REVIEW 3 major objections 5 minor 1 references
How to Fight Fraudulent Publishing in the Mathematical Sciences: Joint Recommendations of the IMU and the ICIAM
T0 review · 3 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash
Pith's one-line read The paper asserts that predatory journals and citation cartels are a reaction to bibliometric ranking, and that replacing bibliometrics with expert-led assessment is the way to fight fraudulent publishing.
desk verdict A clear, well-intentioned institutional policy statement; the joint IMU/ICIAM endorsement is new, but the evidence for its core causal claim is borrowed from a companion paper, so treat it as a call to action rather than a research contribution. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The key mechanism is an incentive chain: bibliometric ranking systems create pressure to maximize countable outputs, and that pressure is the soil in which predatory journals and citation cartels grow. The proposed counter-mechanism is expert-led qualitative assessment, replacing numbers such as journal rankings and citation counts with careful reading of a researcher's best papers by informed evaluators. Central to the argument is the claim that, in mathematics, citation counts are low and therefore unusually vulnerable to manipulation, which is what makes the bibliometric regime both untenable and uniquely damaging in the field.
What would settle it
Compare the share of mathematics papers published in known predatory venues by authors at institutions with strict quantitative publication requirements against those at institutions that ignore bibliometrics in hiring and promotion; if the shares are similar, the claim that metrics drive predatory publishing would be undermined.
Extended reading notes
Core claim
The paper's central assertion is that the rise of predatory journals and citation cartels is a direct reaction to the drive to exactly quantify and rank research quality through bibliometric performance indicators. In the authors' account, the pressure to publish and the competitive job market push even serious scientists to improve their numbers artificially, and predatory publishers turn that desire for measurable output into a commercial scheme. The remedy follows from the diagnosis: instead of refining metrics, research evaluation should be based on expert-led assessment, with hiring, promotion, and funding decided by reading researchers' actual best work. The paper also takes the positi
Load-bearing premise
The load-bearing assumption is that bibliometric incentives are a primary cause of fraudulent publishing and that reducing reliance on bibliometrics will proportionately reduce fraud; the paper states this as its central diagnosis but defends the evidence in a separate companion article.
Editorial extensions
If this is right
- Funding bodies and institutions that adopt expert-led assessment reduce the financial and career rewards of inflating publication counts, shrinking the market for fake venues.
- Abolishing mandates to publish a fixed number of papers for PhDs and promotion removes a concrete incentive that drives researchers toward predatory outlets.
- If bibliometric rankings are no longer used in decisions, individual researchers can cite only what is relevant and can report questionable venues without fear of losing out on quantitative comparisons.
- Community-supported screening and transparency tools make it harder for fraudulent journals to hide, buy credibility through fake editorial boards, or profit before being exposed.
- The recommendations imply that good publishing practices must be explicitly defined and taught, so that young researchers do not accidentally enter the predatory ecosystem.
Reading between the lines
- Beyond the paper: if bibliometric pressure is the driver, then fraud should migrate to whatever quantitative proxy an institution uses next, so the reform must be qualitative, not just metric-swapping.
- Beyond the paper: the causal claim could be tested by comparing submission rates to known predatory venues across institutions or countries with and without strict quantitative publication requirements.
- Beyond the paper: the recommendation to use expert-led assessment implicitly depends on a robust pool of willing referees and editors; an underserved consequence is that funding and recognition for reviewing work may be needed to keep that pool alive.
- Beyond the paper: the paper's focus on mathematics generalizes to other low-citation fields, where the same vulnerability to bibliometric gaming likely applies.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. This short six-page document presents joint recommendations from the IMU and ICIAM for countering predatory publishing and citation manipulation. After stating in §1 that these phenomena are reactions to bibliometric performance indicators, it makes recommendations for policymakers (§2.1), institutions (§2.2), and individuals (§2.3), and supplies a curated set of internet resources (§3). The paper is explicitly a companion to a longer investigation [1], to which all supporting analysis and references are deferred.
Significance. If the stated causal diagnosis is correct, this is an important high-level consensus statement from two major international mathematics organizations. Its strengths are clarity, the concrete resource list, and the emphasis on expert-led assessment, best-paper evaluation, and individual responsibility; the alignment with DORA/COPE-type principles is sensible. As a scholarly document, however, it contains no new data, no systematic review of evidence, and no test of whether the proposed measures reduce fraud; its central causal claim is deferred to [1]. The document's practical value depends strongly on that external source.
major comments (3)
- [§1, §2.1.2] The paper's central assertion is that 'predatory journals and citation cartels are reactions to the effort to exactly quantify and rank the quality of research through scientific performance indicators' (§1). This is a strong causal claim about the etiology of a complex phenomenon. It is load-bearing: the recommendation to replace bibliometrics with expert-led assessment and to discourage university rankings (§2.1.2) presupposes that bibliometrics are a, or the, primary driver. The paper gives no evidence for this direction of causation and explicitly defers to [1] for details and references. As it stands, a reader cannot verify the premise. The authors should either incorporate a compact evidence summary (e.g., describing available case studies, correlations, and natural experiments) or reframe the passage as a hypothesis/consensus position rather than an established causal relation. Wi
- [§2.2.2] The bullet 'Educate researchers about the low correlation between the quality of research and bibliometrics such as journal impact factors and citations' is an empirical assertion. No source is given, and the statement is contested (e.g., citation counts correlate with some measures of research impact, though certainly not perfectly). At minimum, cite the literature or qualify with 'in mathematics' and 'as commonly measured.' This matters because the recommendation is used to justify concrete policy changes in hiring and promotion.
