What Is Peer Review?

Peer review is the step where a journal asks other researchers to read a paper before it is published, and to say whether it should be. That is the whole idea in one sentence. Everything interesting is in the details — who those readers are, what they are paid, what they are able to check and what they are not — and those details are almost never explained to anyone outside the system.

This article is about the machinery rather than the judgement: what physically happens to a paper between a researcher finishing it and you being able to read it, and what that process can and cannot catch. The phrase peer-reviewed is worth a great deal. It is worth considerably less than most people assume.

What actually happens between finishing a paper and publishing it?

A researcher finishes a study and writes it up. At that moment nothing is public. The paper goes to a journal — a publication that prints research in a particular field — and the first person to read it is an editor, not a scientist going through it line by line.

The editor's first job is a quick sort. Most papers arriving at a well-known journal are never sent out for review at all. They are turned down within days, sometimes because the work is weak, more often because it does not fit the journal or is not thought important enough for it. This is called a desk rejection, and at the biggest journals it is what happens to the large majority of everything submitted.

If a paper survives that first sort, the editor goes looking for reviewers. Usually two or three: researchers who work on similar problems and have no connection to the study. They receive the paper and are asked for a written opinion, normally within a few weeks. In most fields the authors are never told who reviewed them, while the reviewers do know whose work they are reading. That asymmetry matters more than it sounds. A famous name gets a different reading from an unknown one, and nobody has to decide that it should.

The reports come back, and they are rarely a simple yes or no. The commonest outcome is a request to revise and resubmit: a list of objections, requests for clarification, sometimes a demand for an additional experiment. The authors rewrite the paper, answer every point in a covering letter, and send it back. The reviewers look again. A second round is normal. A third is not unusual.

The editor, not the reviewers, makes the final decision — reviewers advise, they do not vote. From first submission to publication, six months to a year is ordinary and longer is common. That slowness is the price of the filter, and it is precisely the cost that preprints, further down this page, exist to avoid.

Who are the reviewers, and what are they paid?

Nothing. This surprises almost everybody. The researchers who review papers are not employed by the journal and receive no fee from it. They are scientists with their own laboratories, their own students and their own grant deadlines, and they read the paper in the evening or at the weekend, in among everything else they owe someone.

There is no qualification for the job. No exam, no licence, no register you join. You are invited because an editor knows your work, or because two other people said no first. You can decline, and busy people frequently do, which means a paper's reviewers are partly a matter of who happened to be free that month.

Those who accept then vary enormously in what they do. One reviewer might spend a full day on a paper, recheck the arithmetic, and return four pages of specific objections. Another might skim it and write three lines. Both count as peer review, and the published paper looks identical either way. Attempts to fix this have been sobering: a randomised study that trained reviewers and removed their anonymity found neither change meaningfully improved the reviews that came back 2.

That anonymity is the other odd feature. Hiding the reviewer's name lets a junior researcher criticise a powerful one without fear. It also means nobody is publicly accountable for a lazy report, or for a rival's paper mysteriously held up for months.

So when a claim arrives with the words peer reviewed, here is what happened. Two or three unpaid, anonymous, uncredentialled and probably tired specialists read the paper and thought it good enough to appear. That is genuinely worth something. It is not the institutional inspection most people picture.

What review checks, and what it cannot

What a reviewer can do is read. So the checks are the ones you can make from a written description: is the method explained clearly enough that another laboratory could repeat it; does the design support the conclusion drawn from it; are the numbers internally consistent; is the claim in the final paragraph bigger than the result in the middle; has relevant earlier work been quietly ignored. These are real checks, and papers are genuinely improved by them.

What a reviewer almost never does is see the underlying data. They receive a description of what was found, not the measurements themselves, though a growing number of journals now ask authors to post their data publicly. Nobody repeats the experiment. There is no laboratory visit, no audit of the notebooks, no independent count of the animals. A reviewer who suspects that data were invented has very little to work with, and reviewers are not chosen for suspicion.

