Manuscript Submissions Are Up! That’s Good, Right?

Academic publishing has long been a popularity contest: each journal is an exclusive club to which many are drawn, but only a few get invited in. The best clubs feature the most brilliant editors, the most interesting research, and – by extension – the choosiest manuscript selection process.

It therefore makes good sense to measure the success of your club by the length of the queue outside, quantified by the number of manuscripts submitted by hopeful authors each year. More manuscripts equate to being more popular; more manuscripts also mean the editors have a wider range of research to choose from. Both good things. And, some decades ago, steadily increasing submission numbers may have been a reliable indicator that the journal was headed in the right direction.

I argue here that the total number of submissions a journal receives each year has long ceased to be a useful performance indicator. In fact, chasing ever more submissions is having a serious deleterious effect on our industry, and we need to stop counting rising submissions as success as soon as possible.

1. The Money

First, the number of articles published each year has risen for decades. While some of the extra manuscript volumes do represent additional valuable research, the biggest area of recent growth has been in weak, flawed, or even fraudulent work. The second half of 2025 has seen dramatic increases in submissions at many journals, likely because Large Language Models make drafting (or outright fabricating) manuscripts so easy.

Receiving more submissions now means feeding ever more junk into the journal’s peer review pipeline. In many cases, it’s not just the journal pipeline: often, the goal is to keep the manuscript within the publisher’s publication cascade so that it will eventually be published somewhere and the editorial costs recouped via an APC. Keeping junk within your cascade means the costs discussed below are even higher, particularly if the manuscript never gets published.

Each submission has a cost. Journals have to pay for journal management infrastructure on the basis of how many submissions they receive. Increasing submissions also means more person-hours in the Editorial Office (EO) on basic manuscript checks, or hiring vendors to help with screening. More submissions mean more email from authors that the EO has to deal with.

Since there is considerable Editorial Office work before a new submission reaches an Editor (e.g., checking that all the figures have been provided), desk-rejecting all of the junk still imposes an expense. Moreover, the junk tends to need more Editorial Office time & attention before it meets the threshold to be passed to an Editor. The costs rise sharply for each manuscript that makes it into peer review, as much more Editorial Office time is spent on managing reviewers, sending emails, and monitoring the review process.

Multiplied across 10,000+ journals, chasing ever more submissions (even when many get immediately rejected) imposes massive costs on our industry. We may publish about 4.5 million articles per year, but — assuming an average acceptance rate of 30% — we’re collectively receiving about 15 million submissions.

Assume that 50% of submissions get a desk rejection, and each of those desk rejects costs $20, while 20% of articles going into peer review and being rejected cost $100, and the 30% getting accepted cost $250 (before production). That’s $1.5 billion (with a ‘B’) in peer review costs, and I suspect those dollar values are underestimates.

Fast forward to the end of 2026 or mid-2027, and it’s not implausible that there’s another five million junk manuscripts out there, hitting our journals over and over. That’s an additional $100 million in peer review costs just to let them get submitted. Since those authors pay nothing to submit, that cost will have to be added onto APCs, subscriptions, or whatever model we’re using by then to cover these peer review costs.

As you can imagine, celebrating an extra $100M in costs as a win because submissions went up is a recipe for communal disaster.

2. The Innovation Shackles

Second, when ‘number of manuscripts submitted’ is treated as a key performance indicator, any initiative that might deter authors from submitting is deemed too risky. These initiatives range from requiring authors to share their data prior to submission to basic author-facing automated checks for compliance with journal submission guidelines. Anything that could prompt the author to abandon the submission process and go elsewhere is viewed as a threat to submission targets. Submission fees are unpopular for the same reasons.

To abuse the exclusive club metaphor from above a little more: our club has a strict dress code, but our success metrics demand we let everyone in and then have the bouncers wrestle the hippies in flip-flops off the dance floor. Someone turning up with a device to assess dress code compliance for punters in the queue is turned away because we dare not do anything to discourage the hippies in the queue.

[Minor aside: letting the hippies into the club also runs the risk that they’ll end up doing interpretive dance on the main stage (i.e., published in your journal), leaving your reputation as a sophisticated, James-Bond-esque club in tatters.]

3. The Fix

I’ve largely foreshadowed this, but as an industry, we need to make authors demonstrate that they care about doing good science before we let them submit their manuscript to a journal. This logic also applies to preprint servers: since submission requirements for a preprint server are minimal, they’re receiving a growing avalanche of manuscripts. If these servers are to have a value for readers, then multiple quality control steps need to be in place.

Decades ago, we forced authors to demonstrate their desire to send in their manuscripts to our journals by making them comply with bewildering formatting requirements at initial submission. While these requirements normally served no functional purpose, they imposed a sufficient time cost on authors whose papers were unlikely to be published that they did not bother to submit in the first place. Since each journal had its own idiosyncratic submission format, shopping a manuscript around multiple journals was far too much work to be worthwhile.

Fortunately, technology has moved on, and specialist tools like blueberg.ai will be able to automatically assess manuscripts against journal submission requirements and let authors address issues before they get submitted. Surfacing these issues before submission makes sense for everyone: authors are less likely to have their manuscript kicked back, and a much higher proportion of manuscripts will meet journal standards the first time, saving the Editorial Office a great deal of hassle.

Lastly, prompting authors to do improvement work on their manuscript prior to submission acts as a prequalification step: are they sufficiently committed to submitting to your journal to do what you’ve asked? If not, then you’re much better off if they take their work elsewhere.

The prequalification step is doubly important when the journal has requirements that relate to research integrity. Note: I am not talking here about the negative signals of integrity, such as image manipulation or paper mill activity, where it is vital that authors not be able to ‘test run’ their problematic manuscripts against the detection tools.

Instead, when the journal requires ‘honest signals’ of research integrity — time-consuming actions like sharing data and code, including RRIDs for reagents, or using ORCIDs — then helping authors to comply with these prior to initial submission accomplishes several good things.

1. First, the journal sends a powerful message that it cares about research integrity.
2. Second, by itemizing the research integrity actions the authors need to take prior to submission, the authors are accountable for taking those actions. It’s easy to ignore a generally worded blandishment like ‘please share your data’; it’s much harder to ignore a concise list of the individual datasets from your manuscript that the journal expects to see on a public repository.
3. Third, authors who fear sharing their data and code, perhaps because they’re fabricated, or because the work was done in such a rush that the datasets are a mess, these authors will quietly take their weak submissions elsewhere. Journal submissions will go down, but the submissions that do arrive will be of higher quality.
4. Fourth, it is much easier to maintain Editor enthusiasm and find willing reviewers when the work is by authors who clearly care about the quality of their science.

So, to recap, the publishing industry has led itself into a mess by focusing on rising submissions as a positive indicator of journal performance. Journals are now receiving a flood of low-quality work, and the Editorial Office, Editors, and reviewers are having to wade through it, and we’re wasting huge sums of money in the process. The time has come to close the floodgates and require that authors demonstrate their commitment to quality science before we let them in the door.