Most authorship policy — CRediT, the ICMJE four criteria, COPE’s guest/gift/ghost framework — assumes a project has a known, bounded set of named individuals whose relative contribution can be assessed and recorded. The Polymath Project broke that assumption on purpose. Started by mathematician Timothy Gowers in January 2009 as an open, blog-based experiment in “massively collaborative mathematics,” it produced published papers with dozens of anonymous or pseudonymous contributors credited under a single collective byline: D.H.J. Polymath. This page explains what the Polymath Project actually was, how it decided who counted toward that byline, and what the model does and doesn’t offer as a reference point for CASRAI’s broader authorship-criteria content.
What the Polymath Project is
The Polymath Project began with a blog post by Timothy Gowers, then a Cambridge mathematician, titled “Is massively collaborative mathematics possible?” (January 2009). Gowers proposed working on a hard, unsolved mathematics problem entirely in public, in the comment section of his own blog, open to anyone willing to contribute — professional mathematicians, graduate students, and interested amateurs alike, with contributions ranging from a fully worked partial proof to a single corrective observation. Terence Tao ran a parallel discussion thread on his own blog, and a public wiki (later consolidated at polymathprojects.org) was used to organize partial results, terminology, and a running summary of where the collective proof attempt stood.
The first problem tackled this way, retroactively named Polymath1, sought a new combinatorial proof of the density Hales–Jewett theorem, a result in Ramsey theory. The collaboration succeeded: it produced a genuinely new, more elementary proof with an explicit quantitative bound, later published as “A New Proof of the Density Hales-Jewett Theorem” in Annals of Mathematics, vol. 175, no. 3 (2012), pp. 1283–1327. Further numbered Polymath projects followed on other open problems, run with the same open-comment-thread format, of varying scale and success. The best-known later project, Polymath8, took up bounding gaps between consecutive prime numbers immediately after Yitang Zhang’s 2013 breakthrough proof of a finite bound; Polymath8a tightened Zhang’s bound to 4,680, and Polymath8b, folding in a separate improvement by James Maynard, brought it down to 246.
The D.H.J. Polymath pseudonym
Rather than list some or all participants as individual authors, papers resulting from Polymath projects are typically published under a single collective byline: D.H.J. Polymath, initials taken from the density Hales–Jewett theorem that Polymath1 first tackled. The name stuck as a general project pseudonym and has been used across multiple subsequent Polymath papers, not only the one that originated it — so “D.H.J. Polymath” functions less like a literal acronym describing each paper’s subject and more like a standing collective identity, comparable in function (though not in mechanism) to a large collaboration’s group name on a physics paper.
This is a genuinely unusual choice in a scholarly-publishing landscape built around individual bylines. The project’s own published rules for how a paper gets written state the reasoning directly, and it is worth quoting precisely because it is the actual stated rationale, not an inference: the General Polymath Rules page explains that once a problem is judged solved, “the paper will be written collaboratively (using the wiki and blog as appropriate), authored under a polymath pseudonym,” because “the question of assessing how much of a contribution each participant had to a project seems impossible to answer with any degree of objectivity, so the pseudonym approach to authorship is the simplest solution.”
How authorship was actually decided
This is the part search intent around this topic usually gets wrong: the Polymath model is not a threshold-based individual authorship system where contributors accumulate points and cross a line onto a named byline. It is closer to the opposite — a decision to route around the problem of individually assessing contribution altogether, by publishing under a shared pseudonym instead of a named list. Two mechanisms operate side by side, and it matters not to conflate them:
- Formal authorship of record is the collective pseudonym itself. The paper’s author, as it appears in the journal and in citation databases, is “D.H.J. Polymath” — a single entity, not a list of names. No individual assessment of “does this person meet the bar for co-authorship” happens at the point of submission, because there is no individual byline to populate.
- Attribution and recognition happen separately, and informally. Every substantive contribution — a blog comment, a wiki edit, a partial argument, a counterexample — is public and timestamped by the platform itself, which the project’s own rules note establishes precedence of ideas even without a formal authorship credit attached to it. Mathematicians who have written about participating in Polymath projects describe a companion practice on top of this: a wiki page maintained alongside the project listing which individuals contributed at roughly the level expected of a co-author, compiled informally and relied on as an honor-system record — used, for instance, by a contributor wanting to cite the work on a CV — even though it carries no formal weight with the journal or publisher the way an ICMJE-qualifying byline would.