Nor can review rescue a study that was too small, or supply a comparison that was never made. A good reviewer will point out that fourteen participants cannot support a confident conclusion, and will ask for the wording to be softened. If the authors soften it and the editor is satisfied, the paper is published, and the fourteen people are still fourteen people.

This is a large part of why so many reviewed findings later fall apart. In an influential analysis, the physician and researcher John Ioannidis set out why a substantial share of published research findings should be expected to be false: small studies, small true effects, many relationships tested at once, and freedom in how the analysis is done all push in the same direction, and the reward for publishing something striking pushes harder 1. Peer review addresses almost none of that. It happens afterwards, at a distance, by a volunteer with a fortnight.

Predatory journals, and how to spot one

A predatory journal is one that takes a publication fee and provides little or none of the process described above. The paper appears online inside something that looks like a journal, with an official-sounding title and a list of editors. The review either did not happen or consisted of a single approving email a few days later.

This is not a fringe problem, and it is not always obvious from outside. A group of researchers who met specifically to define the term found no single feature separating these journals cleanly from the rest, and settled on describing them by behaviour: entities that prioritise self-interest at the expense of scholarship, marked by false or misleading information, a departure from accepted editorial practice, a lack of transparency, and aggressive soliciting of submissions 4.

How lax the checking can be has been demonstrated directly. A journalist wrote a deliberately hopeless paper — a fabricated study with flaws an undergraduate should catch, in which the data plainly did not support the conclusion — and submitted variations of it to 304 open-access journals under invented names and institutions. More than half accepted it 3.

One caution before the warning signs. Open access, meaning free for anyone to read, is not the same thing as predatory. Several of the most respected journals in medicine are open access, and an author fee is a normal way of paying for that. The problem is journals that take the fee and skip the work.

• An unsolicited email praising your esteemed contribution to a field you have never published in.
• A promise of publication within days, or within a fortnight. Genuine review takes months.
• A fee that only becomes clear after the paper has been accepted.
• A title almost identical to a well-established journal, differing by one word.
• An editorial board with no institutions listed — or researchers who never agreed to be on it.
• A scope covering everything: cardiology, agriculture, civil engineering and management studies in one publication.
• No named editor, no verifiable address, and no listing in any major academic database.

Why this matters here in particular: research peptides attract confident claims, and when you follow one back to its source it lands surprisingly often in a journal nobody in the field has heard of. The most useful check takes about a minute. Search the journal's title in PubMed, the free index run by the United States National Library of Medicine, and see whether it is there at all. A journal absent from every reputable index has told you something. A short public checklist built around three steps — think, check, submit — covers the rest of the signals 5.

Preprints, and what published is really worth

A preprint is a complete paper posted publicly by its authors before any review has taken place. Anyone can read it, immediately, free. Preprints have been ordinary in physics for decades and became familiar to everyone else during the pandemic, when waiting a year for findings was not an option.

The value is speed and openness. Often the version that eventually appears in a journal is barely different, and the delay served nobody. The catch is that nothing has filtered it. A preprint may be first-rate work whose journal is being slow. It may equally be a paper that four journals have already rejected. From the outside, those two look exactly alike.

So a preprint is not a lesser kind of publication. It is an unfiltered one, and it asks you to do the job the reviewers would have done: how many participants, what comparison, who ran it, does anything else point the same way. News coverage that describes a preprint as a new study published today has quietly skipped that step for you.

Which brings us to the honest answer about the word published. It is a floor, not a ceiling. It says the work cleared a minimum: a couple of specialists in the field read it and did not find a reason to stop it. That is worth having, and it is why a reviewed paper beats a forum post or a marketing page without argument. It does not say the study was large, the effect real, the result repeated, or the journal serious.

None of this is a reason to give up on published research. It is the reason to keep reading past the label. One reviewed paper is one card face-up on the table, and a single card rarely settles anything. What happens when somebody gathers every card in the deck — including the ones never turned over — is a subject of its own, and it is where this set of articles ends.