Read the whole book free at the MIT Press
What the book establishes
The book's strongest idea is to move the study of research misconduct, meaning cheating in research, away from what is written in a paper and toward the system that puts a value on papers. The editors separate "epistemic" crimes, which corrupt the knowledge itself, from "bureaucratic" crimes, which corrupt the way research is counted and rewarded (p. 3). They also argue that metrics, the numbers such as citation counts used to rate research, make new kinds of cheating possible (p. 1). Together these ideas give a clear frame for cases that fit poorly into the standard definition of misconduct: fabrication (inventing data), falsification (altering data) and plagiarism (copying others' work). In fake peer review, authors arrange to review their own papers. In a citation cartel, a group cites one another to push up their numbers. A purchased affiliation is a researcher paid to name a university as their home. A ghost-managed paper is planned and written by a drug company behind the names of academic authors. All of these can involve sound science and still corrupt how research is judged. The book makes that point convincingly.
Its second contribution is historical. Csiszar shows that the danger of gaming the numbers was named by the sociologist Merton and discussed with Garfield, the inventor of citation indexing, before citation metrics took hold (pp. 33–37). Add Gingras's account of how the paper became a unit for keeping accounts (p. 44) and Power's study of the UK system for scoring the "impact" of research (pp. 57–65), and the case is solid: gaming is part of how metrics work, not an accident.
Its third contribution is to give the watchdogs a voice. Part III records, in their own words, how Retraction Watch, PubPeer and individual critics worked in the 2010s, what they found, and what it cost them. Brookes describes how the Office of Research Integrity (ORI), the US government office that rules on misconduct, disclosed his name (p. 170), and how he faced legal threats (pp. 173–174). This kind of record rarely appears in academic print.
Where the evidence is strong
The strongest chapters are the ones with real data and clear methods. Fanelli's studies are the most rigorous in the book: one compares retracted papers with similar papers that were not retracted, and another follows researchers' output over a long run of years (pp. 114–116). Several other chapters are specific and can be checked: Oransky's figures on retractions (pp. 144–145), Sismondo's observation of drug-industry conferences and the citation studies he reports (pp. 124, 129–130), and Guaspare and Didier's dated step-by-step account of the Voinnet affair, a dispute over manipulated images in a leading scientist's papers (pp. 160–164). Power's figures from the UK's Research Excellence Framework (REF), its national rating of university research (pp. 57–58), and Kehm's cases (pp. 93–96) are concrete. Ike Antkare, an invented author, and Morgenstern's fake editors are not studies. But they prove that such tricks can be done, which is all they claim.
The book is also honest about its own frame. Fanelli's results go against the common story that pressure to publish drives misconduct, and he revises his own earlier view in print (p. 117). Wouters qualifies the editors' picture of how research is evaluated (p. 67). Jacob says the book itself largely leaves out the Global South, the poorer countries of Africa, Asia and Latin America (p. 254). An edited book that prints its own disagreements is more useful than one that does not.
Where it is thin
Many chapters are short opinion pieces or first-person accounts. Part III is mostly written by the people who run the projects it describes: Retraction Watch by its co-founder, PubPeer by two of the people who run it, altmetrics (scores of online attention to research) by a developer, and the work on detecting computer-generated SCIgen papers by its pen-named author. These accounts are valuable as testimony. They are not independent assessments, and the book offers none. For example, Barbour and Stell say critics can rarely point to a career that the site has unfairly harmed (pp. 152–153), but no data are given either way.
The main thesis, that metrics make new kinds of misconduct possible, is argued through examples rather than tested. The Introduction gathers many striking cases, several from news reports and blogs (pp. 10–13). The book does not try to measure how common these practices are. Fanelli's chapter raises a direct question for the thesis. The most rewarded authors are no more likely than others to have papers retracted (pp. 114–115). So how much of the new misconduct comes from metrics in general, and how much from cash rewards in particular? The book leaves the two views side by side.
Some details differ between chapters. Brookes gives 274 papers discussed on his blog (p. 174), and Lippman gives 275 (p. 272). The difference is small, but readers who cite the book should use the chapter that gives the first-hand account.
Blind spots
Geography. The cases come mainly from the United States, the United Kingdom and continental Europe. Brazil, China, the Czech Republic and Saudi Arabia appear as cases, not as authors. Every contributor whose institution the list names was based in Europe or North America when the book came out (pp. 285–286). Jacob names the gap directly (p. 254). The Arab world appears in two places: Kehm's account of two Saudi universities hiring highly cited researchers (p. 93), and the Introduction's note on King Abdulaziz University (p. 20, n. 19). In both, the universities are described from outside. Their own reasoning, and the national policies behind them, are not examined.
Disciplines. The examples lean toward biomedicine, physics and computer science. The humanities appear mainly through Power's UK case and two hoaxes: Sokal's parody article and the Postmodernism Generator.
Date. Most chapters were written between about 2015 and 2019, and the book appeared in 2020. Beall's list, a well-known blacklist of predatory journals (journals that take authors' fees while doing little or no real review), had closed in January 2017 (p. 108, n. 6). Some later developments lie outside its scope. From about 2022, paper mills, businesses that sell ready-made manuscripts and places on author lists, became a publicly documented concern of publishers and publication-ethics bodies (COPE and STM, Paper Mills, 2022). The book touches the same practice only through reports of authorship sold in prewritten papers already accepted for publication (p. 7), and COPE's finding of organizations selling prewritten manuscripts and fake reviewer contacts (pp. 12–13). The Retraction Watch database, which Oransky mentions as new in October 2018 (p. 144), was bought by Crossref in 2023. Large language models, the AI systems that write fluent text, came after the book. None of this weakens its argument. It means the cases should be read as a record of the 2010s.
Remedies. The book is better at diagnosis than at cures. Its proposals range widely: Wouters's "contextualized judgment" (p. 73), Lin's open data systems (pp. 221–223), Wager's call to give up the journal article (pp. 229–233), Griesemer's hacker journals (pp. 84–86) and PubPeer's rejection of all metrics (p. 155). The editors do not weigh these against each other.
Why it matters to a reader working on research integrity
For anyone who sits on promotion committees, edits a journal, runs a research office or teaches research ethics, the book gives names for practices that the standard definition of misconduct does not cover. Examples are "post-production manipulation" (gaming the publication process rather than the content), "spammish" publishing and "ghost management". It explains why these practices keep coming back: each new metric creates its own way to be gamed, so tighter rules for one number push the problem somewhere else. And it describes the field's tools, with their limits and the risks run by the people who use them: databases of retractions, review of papers after they are published, forensic checks on images, and software that detects machine-generated text.
It also carries a warning for institutions that reward publication by number or by journal. Fanelli found that countries paying cash per paper had the highest risk of retraction (p. 114), and Kehm describes researchers paid to list a university's name (p. 93). Both bear directly on any reward system tied to paper counts, impact factors (scores of how often a journal's articles are cited) or university rankings. Wouters points out that peer review, the judgment of work by experts, and the numbers are not independent of each other (pp. 69–72). That applies to every committee that reads a CV with a citation profile beside it.
The book does not offer a single remedy, and its authors do not agree on one. What it shows clearly is that misconduct in this sense is not only a matter of individual character. It grows from the way research is counted.