its subtitle promises an interdisciplinary synthesis across economics, psychology, biology, philosophy, game theory and spiritual tradition. The abstract says the book synthesises fourteen thinkers, among them Morris, Bowlby, Becker, Frankl, Von Neumann, Bateson, Gottman and the Hindu philosophical tradition. It claims that Gottman's laboratory models predict separation with over ninety percent accuracy. The preface lists sixteen names, including Polson himself, as thinkers who "did not, for the most part, know each other's work".

The blog author was not impressed. Gottman, they write, is the man who conned Malcolm Gladwell and various media outlets into believing he could predict divorces with 94% accuracy, and the paper devotes a whole chapter (chapter 10) to that material. The author wonders what prompts were used and guesses they did not include "Evaluate implausible claims skeptically." The paper also has a Chapter 13, "The Cumulative Model: A Unified Theory of Human Connection", and a "Master Equation" on page 75. The author concludes it is not the sort of thing a non-bullshitting academic might write, while noting that Polson is a statistician who has done lots of solid work over the years.

The post then weighs four explanations. First, an experiment or a joke: the author finds this unlikely because a joke would carry internal clues and an experiment would produce papers that read like straight statistics. Second, someone is impersonating Polson, possibly with a chatbot told to pick the name of a successful academic and flood the internet with papers attributed to him. Third, "intellectual squatting", a conjecture from Michael Makowsky. Makowsky's charitable version is that half-baked papers stake a property right to ideas earlier in the research process than was possible before; the less generous version is dumping haystacks on the plains and claiming the needles probabilistically. He imagines someone spending an hour on prompts per idea and letting an LLM fabricate a full paper. The blog author is unsure what Polson would gain from this. Fourth, education, which the author finds the most plausible: Polson wants to reach a broader audience, picks 258 topics, writes prompts for each, and has a bot write the papers and post them straight to SSRN. That would explain how the 80-page paper got there, since nobody can be expected to write or read 258 papers of that length in under a year. The author admits this is a guess.

On value, the author has looked at only one of the 258 papers and guesses the net value is negative, because the time people have wasted going through them outweighs any positive contribution. The suggestion to Polson: write, without a chatbot, one article explaining why the 258 ideas matter, in the spirit of a paper the author co-wrote, "What are the most important statistical ideas of the past 50 years?". The author also recommends withdrawing all the papers from SSRN, and jokes he could program a bot to do the job.

Two postscripts add detail. The first notes there are two accounts for Nicholas or Nick Polson at the University of Chicago business school, which the author says fits either the "social experiment" or the "impersonation" hypothesis. The second reports that SSRN lists papers in time order. The page linked from Polson's website shows 16 papers: the first, "The Impact of Jumps in Volatility and Returns", was posted on 1 Jan 2001, and the latest, "Deep Learning in Characteristics-Sorted Factor Models", was posted on 23 Sep 2018 and last revised 26 Jun 2023. The other page, with the papers the author calls fake, starts with "Kramnik vs Nakamura or Bayes vs p-value" (7 Dec 2023), then a 25-page "AI and Vivekananda" (5 Mar 2024), then a two-year gap until another AI-related paper on 9 Mar 2026, and "Theories of Human Connection" on 11 Mar 2026. So the two pages do not overlap in time. Polson also has arXiv papers, including "Bayes with No Shame: Admissibility Geometries of Predictive Inference", dated 24 Aug 2026, which the author could not find on SSRN and which looks far more substantive, perhaps a sign of how convincing chatbots have become.

Key facts

• Jeremy Horpedahl counted 258 working papers on Nicholas Polson's SSRN page for 2026, six of them posted on August 26, 2026.
• Horpedahl says the output is obviously aided by AI; the post names no specific AI tool or method.
• The blog author read one paper, the 80-page "Theories of Human Connection", and called it bogus, pointing to a chapter built on Gottman's disputed 94% divorce-prediction claim.
• Four explanations are weighed: experiment or joke, impersonation, "intellectual squatting" (Michael Makowsky) and education; the author favours education and admits it is a guess.
• The author recommends withdrawing all the papers from SSRN and guesses their net value is negative, having looked at only one.

Why it matters

Preprint servers accept whatever is uploaded, so one account can now produce a paper every day or so at normal academic length. The post shows what that looks like in practice: 258 papers by August 2026, six in a single day, under the name of a statistician the blog author calls a producer of lots of solid work. Reading one paper is enough for the author to doubt the whole batch. The open question is not only whether the papers are good but who is actually behind them.

Who it affects

Readers and researchers who search SSRN and may waste time on papers of doubtful quality; the author says people, including themselves, have already lost time this way. It also affects Polson, whose name is on the papers and who is not quoted in the post. Finally, it touches SSRN and the University of Chicago business school, where the author says two accounts exist under Polson's name, though the post reports no action by either.

How to use it

The post offers a practical habit rather than a tool: check a prolific author's SSRN listing in time order, as the author did, and read at least one paper closely before trusting a bulk output. The author also searched the document for a name that rang a bell (Gottman) and found a whole chapter resting on a claim they dispute. Papers that seem out of pattern with an author's earlier work, such as the 2001 to 2018 page versus the 2023 onward page, are a signal to look harder.

How solid is it

The 258 figure is Horpedahl's own count from SSRN, and the post itself once writes 256 elsewhere. The AI involvement is inferred by Horpedahl and the blog author; no tool or model is named. The author read only one of the 258 papers, and that is the basis for the harsh verdict. Who or what produced the papers is not established: impersonation, experiment, squatting and education are all hypotheses, and the author says the whole thing remains a mystery. The bot theory is explicitly a guess.

Risks and caveats

Judging 258 papers by one sample is thin, and the author concedes it. The Gottman critique is the blog author's opinion, while the paper's abstract states the over ninety percent figure as fact. The author also says the arXiv paper from 24 Aug 2026 looks much more substantive, so quality may vary. Attributing the papers to Polson himself is uncertain given the impersonation possibility, so the post's suggestions to him assume he is the author. Polson is not quoted, and no response from him appears.

“My guess is that the net value of the papers is negative, in that the amount of time that people (including me) have wasted going through them outweighs any positive contributions that might have been there.”

— The blog author, on the 258 papers