Critically, there was no single, codified, cross-project rule that fixed this in advance. Accounts from mathematicians involved describe it as a consensus decision made freshly for each paper rather than a standing editorial policy — which is consistent with the wide variation in scale across Polymath projects: Polymath1 alone drew contributions from more than 40 participants over the course of the collaboration, with a small number of especially active contributors, prominently including Gowers and Tao, producing a large share of the content, while most other participants contributed material that meaningfully shaped the final proof without individually producing most of it. A model built around a fixed numeric contribution threshold would have had to adjudicate every one of those 40-plus cases individually; the pseudonym model simply did not require that adjudication to happen at all in order to publish.
Why this is a genuine alternative model, not just a variant of “many authors”
It’s tempting to file Polymath alongside other large, many-contributor collaborations and assume the underlying logic is the same, just at a different scale. It isn’t. CASRAI’s guide to authorship in high-energy-physics megacollaborations such as ATLAS, CMS, and LIGO covers a structurally different answer to the same “too many contributors to assess individually” problem: those collaborations still publish a full, individually named author list — sometimes thousands of names — typically in a fixed order (often alphabetical) that deliberately avoids ranking contribution, but every name on it is still an individually accountable, ICMJE-adjacent author of record. Polymath goes further: it removes the individual byline from the paper entirely and substitutes a collective identity, with individual recognition demoted to an informal, non-binding side record.
Put against CASRAI’s own reference frameworks, the contrast is precise. The ICMJE’s four authorship criteria require that everyone credited as an author be individually named, individually accountable, and able to defend the work publicly — a structure built around the idea that a byline is a personal, professional claim of responsibility. The CRediT taxonomy (see CASRAI’s overview of how CRediT applies, and doesn’t apply cleanly, to non-traditional contributors) goes further still, recording not just that someone contributed but specifically what they did, role by role. Polymath’s pseudonym approach satisfies neither: no individual is named as accountable on the paper, and no role-level contribution record is attached to the formal publication at all. That is a deliberate trade, not an oversight — the project’s own stated reasoning is that assessing relative contribution with the rigor either of those systems implies was not achievable at the honor-system, open-comment-thread scale the project operated at, so it opted out of trying rather than doing it badly.
The nearest points of comparison elsewhere on this site are cases where a project’s scale or structure similarly strains conventional, individually-accountable authorship: see CASRAI’s coverage of how the Journal of Open Source Software decides software-paper authorship and of citizen science more generally, where volunteer, non-professional contributors routinely complicate a conventional byline in ways closer in spirit, if not mechanism, to Polymath’s own honor-system attribution page.
What this means if you’re weighing it as a model for your own project
Polymath’s approach worked for the specific conditions it was built for: pure mathematics, where “the work” is a public proof argument that can be independently checked by anyone regardless of who wrote which line, and where the norms of the discipline already tolerated an unusual, low-stakes experiment in attribution. It is a poor fit, and CASRAI does not recommend it, for research contexts where individual accountability is the point rather than an inconvenience — clinical research, human-subjects work, anything subject to funder or institutional authorship-disclosure requirements, or any output where a reader needs to know specifically whom to ask about a specific claim. In those settings, the ICMJE criteria and a proper CRediT contribution statement exist precisely to preserve the accountability that a collective pseudonym deliberately discards. Polymath is best read on this site not as a template to copy but as a genuine edge case that clarifies, by contrast, what conventional authorship criteria are actually for.
Frequently asked questions
Is D.H.J. Polymath a real, registered author name?
It functions as one for publication and citation purposes — journals list it as the paper’s author, and citation databases index papers under it — but it is a collective pseudonym, not a legal or institutional identity, and it does not correspond to any single accountable individual the way a normal byline does.
Did everyone who commented on a Polymath blog thread become a co-author?
No. Casual or minor comments did not confer authorship-equivalent status. Recognition as having contributed at the level expected of a co-author was reserved, informally and by consensus rather than a fixed rule, for participants judged to have made a substantive contribution to the eventual proof — and even then, that recognition lived on an informal companion page, not on the paper’s own byline, which remained the collective pseudonym.
Why didn’t the Polymath Project just use a long list of individual names, like large physics collaborations do?
Its own stated rationale, from the project’s General Polymath Rules, is that assessing each participant’s relative contribution with enough objectivity to justify a specific named list was not achievable given how the project ran — open blog comments from dozens of participants of widely varying involvement — so it used a shared pseudonym instead of attempting that assessment.
Has this model been adopted outside the Polymath Project?
Not widely as a formal authorship policy. It remains a distinctive case study rather than a standard other fields have taken up; most large-scale collaborative research, including other crowdsourced or citizen-science projects, still resolves the same underlying problem through individually named, ordered author lists or through formal contribution-role systems like CRediT rather than a collective pseudonym.
