A wave of board departures is forcing questions about who actually controls scholarly publishing
In April 2023, more than 40 editors resigned from NeuroImage, a leading brain-imaging journal published by Elsevier, after the publisher refused to reduce the article processing charge from $3,450. The board had asked for a fee below $2,000. The editors then announced a replacement, Imaging Neuroscience, with MIT Press. That single departure became the reference point for a longer argument about what boards can do when publishers set fees and production systems without them.
The arithmetic mattered because $3,450 arrives after a lab has already spent months on experiments, analysis, revisions, and resubmission. Editors called the fee unethical and unsustainable. The resignation did not end the journal; Elsevier continued it under new editors. But the old board moved together, and enough reviewers followed that the replacement journal became a live option in the field.
NeuroImage walked first, and the arithmetic was blunt
Why fee increases produce board resignations is straightforward. An unpaid editor can accept a high fee as a problem for someone else until the waiver requests and reviewer complaints pile up in their inbox. At NeuroImage, the editorial request was not vague: reduce the APC below $2,000. When Elsevier said no, the board treated the fee as a condition of its labour, not a separate publisher concern.
The replacement, Imaging Neuroscience, launched under MIT Press and set a lower APC, reported in Nature and other outlets. The new journal's pitch was simple: the same scope, the same editorial team, and a lower cost to authors. That model has since been discussed as a template by other unhappy boards, though it requires an established field and a university press or society willing to act quickly.
When automated production lands on a board's desk
A colleague I'll call Dr. K served a society journal board for six years. When the publisher moved to an automated proofing queue, she started seeing page proofs where a Greek letter had been rendered as a comma and a species abbreviation was silently corrected to a different genus. She spent two evenings re-checking one proof. Multiply that across a thirty-person board and the invisible labour stops being invisible.
The Journal of Human Evolution, also published by Elsevier, became a sharper public example in late 2024, when editors and advisory board members resigned after production changes left accepted manuscripts without standard copyediting. Reports in Retraction Watch described automated workflows introducing errors in genus names, anatomical terms, equations, and figure symbols. The base rate is that most journals still pay a human copyeditor; the exception is the commercial production pipeline that lets software make the first pass. The difference between those two paths is what boards have begun refusing to absorb.
AI and automation are not the same thing, and publishers use both. A machine-learning tool that flags mismatched statistics in an accepted file can be useful. The problem starts when the automated step replaces the human check instead of supporting it. Our earlier look at generative AI in peer review and the publisher AI deals covers the policy side of that boundary.
Fee pressure did not start with AI
Article processing charges at several major commercial journals have moved from a convenience to a mandatory budget line. A lab that publishes three open-access papers a year at $3,000 to $5,000 each is committing between $9,000 and $15,000 before page fees and colour charges. The alternatives include diamond open-access journals, which charge neither readers nor authors because a funder, a library, a scholarly society, or a consortial agreement absorbs the cost. The Directory of Open Access Journals keeps a filter for journals without APCs, and the list is longer than many researchers assume. The NeuroImage successor, Imaging Neuroscience, operates from the other side of that equation: an established field, a lower APC, and a production line run by a university press.
Fee disputes and workflow disputes often arrive together because automated production is one of the ways a publisher tries to control the cost of handling a manuscript after acceptance. That is the base rate: commercial publishing at the production stage is high-volume and cost-sensitive, and software is cheaper than copyeditors. The exception is the small society journal that still sends every accepted file to a named production editor. Many academic societies are now revisiting their publisher contracts because editorial boards can no longer guarantee that the old production standards will hold.
Photo by Markus Winkler on Unsplash
What this means for your lab
The editorial board departures are episodic, not a formal movement. But the practical risk to authors appears before any mass resignation. A submitted manuscript can be accepted, invoiced, and then pushed through automated checks that alter figures, symbols, species names, or author names after the corresponding author has already approved a clean version.
Four checks are worth making before you submit:
- Ask the journal who copyedits and typesets the accepted file, and whether the editorial board can override automated changes.
- Put the APC in the grant budget at submission, using the higher end of the publisher's listed range, not the promotional rate.
- Keep the final accepted PDF you approved, because the version of record may differ in ways that a correction request cannot easily unwind after publication.
- Check whether the editorial board is stable; a board that resigned at a related journal is a signal about the publisher, not necessarily the field.
Retraction Watch has detailed coverage of several board exits, and the Committee on Publication Ethics maintains guidance on who is responsible for the integrity of the published record. The practical rule is simpler: do not treat the accepted manuscript as final until you see the production proof.
What to ask before the next submission
Before you sign another author agreement, email the journal editor one question: who will copyedit and typeset the accepted manuscript, and can the board override automated changes if the proof comes back wrong? If the answer is vague, consider a society journal or a university-press journal that names a production editor on its site. The research jobs and postdoc positions pages on AcademicJobs are built around institutions where a realistic publication budget matters.
Board resignations get attention because they are rare. Most editors stay, absorb the extra work, and hope the next production update is better. The long-term cost of that hope shows up in people like Dr. K, who stopped accepting new review requests for two months because a single automated proof had already eaten her week. The next step is a specific email, not a general complaint: write it before the next submission goes out.