Scientific journals are the primary vehicles through which researchers share findings, and the system surrounding them is more varied and more contested than most people realize. There is no single type of journal, no single way papers get reviewed before publication, and no single number that reliably tells you how good a journal is. Understanding the landscape means knowing the differences between subscription and open-access models, how peer review actually works (and where it fails), what metrics like the impact factor measure and what they miss, and why threats like predatory journals and paper mills have become serious problems for the integrity of science.
How Journals Make Money and Who Pays
For most of the twentieth century, journals operated on a subscription model: libraries and individuals paid to read the content. Researchers submitted papers for free, volunteer reviewers evaluated them for free, and publishers packaged the finished product and sold access. That model still exists, but a significant shift has been underway for the past two decades toward open access, where anyone can read published papers without a subscription. The trade-off is that someone else has to cover the costs.
The most common way open-access journals fund themselves is through article processing charges, or APCs, which are fees the authors (or their funders or institutions) pay when a paper is accepted. APCs apply to fully open-access journals and also to “hybrid” journals, which are subscription-based but allow individual articles to be made open access for a fee.1arXiv. Estimating global article processing charges paid to six publishers for open access between 2019 and 2023 There is a clear pattern in how these fees are set: journals with higher prestige metrics tend to charge more, and those charges have been rising faster than at lower-ranked journals. Authors report mixed feelings about the system. Researchers in fields where grants routinely cover publication costs are more comfortable with APCs than those in disciplines where funding is scarce, leaving some scholars to pay out of pocket or lean on library funds.2Learned Publishing. Article processing charges for open access journal publishing: A review
A third category, sometimes called “diamond” open access, charges neither readers nor authors. These journals are typically run by scholarly societies, universities, or volunteer communities. They are widely praised in principle, but the reality is that most open-access articles still end up in commercially run journals that charge APCs.2Learned Publishing. Article processing charges for open access journal publishing: A review The economics of the system shape which journals thrive and which struggle, and that has downstream effects on who can afford to publish and where.
Mega-Journals and Preprint Servers
Not every journal works the same way editorially, either. A category that emerged in the 2000s and 2010s is the mega-journal, with the two largest each publishing more than 20,000 articles per year.3Online Information Review. Publishing speed and acceptance rates of open access megajournals What makes mega-journals distinctive is their review philosophy: they assess whether the science is methodologically sound, but they deliberately do not judge whether the findings are exciting or important enough to publish. That judgment is left to readers. The result is acceptance rates in the range of 50 to 70 percent and publication timelines of roughly three to five months, both significantly better than what most traditional journals offer.4PubMed Central. Have the “mega-journals” reached the limits to growth?
Preprint servers sit even further from the traditional model. Platforms like bioRxiv and arXiv let researchers post manuscripts before they have been peer-reviewed at all. The appeal is speed: the traditional publication process can take months or years, while a preprint goes live in days. A survey of bioRxiv users found that about 30 percent post preprints weeks to months before submitting to a journal, while 55 percent post around the same time they submit.5bioRxiv. bioRxiv: the preprint server for biology Preprints are not peer-reviewed, so they carry more uncertainty than published papers, but they have become an increasingly normal part of the research communication ecosystem. Many journals and funders now explicitly encourage preprint posting, and some new review initiatives have grown up around preprint servers to experiment with alternative forms of evaluation.5bioRxiv. bioRxiv: the preprint server for biology
How Peer Review Works
Peer review is the process by which other researchers evaluate a manuscript before a journal decides whether to publish it. At most journals, an editor first screens the submission and then sends it to two or three external reviewers with relevant expertise. Those reviewers read the paper, flag problems, and recommend whether the journal should accept, revise, or reject it. The editor makes the final call. The whole system runs on volunteer labor: reviewers are typically unpaid, and even editors at many journals are working scientists who do the job part-time.
The most common format is single-blind review, where the reviewers know who wrote the paper but the authors do not know who reviewed it. Double-blind review hides identities in both directions. And open peer review is an umbrella term for several newer approaches, including publishing reviewer reports alongside the paper, revealing reviewer identities, or allowing the broader community to comment on manuscripts.6PubMed Central. What is open peer review? A systematic review
Post-publication review platforms take the idea further. On these platforms, a paper is published first and then reviewed openly. One analysis of F1000Research found that 80 percent of articles had successfully undergone peer review within six months of posting, and about three-quarters passed on the initial submission without needing revisions. Authors cited the open access policy, open peer review, and speed of publication as the top reasons for choosing the platform.7PubMed Central. Who and why do researchers opt to publish in post-publication peer review platforms? – findings from a review and survey of F1000 Research
Where Peer Review Falls Short
Peer review is treated as the gold standard of scientific quality control, but the evidence on how well it actually works is sobering. In one study, researchers inserted nine deliberate major errors into manuscripts and sent them to reviewers. On average, reviewers caught fewer than three of the nine errors. One type of error, biased randomization, was spotted by over 60 percent of reviewers who ultimately rejected the paper, but fewer than 40 percent of reviewers who recommended acceptance noticed it. The study’s conclusion was blunt: editors should not assume reviewers will detect most major errors.8PubMed Central. What errors do peer reviewers detect, and does training improve their ability to detect them?
