An editor at a clinical journal receives a decent manuscript. It needs two reviewers with relevant expertise.
She sends five invitations. Over the following ten days, three are ignored entirely, one declines, and one accepts.
She sends five more. Two decline, three ignore. She sends five more.
Three weeks after submission, she has two acceptances. One of those reviewers will never deliver, which she will discover in another three weeks, at which point she starts again.
The authors, meanwhile, are watching a submission portal that says "under review" and wondering whether anyone has looked at their paper. They have a grant renewal that depends on it.
Everyone in this story is behaving reasonably and the system is failing anyway.
The trend line
The numbers describing this have moved sharply in a short period.
Invitations per manuscript. An editorial in Acta Odontologica Scandinavica documented invited reviewers per manuscript rising from two or three in the 2006 to 2010 period to six or seven in 2019 to 2024, noting it is "not uncommon to invite 20 researchers or more to recruit 2."
Acceptance rates. Data published in PRiMER covering 2,951 invitations across 459 manuscripts tracked invitation acceptance at 44.69 percent in 2017, rising to 56.14 percent in 2020, then falling to 35.71 percent in 2024 and 38.58 percent in 2025.
Completion. The same dataset found completion rates among those who accepted falling from 89.31 percent in 2021 to 76.19 percent in 2025.
And non-response. Statistics published by one journal in 2025 found 58 percent of invitees simply ignored the request and 29 percent declined, meaning 86 percent produced no positive response, with a further 14 percent of those who accepted never delivering.
Notice the pandemic spike in the middle of that acceptance series. Acceptance rose in 2020, when many academics were grounded and had unusual amounts of time, then fell well below the pre-pandemic baseline afterward. That pattern is a strong hint about what is actually being measured, and we will return to it.
The counter-evidence, which is the important part
Before accepting the standard "reviewer shortage" framing, it is worth engaging with evidence that contradicts it, because it substantially changes the diagnosis.
An analysis published through Springer Nature research communities examined invitation patterns at the Journal of Comparative Physiology A from 2014 to 2021 and found no increase in the number of invitations required to secure reviews. The piece was titled, pointedly, as debunking the myth of reviewer shortage.
So at one journal, over eight years, the reviewer recruitment burden was stable.
How can that coexist with editors elsewhere sending twenty invitations to secure two?
The most coherent explanation is that the two journals differ not in the availability of reviewers but in the editors' ability to identify the right ones. A specialized journal with a well-defined community, where editors know the field personally and invite people who are genuinely relevant and likely to say yes, does not experience a shortage. A broad clinical journal with high submission volume, where editors search a database for anyone who has published on the topic, does.
Which means this is not primarily a supply problem. It is a matching problem.
That reframe changes everything about what would fix it.
What is actually scarce
Be precise about what an editor lacks at the moment of invitation.
They can identify people who have published on the topic. That information is public and easy.
What they cannot see is:
Availability. Is this person free this month? Reviewer capacity is a time-varying private state. A researcher who happily reviewed three papers in March may be on grant deadline in April and will not respond at all.
Current load. How many reviews is this person already doing, for how many journals? Nobody can see this, including the researcher's own institution.
Willingness for this topic. Having published on something years ago does not mean wanting to review it now.
Reliability. Will they deliver? The 14 percent who accept and never complete are invisible until the deadline passes.
And relevance beyond keywords. The most appropriate reviewer for a methods-heavy clinical paper may not be the person with the most topical publications.
Every one of those is knowable, and none of them is visible at invite time. So editors invite on the only signal available, publication history, and the resulting hit rate is roughly one in three and falling.
Why the ledger is missing
The deeper structural problem is that peer review is a professional obligation with no accounting.
The norm is real and widely accepted: you review because you submit. A researcher who publishes several papers a year, each requiring two or three reviews, incurs a debt of several reviews a year to the community.
Nobody tracks it. Not journals, not institutions, not funders.
So the researcher who reviews twelve papers a year and the one who reviews zero appear identical to everyone. Neither is recognized nor asked about it. Promotion committees do not count it. Funders do not weigh it.
And the consequence is predictable. Reviewing is unpaid, uncounted, invisible work with a real time cost and no professional return. The remarkable thing is not that acceptance rates are falling. It is that they remain as high as they are, which is a genuine credit to the profession.
