I have been publishing in academic journals for 25 years. I have published in dozens of them across multiple disciplines. I have refereed more than 100 articles for 41 unique journals. I have served as an editorial board member on five. I believe in peer review. I think the enterprise of having experts vet scholarly work before it enters the public record is one of the better ideas we have for managing knowledge, and I do not want to see it abandoned.

What I want is for the people who run journals to take their own systems as seriously as the people who depend on them. Because recently, I have had a run of experiences so bewildering that I started keeping notes. What follows are five of those experiences, in order from annoying to catastrophic. I then look at the research on why they matter and the unwritten rules that determine who bears the cost. The evidence reveals structural problems that shape tenure files, graduate training and who stays in academia—and which fall hardest on scholars who can least afford them.

Dispatch 1: Too Direct

My students and I submitted a paper to an open-access journal affiliated with a major professional society in the social sciences. The editors returned it without sending it out for external review. Among their feedback: my writing “comes across as somewhat too direct.”

Too direct. Think about what that means as editorial feedback. In most contexts, directness is a virtue in scientific communication. It means writing clearly, making your argument accessible across disciplinary boundaries and avoiding the hedged, jargon-laden prose that makes academic writing impenetrable to the public. And here was a journal telling me, on the basis of an internal read with no external review, that directness was a problem.

Maybe the editors simply preferred a different disciplinary register. But directness norms are not equally distributed. Scholars trained in traditions that prize accessible, plain-language communication—and scholars from backgrounds where directness is a cultural norm rather than a breach of etiquette—are disproportionately likely to hear this kind of feedback. Whatever the intent, the effect can be gatekeeping dressed up as editorial preference.

Dispatch 2: The Journals That Don’t Want What They Say They Want

A well-regarded open-access flagship journal of a social sciences association describes itself on its website as encouraging “short papers and those providing interesting empirical findings that may spark innovation and future work.” My students and I submitted a short, timely empirical paper. One reviewer acknowledged the “creative and unique research design” and the solid core findings but recommended rejection, in large part because the paper had too few citations. Articles in “a journal like this,” the reviewer wrote, typically have “40 or more academic citations, with potentially 80–100 in some cases.” The second reviewer recommended revisions. The paper was rejected.

I wrote to the journal editor to flag the disconnect between what the journal advertises online and how it actually evaluates submissions. His reply was candid: the encouragement of short papers was “one of the great intentions likely by the founding editors” that “has rarely, if ever, materialized.” He planned to raise the issue at the next board meeting.

Later, we sent the same paper to a different social sciences journal. This one explicitly lists a “Brief Research Report” format on its website, with its own word limit, structure and submission guidelines. Our paper fit the format. I emailed the editors anyway to confirm that it was a format they accept. They never replied. I went ahead and submitted our brief report. The rejection arrived as two sentences, with the main complaint that the work was “not sufficiently developed.” There was no engagement with the question, data or findings—and no acknowledgment that we had submitted in a format the journal actively solicits.

In both cases, the journal’s website said one thing and its editorial practice did another. In the first case, the editor admitted it. In the second, nobody even replied to the presubmission inquiry that would have saved everyone’s time. I knew to push back, to email the editor, to name the contradiction. A graduate student or early-career scholar submitting without that confidence would likely have absorbed the rejection and moved on, learning only that the system is not for them.

Dispatch 3: The Portal Versus the Journal

A social sciences journal states on its website that it allows simultaneous submissions. I went to submit a manuscript, and the publisher’s online system asked me to attest that I had not submitted the paper elsewhere. I had—to another journal that also allows simultaneous submissions. I emailed the editor, who confirmed simultaneous submissions were allowed and said the attestation was solely controlled by the publisher. The journal could not modify its submission portal on its own. I knew to question the form. A less experienced scholar might simply check the box, or decline to submit, having been taught to follow instructions precisely. The system does not distinguish between a rule and a glitch.

Dispatch 4: 519 Days

My co-author and I submitted a paper to a multidisciplinary journal in a major publisher’s portfolio. The process took more than 500 days. After our final revision, I could see that the last reviewer accepted the assignment and returned their report almost immediately. At that stage, the most likely explanation was an exceedingly brief report recommending acceptance, since all issues had been addressed. Then the handling editor went silent. I emailed the editorial office. The responses came back vague: “The coming week.” Nothing. Then: “The coming few weeks.” Nothing. Finally: “Soon.” The assistant editor was responsive and sympathetic but could not compel the handling editor to log a decision. When I asked whether the journal had a procedure for unresponsive editors, they said they did. It was not working.