Bias is another persistent issue. When reviewers can see author names and affiliations (single-blind review), their judgments are measurably influenced by prestige. A randomized experiment found that reviewers recommended acceptance 87 percent of the time when prestigious authors’ identities were visible, compared to 68 percent when those identities were hidden, with higher ratings for methodology as well.9JAMA. Single-blind vs Double-blind Peer Review in the Setting of Author Prestige A separate large-scale study at a computer science conference found similar patterns: single-blind reviewers were significantly more likely to recommend acceptance for papers from famous authors, top universities, and top companies, with the strongest effect for well-known companies, where the estimated odds of acceptance roughly doubled compared to double-blind conditions.10PubMed Central. Reviewer bias in single- versus double-blind peer review
These findings do not mean peer review is useless. It catches some errors, encourages authors to improve their manuscripts, and serves as a basic filter. But it is far from the airtight quality guarantee that the public sometimes imagines. Problems regularly slip through, and reviewer decisions are shaped by factors that have nothing to do with the quality of the science.
What the Impact Factor Measures and What It Misses
The Journal Impact Factor is the most widely recognized metric in academic publishing. It measures how often articles published in a journal over the previous two years were cited by other papers in the current year, divided by the total number of citable articles. A journal with an impact factor of 5 means its recent articles were cited an average of five times each. It was originally designed as a tool for librarians deciding which journals to subscribe to, not as a measure of individual paper quality or researcher merit.
That distinction matters, because the impact factor has been repurposed far beyond its original intent. Hiring committees, promotion boards, and funding agencies routinely use journal-level metrics as shortcuts for evaluating individual researchers. The pressure this creates has been widely criticized. Over-reliance on impact factors can be especially harmful to early-career scientists who may not yet have access to the networks and resources needed to publish in the highest-ranked journals.11PubMed Central. Impact factor: Mutation, manipulation, and distortion
Several alternative metrics have been developed to address the impact factor’s limitations. The Eigenfactor Score weights citations by the prestige of the citing journal, so a citation from a highly cited journal counts more. The SCImago Journal Rank uses a similar approach based on the Scopus database. These alternatives correlate with the impact factor in some fields but diverge in others. In pediatric neurology, for example, a comparison found a high correlation between the impact factor and one alternative metric (the Article Influence Score) but weaker correlations with the others, suggesting the metrics are capturing somewhat different things.12PubMed Central. Comparison Between Impact Factor, Eigenfactor Metrics, and SCimago Journal Rank Indicator of Pediatric Neurology Journals Factors like self-citation rates, the proportion of review articles, and total article volume also influence metric rankings in ways that have nothing to do with research quality.12PubMed Central. Comparison Between Impact Factor, Eigenfactor Metrics, and SCimago Journal Rank Indicator of Pediatric Neurology Journals
In response to these concerns, the San Francisco Declaration on Research Assessment (DORA) was launched in 2012. It urges institutions and funders to stop using journal-based metrics as proxies for the quality of individual research outputs. The movement has gained traction across disciplines, including fields like nursing science, where DORA principles are being adopted to reshape how researchers are evaluated.13PubMed Central. The San Francisco Declaration on Research Assessment (DORA) in Nursing Science Progress has been uneven, though. Many institutions pay lip service to DORA while continuing to weight impact factors heavily in practice.
Predatory Journals and How to Spot Them
One of the more troubling developments in scientific publishing is the rise of predatory journals. These are outlets that present themselves as legitimate scholarly journals but misrepresent their practices. Common tactics include falsely claiming to provide peer review, hiding or misrepresenting article processing charges, listing fabricated editorial boards, and violating copyright norms.14PubMed Central. Predatory Journals: What They Are and How to Avoid Them The business model is simple: charge authors a fee, publish whatever they submit with minimal or no review, and pocket the money.
For researchers, especially those early in their careers or working in countries with strong “publish or perish” pressure, predatory journals can be a trap. A paper published in one carries little weight with informed readers and can actively damage a researcher’s reputation. Red flags include aggressive email solicitations, unusually fast promises of publication, vague or absent information about the review process, and a journal title that closely mimics a well-known legitimate journal. Checking whether a journal is indexed in established databases and whether its editorial board members are real, active researchers in the field are basic precautions.