- [§2.1.2, §3.3] Operationalization of 'predatory' is unresolved. §2.1.2 recommends 'endorse good journals and discourage publishing in predatory journals,' but §3.3 says it is 'virtually impossible to compile a complete list' of such journals and gives several lists with caveats. Without a working definition or triage procedure, institutional implementation is difficult. The paper could propose minimal criteria (e.g., presence in zbMATH/MathSciNet with verification of peer review, COPE membership, transparent APC policies) and note exceptions. This is not a fatal flaw, but for a recommendation document aimed at policy makers, it needs addressing.
minor comments (5)
- [§2.3.4] Typographical errors: 'accross' should be 'across' and 'fundserious' should be 'fund serious'.
- [§3.5] In 'keep away of OA journals that do not have the seal,' 'of' should be 'from.' Also 'OA' is not expanded on first use.
- [§2.1.2, §2.2.2] Some abbreviations (SJR, JCR, APC, ORCID) are not expanded at first use. Since the intended audience includes policy makers, a short glossary or expanded first use would help.
- [§2.3.3] The first bullet says scientific phishing emails are 'not to trick the recipient into revealing sensitive information' but to draft them into fake science. The word 'not' makes this too absolute; 'are often not primarily intended' would be more accurate.
- [References] Reference [1] lists both a print version (Notices of the AMS) and a digital version (arXiv). If the two versions differ in the list of references, the authors should clarify which version supports the recommendations in this note.
Circularity Check
No circularity: the paper is a policy/editorial statement whose recommendations are not derived from themselves or from fitted quantities; reliance on companion paper [1] is external evidence, not a circular reduction.
full rationale
This is a recommendations/position paper, not a derivation. There is no equation, fitted parameter, or formal model whose output is equivalent to its input. The opening causal claim—'Predatory journals and citation cartels are reactions to the effort to exactly quantify and rank the quality of research through scientific “performance indicators”'—is an asserted empirical/sociological premise. The subsequent recommendation to rely on expert-led assessment rather than bibliometrics follows from that premise, but it does not assume the truth of the recommendation itself; it is an ordinary policy argument. The paper refers to the authors' companion article [1] for 'a detailed description of fraudulent publishing... and an extensive list of references,' but this is a self-citation for background evidence, not a self-referential proof: the recommendations would stand or fall on the adequacy of that evidence, and the citation is not used to forbid alternatives or as a uniqueness theorem. The unsupported assertion about 'the low correlation between the quality of research and bibliometrics' (Section 2.2.2) is a weakness of empirical support, not circularity, since it is not derived from the conclusion. No quantity is defined in terms of another quantity being predicted; no fitted parameter is renamed as a prediction; no known result is repackaged as new. The document is self-contained in the sense that its recommendations are not circular, even though their evidential basis is partly external.
Assumptions & free parameters
assumptions (2)
- domain assumption Bibliometric metrics are a major cause of fraudulent publishing.
- domain assumption Expert-led research assessment is more reliable than bibliometric measures.
Cite this review
Pith. "Pith review of How to Fight Fraudulent Publishing in the Mathematical Sciences: Joint Recommendations of the IMU and the ICIAM." pith.science (2026). https://pith.science/paper/JS7RAD3B
@misc{pith2026250909877,
author = {Pith},
title = {Pith review of: How to Fight Fraudulent Publishing in the Mathematical Sciences: Joint Recommendations of the IMU and the ICIAM},
year = {2026},
howpublished = {\url{https://pith.science/paper/JS7RAD3B}},
note = {Machine review of arXiv:2509.09877}
}
read the original abstract
These recommendations were formulated by the authors in close collaboration with the IMU Committee on Publishing (chaired by Ilka Agricola) and have been endorsed by the Executive Committee of the IMU and the Board of the ICIAM in May/June 2025.
Reference graph
Works this paper leans on
-
[1]
[1] Ilka Agricola, Lynn Heller, Wil Schilders, Moritz Schubotz, Peter Taylor, Luis Vega,Fraudu- lent Publishing in the Mathematical Sciences. Print version with partial references: Notices of the AMS, October 2025,https://www.ams.org/cgi-bin/notices/nxgnotices.pl?fm=main& current=202509. Digital version with all references: arXiv, math.HO, September 2025,...
arXiv 2025
Reviewed August 4, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.