Acceptance arrived on day 519. My co-author is an untenured scholar for whom timely publication directly affects career advancement. The preprint had been downloaded more than 1,000 times on a preprint server. The reviewer had done the work almost immediately. But the system could not turn a finished review into a decision for nearly two months, and could not move a paper from submission to decision in less than 17 months. For me, this was a frustration. For my co-author, it was a career risk imposed by other people’s inaction.

Dispatch 5: The Review, the Runaround and the Watchdog

My students and I submitted a paper to a general social sciences journal from a major commercial publisher. The paper was rejected on what appeared to be the basis of a single review by an associate editor rather than an external expert. The review was remarkable. One critique consisted of the single word “Specify.” Another offered “not really convincing” as the complete assessment of pages of detailed appendix material. There was no elaboration, no counterevidence: just three words. Our core findings were “not unexpected,” the reviewer wrote. The comment was that “you don’t need sophisticated analyses to show this.” But if researchers were barred from publishing results where they suspected the direction of an effect in advance, a great deal of quantitative social science would be invalidated. The paper’s value, the reviewer concluded, “only lies in the method, but absolutely not in the content.” We had applied a method to a substantive question. What did the reviewer think content was?

The problems went beyond tone. The review got basic facts wrong about works we had cited, misrepresented our research question and raised concerns we had already addressed in our limitations section—all while ignoring our mathematical framework, model selection, diagnostic tests and appendix entirely.

I sent a detailed, point-by-point rebuttal on the day of rejection, copying the publisher and executive editors. For 10 days, nobody replied. Over the next two weeks I was bounced from a project manager to an assistant editor to a senior editor. Nearly a month after my initial rebuttal, the journal finally responded on the merits. Its remedy was to have me resubmit as a brand-new submission. Then the story shifted. The journal claimed, for the first time, that two people had reviewed the paper: an external reviewer and the associate editor. I had only ever been sent one set of comments. Either a review existed and was withheld from us, or the account changed after the fact.

The review process had failed. The review was indefensible. But the proposed fix required my entire team, including my students, to absorb the full cost: more queue time, another review cycle and months of delay on top of months already lost. I filed a complaint with the leading international body for publication ethics, an organization with more than 15,000 members. They told me the journal had “given the concerns raised due consideration” and that the matter fell “within the remit of editorial decision making.” They could not “comment on or interfere with editorial decisions.”

My students watched all of this. They learned that a journal can deliver a manifestly deficient review, stonewall the corresponding author for weeks, propose a remedy that punishes the authors for the journal’s failure and face no consequences. That is not a lesson about peer review. It is a lesson about power.

The Unwritten Rule Book

Every one of these experiences required knowledge that is not written down anywhere: knowing that “too direct” is a judgment about conformity, not clarity. Knowing to email an editor before submitting, to check whether a journal’s website reflects its actual standards. Knowing to question an official-looking submission form rather than simply comply. Knowing that when a review process drags on for months, your only recourse is polite emails that may or may not be answered. Some of this I knew going in. Some of it I did not. I filed a formal complaint with the leading international body for publication ethics, fully expecting them to take seriously a review containing factual errors and a process that could not keep its own story straight. They told me the matter fell “within the remit of editorial decision making.” Even 25 years in, the system had lessons left to teach me.

Scholars of higher education call this the “hidden curriculum”—the unwritten rules and norms that govern a system but are never made explicit. In academic publishing, the hidden curriculum determines who can decode a confusing editorial letter, who knows when to push back on a bad review, who understands that a two-sentence rejection is not a verdict on their worth as a researcher. This knowledge is distributed unevenly, and the unevenness tracks every axis of disadvantage that already structures academia: race, gender, institutional prestige, national origin, career stage.

The philosopher Miranda Fricker names this dynamic “epistemic injustice”: a condition in which certain knowers are systematically denied credibility or excluded from the frameworks used to make sense of knowledge. In scholarly publishing, as Denisse Albornoz, Angela Okune and Leslie Chan argue, these injustices are built into the infrastructure itself—through criteria for “quality” and “academic literacies” that construct hierarchies of expertise, while rendering other ways of knowing invisible. The result is a system that looks meritocratic on its surface but functions, in part, as a filter for cultural capital.