Recognition systems have tried to address this by counting reviews after they happen. That is useful for the reviewer's record and it does nothing for the editor's problem, which is knowing who is available before sending the invitation.
Journals also hoard their own reviewer histories. Each journal knows which reviewers are reliable for it. None of that is shared, so every journal rebuilds the same knowledge independently, and a reviewer who is excellent and available is invisible to every journal that has not used them before.
Why this is getting worse now
Two pressures are converging.
Submission volume keeps rising, including a growing volume of manuscripts produced with substantial machine assistance, which increases the number of papers requiring assessment without increasing the number of people qualified to assess them.
And the same names are being over-invited. When an editor cannot see availability, they default to people they know are qualified, which concentrates the burden on a small group of senior researchers who are consequently the most likely to decline. That concentration is self-reinforcing: the more reliably someone reviews, the more they are asked, until they stop.
Meanwhile, an entire pool of capable reviewers is never invited at all: early-career clinicians and researchers who would review, would do it well, and are not in any editor's contact list.
And a second, larger pool: retired and semi-retired clinicians and academics, who have deep expertise, current-enough knowledge in their areas, substantial available time, and no competing grant deadlines. They are approaching invisibility to editors for exactly the reasons this series has documented elsewhere: their institutional affiliation has lapsed and their publication activity has slowed, so keyword searches no longer surface them.
The one pool of expert time that is growing is the one the discovery mechanism cannot see.
The cost of the delay, which nobody bills
It is worth stating what the recruitment failure actually costs, because the discussion usually stops at editorial inconvenience.
Authors wait. Every failed invitation wave adds one to two weeks. An editor who needs three waves has added a month to the review timeline before a single reviewer has read anything. For a paper that eventually requires revision and a second review round, the recruitment overhead alone can add several months to publication.
Careers are timed against those months. Early-career researchers are assessed on publication records at fixed points: fellowship applications, grant renewals, promotion review. A delay that is merely annoying to a senior author can be materially consequential to someone whose next position depends on what is in press by a particular date.
Evidence reaches practice later. For clinical research, the interval between submission and publication is time during which the finding is unavailable to the clinicians who would act on it. That is a real cost paid by patients, and it is entirely invisible because nobody counts the counterfactual.
And desk rejection rises as a coping mechanism. An editor who cannot reliably recruit reviewers has one lever that always works: reject more manuscripts without review. That is a rational response to a broken recruitment process and it means some proportion of the papers rejected without review are being rejected because reviewing them was operationally difficult rather than because they lacked merit.
That last effect deserves more attention than it receives. A reviewer discovery failure is quietly reshaping what gets published, and it does so invisibly, because nobody can distinguish a manuscript rejected on quality from one rejected on friction.
What would actually work
A declared capacity signal. A researcher stating "I will do two reviews per quarter, in these areas" turns a private, time-varying state into a public one. This single change addresses the largest component of the failure, which is that editors are guessing about availability.
Matching on relevance plus availability, not publication history alone. The best available reviewer is the intersection of qualified and free, and current systems can only see the first term.
A visible ledger of contribution. Reviews completed, counted, and recognized as professional service. Not for its own sake, but because an obligation with no accounting is systematically undersupplied. Recognition after the fact is worth something; recognition that also signals availability is worth much more.
Reciprocity that is actually enforced socially. The norm that you review because you submit works when it is visible. Right now a researcher can publish extensively and review nothing with no consequence, because nobody can see it.
Deliberate recruitment of the invisible pools. Early-career reviewers, who want the experience and are rarely asked. And retired experts, who have the time and are no longer surfaced by any search.
And shared reliability signals, carefully. Journals collectively know who delivers. Some mechanism for sharing that, with obvious care about fairness and the ability to correct the record, would prevent every journal from rediscovering it independently.
What you can do now
If you are an editor
Ask about availability separately from expertise. A brief note asking whether someone is taking reviews this quarter, before formally inviting, has a materially better response rate than a full invitation, because it is a smaller ask.
Invite early-career researchers deliberately. They accept at higher rates, frequently produce more thorough reviews because they take the responsibility seriously, and they are the reviewer pool of the next twenty years.