Paper Mills and Large-Scale Fraud
Predatory journals are not the only integrity threat. Paper mills are organizations that produce fabricated manuscripts for profit, selling authorship slots to researchers who need publications on their CVs. This is not a fringe problem: many thousands of fake papers have made it into peer-reviewed journals.15PubMed. Paper mill challenges: past, present, and future The first identified paper mill manuscript dates to 2004, with the first retraction not coming until 2016. By 2021, paper mill retractions accounted for roughly one in five of all retractions that year.16BMJ. Retracted papers originating from paper mills: cross sectional study
The scale is troubling because it suggests the traditional peer review system was not designed to catch coordinated fraud. A single researcher fabricating data is one thing; an industrial operation mass-producing plausible-looking manuscripts and distributing them across dozens of journals is a qualitatively different challenge. Some researchers have argued that the combination of internet-era communication, open-access publishing models, and intense career pressure has created fertile ground for these operations to grow.17PubMed Central. The entities enabling scientific fraud at scale are large, resilient, and growing rapidly Recent evidence indicates the entities behind this kind of fraud are not shrinking but expanding.
Detection has increasingly relied on post-publication scrutiny. Volunteer sleuths and dedicated integrity investigators comb published literature for patterns that signal mill activity: suspiciously similar figures across papers from unrelated groups, boilerplate text, or implausible author networks. This kind of post-publication peer review has unraveled many cases of misconduct and helped correct the scientific record, though it remains largely uncompensated and undervalued by institutions.18PubMed. Post-publication Peer Review with an Intention to Uncover Data/Result Irregularities and Potential Research Misconduct in Scientific Research: Vigilantism or Volunteerism?
Publication Bias and What Gets Published
Even in legitimate journals with rigorous review, the published literature does not perfectly represent the state of knowledge. Publication bias, the tendency for studies with positive or statistically significant results to be published more readily than null or inconclusive findings, has been a recognized problem for decades. If ten research teams independently test a hypothesis and only the two that find a statistically significant effect get published, readers of the literature come away with a distorted picture.
The issue is more nuanced than it first appears, though. A recent analysis argued that when studies are conducted at reasonable levels of statistical rigor and cover a range of different topics, a preference for publishing significant results can actually increase the proportion of genuine findings in the literature without dramatically inflating false positives.19PubMed. Balanced examination of positive publication bias impact That is a more optimistic framing than the usual alarm, but it depends on conditions that are not always met in practice, particularly the assumption that individual studies are well powered and properly designed. In fields where small sample sizes and flexible analysis methods are common, publication bias remains a serious concern.
Preregistration, where researchers publicly commit to their study design and analysis plan before collecting data, has emerged as one practical response. It makes it harder to cherry-pick results after the fact and gives journals and readers a way to check whether the published analysis matches the original plan. Some journals now offer “registered reports,” a format where peer review happens before results are known, which removes the incentive to produce flashy findings entirely.
AI in the Review Process
Artificial intelligence tools are beginning to change how peer review and editorial work are done. At the most basic level, AI can help editors screen submissions for formatting compliance, detect plagiarism, or flag statistical anomalies. More ambitiously, some platforms are experimenting with AI-assisted reviewer matching, where algorithms suggest appropriate reviewers based on their published expertise and availability. The potential benefits for efficiency are real, but so are the concerns about transparency, accountability, and whether algorithmic tools might introduce new biases or be gamed by bad actors.20PubMed Central. Artificial Intelligence in Peer Review: Enhancing Efficiency While Preserving Integrity
On the fraud detection side, AI shows more unambiguous promise. Automated tools can scan manuscripts for image manipulation, detect text patterns associated with paper mills, and flag citation networks that look artificially constructed. These tools are not replacing human judgment, but they can process the volume of submissions that no editorial team could manually review. As paper mills grow more sophisticated, the arms race between fabrication and detection is likely to become increasingly computational on both sides.
How Scholarly Publishing Became a Business
The tension between science as a public good and publishing as a commercial enterprise is not new. In the decades after the Second World War, learned societies that had traditionally published journals as part of their scholarly mission found themselves facing a choice. The volume of research was expanding rapidly, and commercial publishers offered the infrastructure to handle it. What followed was a transformation: scientific research became something to be packaged and sold to libraries rather than circulated as part of a nonprofit scholarly exchange.21PubMed Central. Self-help for learned journals: Scientific societies and the commerce of publishing in the 1950s For some societies, this was an opportunity to stabilize fragile finances. For the research community as a whole, it set the stage for the subscription pricing crisis that would come decades later, when journal prices rose far faster than library budgets.
That history shapes everything about the current system. The open-access movement is, in many ways, an attempt to reverse the mid-century commodification of research. But the commercial incentives that drove consolidation in the subscription era are still present in the APC era, just configured differently. A handful of large publishers still dominate the market, and the most prestigious journals still command the highest prices, whether those prices are paid by subscribers or by authors. Understanding scientific journals means understanding that they are not just containers for knowledge. They are businesses operating within an economic system that rewards prestige, speed, and volume in ways that sometimes align with scientific quality and sometimes work against it.