Who Bears the Cost

The dysfunction is not distributed equally. Researchers in emerging regions produce nearly 30 percent of the world’s published scholarship but perform less than 19 percent of peer reviews, while editors disproportionately select reviewers from their own region and from established institutions. A study of editorial behavior across a large multi-journal platform found that male editors chose male reviewers 73 percent of the time. Biases in reviewing track institutional prestige, national origin and gender, and they compound through the system into disparities in whose work gets published.

Early-career researchers on precarious contracts face the sharpest version of these costs. A 519-day review timeline is an inconvenience for a full professor. For an untenured scholar assembling a tenure file, it can mean a missing line on a CV at the worst possible moment. For a researcher on soft money, it can mean a grant renewal that arrives too late. The system asks its most vulnerable participants to be the most patient and offers them the least recourse when things go wrong.

The Numbers

My five dispatches are anecdotes, but they are not anomalies. Peer review is unreliable. A meta-analysis of 45 studies found that the average correlation between two reviewers’ ratings of the same manuscript was 0.34. That is a modest level of agreement for a process often treated as authoritative.

The reviewer pool is also under strain. The last comprehensive measurement, from 2018, found that the number of invitations needed to secure one completed review had already risen from 1.9 to 2.4. Three-quarters of editors said finding reviewers was the hardest part of their job. Every indication is that the problem has worsened since. An estimated 10 percent of reviewers perform about half of all reviews. The global research community pours an estimated 100 million-plus hours a year into reviewing—roughly 15,000 person-years—almost entirely uncompensated.

Fraud is scaling faster than oversight. More than 10,000 papers were retracted in 2023, a record driven by fake reviews and paper mills. More than 35,000 retracted documents were identified in Scopus between 2001 and 2024. And machine-learning algorithms can now predict review outcomes from formal and textual manuscript features—word distribution, readability scores, formatting—without engaging with the substance of the research.

Reforms With Evidence

The journal publishing process—editorial decisions, submission infrastructure and peer review alike—is too important to abandon. But it is also too important to leave on autopilot. Every failure mode in the dispatches above implies a concrete fix: opaque stylistic policing calls for clearer desk-rejection criteria; mismatches between websites and practice call for periodic audits of author-facing policies; contradictory submission portals call for publisher-journal policy synchronization; unresponsive editors call for automatic escalation procedures; and deficient reviews with no recourse call for independent appeals processes.

One promising strategy for improving publication processes is open peer review. One study of more than 9,000 submissions found that publishing reviews alongside the paper did not reduce reviewer participation, slow turnaround time or substantially change quality. Only 8 percent of reviewers chose to sign their names, showing transparency and anonymity can coexist. A systematic review cataloged at least 122 definitions of “open peer review,” but the core idea—visible reviews, accountable reviewers—has evidence behind it.

Registered reports—in which journals conditionally accept a study’s design before results exist—are another strategy, one now offered by more than 300 journals. Early evidence is encouraging: About 60 percent of registered reports present null findings, versus roughly 5 percent in conventional publishing, suggesting reduced publication bias. A “Publish-Review-Curate” model, separating dissemination from evaluation, has also been explored. Reviewer training, the most intuitive fix, largely does not work: Randomized trials have found limited effects on quality. The problems are structural.

Worth Fixing

There is a risk in telling these stories publicly. A reader might conclude that I am someone who cannot publish, who cannot handle rejection, who does not understand how the game works. I am comfortable with that risk. I am a tenured full professor at a selective liberal arts college, and my publication record can speak for itself. But that is exactly the point. I can afford to say this out loud. Most of the people who experience what I have described here cannot. An untenured scholar who writes an essay like this is not being candid—they are being reckless. A graduate student who names a journal’s dysfunction publicly is not speaking truth to power—they are jeopardizing a career that has not started yet. The system’s failures stay hidden not because they are rare, but because the people who know them best have the most to lose by talking about them.

Scholarly publishing is a multibillion-dollar industry sustained by donated labor. The research on how to improve it exists. The question is not whether better models are available. It is whether the people who run journals will hold their own processes to the same standards of coherence, rigor and transparency that they demand of authors.

Chad M. Topaz is a professor of complex systems at Williams College, cofounder of the nonprofit QSIDE Institute and author of the new book Unlocking Justice: The Power of Data to Confront Inequity and Create Change (Princeton University Press). Reach him on Bluesky at @chadtopaz.bsky.social.

Next Story

More from Advancing in the Faculty