Ask declining reviewers to name a substitute. A researcher who cannot review will often suggest someone appropriate, and that suggestion carries real signal about relevance that no database search produces.
Tell reviewers what happened. A note explaining the decision and how their comments were used costs a minute and is one of the very few forms of recognition available in this system.
If you review
Declare your capacity publicly. A line on your profile stating what you review and roughly how much per quarter costs nothing and makes you findable to editors who currently cannot tell whether you are free.
Decline fast and suggest someone. A same-day decline with a name attached is enormously more useful to an editor than a week of silence, and it takes ninety seconds.
Count what you do. If you review ten papers a year, that is a substantial professional contribution that currently appears in no record of your work. Keep the record yourself.
Consider your own ratio. Papers submitted, multiplied by two or three, against reviews completed. Most researchers have never calculated it, and the calculation is clarifying.
If you are retired or semi-retired
Tell journals in your field you are available. You are the most underused expert reviewing capacity in existence, and editors cannot find you because their search tools index recent publications and current affiliations. A direct email to the editors of two or three journals in your area is all it takes.
If you lead an institution or a funder
Count peer review as service. It is genuine professional labor sustaining the evidence base your institution depends on, and treating it as invisible is why it is undersupplied.
Ask about it in promotion review. Nothing would change reviewer behavior faster than the field's recognition systems noticing who does the work.
Frequently asked questions
Is there a peer reviewer shortage? The evidence is contested and the framing may be wrong. Invitation acceptance rates at some journals fell from around 56 percent in 2020 to roughly 36 to 39 percent in 2024 and 2025, and editors report inviting six or seven reviewers per manuscript versus two or three a decade earlier. However, an analysis of one specialized journal found no increase in invitations required between 2014 and 2021, suggesting the problem lies in reviewer discovery and matching rather than absolute supply.
How many reviewers do editors have to invite? Published editorial accounts describe six to seven invitations per manuscript in recent years, up from two to three in the 2006 to 2010 period, with cases requiring twenty or more invitations to secure two reviews.
Why do researchers ignore review invitations? Reviewing is unpaid, uncounted, and largely invisible professional work with a real time cost. Institutions rarely weigh it in promotion, funders do not count it, and no record exists distinguishing a researcher who reviews twelve papers a year from one who reviews none. Journal statistics have found 58 percent of invitees simply not responding.
Do reviewers who accept always complete the review? No. One journal reported 14 percent of those who accepted never delivering, and tracked completion rates falling from about 89 percent in 2021 to 76 percent in 2025.
What information do editors lack when inviting reviewers? Availability this month, current review load across all journals, willingness to review this specific topic now, and reliability of delivery. Publication history, the only signal readily available, correlates weakly with all of them.
Who is not being asked to review? Two large pools. Early-career researchers and clinicians who would accept at higher rates and are not in editors' contact lists, and retired or semi-retired experts who have deep knowledge and substantial available time but no longer surface in searches based on recent publications and current institutional affiliation.
The bottom line
An editor sends twenty invitations to find two reviewers while, somewhere in the same field, a recently retired professor with forty years of expertise and an empty calendar has not been asked to review anything in three years.
The framing of this as a shortage has produced a decade of appeals to professional duty, which have not worked, because the problem was never a deficit of willingness.
The problem is that capacity is a private, time-varying state that nobody publishes, so editors invite on publication history, which tells them who is qualified and nothing about who is free. And because the obligation has no ledger, the people who do the work receive nothing for it, while the people who do none are indistinguishable from them.
We have built an evidence base that depends entirely on unpaid expert labor, and we have never built the two things that would make that labor findable: a signal for who is available, and an account of who has contributed.
The reviewers exist. Nobody can see which ones are free.
Part of a series on the missing professional infrastructure of healthcare. Previously: The Home Program Gap
Evidence note: acceptance and completion rate data come from STFM PRiMER (2025) covering 2,951 invitations across 459 manuscripts at a single journal. Invitation volume trends come from an Acta Odontologica Scandinavica editorial (2024). Non-response figures come from journal-published statistics (2025). Counter-evidence on invitation stability comes from an analysis of the Journal of Comparative Physiology A published via Springer Nature research communities. Single-journal data may not generalize across fields, and the divergence between these sources is discussed explicitly in the text.
