Editors’ Note: Today’s post is by Julia Romano. Julia is a Licensed Professional Counselor in private practice in West Virginia, with graduate training in clinical psychology and yoga therapy. 

I am not a paper mill. I am a Licensed Professional Counselor in private practice in West Virginia, with graduate training in clinical psychology and yoga therapy, three master’s degrees, and a book on postpartum healing published by Singing Dragon in 2024. I see patients every week. I am also, by the standards increasingly being applied across scholarly publishing infrastructure, no one in particular. I have no current institutional affiliation. I have no prior peer-reviewed publications in psychology. When a preprint server’s moderation system runs a database check on my name, it finds nothing. If that search is a gateway to publication, then that gate is closed to me.

My goal in writing this is to share something important with the people who are building these systems. This is what the gate sounds like from the other side — especially right now, as the field moves toward stronger researcher identity verification in response to real and serious fraud. Tim Lloyd’s recent post in this space laid out the case clearly: editorial platforms were designed for trust, not verification, and the rise of paper mills and generative AI has made that trust exploitable at scale. He is right; the fraud problem is real. But Lloyd also named the risk on the other side of the ledger: “we risk replacing an integrity problem with an exclusion problem.”

I am one version of the exclusion problem. It’s likely there are others who do not write essays about it. They feel dejected and powerless — or maybe they just don’t know that The Scholarly Kitchen exists, and that the policies shaping their professional lives are being debated here.

What Happened

Last month I submitted a paper to PsyArXiv — a postpartum screening instrument I have spent several years developing, grounded in a literature review of the existing tools and their gaps. I am not going to re-litigate the paper’s content here, because the content turned out to be beside the point. The moderator rejected the submission without engaging with the work at all. The decision read, in part:

“For submissions in this format, we need to look at the author’s expertise via past publications… As we found no record of previous relevant peer-reviewed publications by the author(s) in a brief search, we cannot accept the work into the repository. This moderation decision is final.”

I understand the need this policy is meeting. Preprint servers operate at volume, with limited moderator time, against a rising tide of fraudulent and AI-generated submissions. A quick check of an author’s publication record is cheap, and it is likely mostly correct: most bad-faith submissions come from identities with no legitimate research history. The past-publication-as-gate policy catches them.

The danger with this particular filter is that it does not distinguish between a paper mill and a practitioner-researcher who has spent years building a clinical tool and simply has not yet entered the publication pipeline. The filter treats both of us identically, because the filter is not evaluating work. It is evaluating pedigree. With a legitimate contribution to share, I stood on the wrong side of that filter. It wasn’t that I was evaluated and found wanting — the substance of my submission wasn’t evaluated at all.

The Unaffiliated Researcher in an Era of Identity Verification

Lloyd’s post describes a world in which scholarly publishing is moving toward stronger mechanisms for confirming that the people submitting research are who they say they are. Validated institutional emails, federated authentication, organizational affiliation checks, publication history as a trust marker — these are reasonable tools for catching fraudulent work flooding editorial workflows at industrial scale.

But each of these tools carries an assumption about what a legitimate researcher looks like, and the assumption is not neutral. A legitimate researcher, in this emerging framework, has an institutional email address, a recognized organizational affiliation, and a publication history verifiable through ORCID or a similar system. They exist, in other words, inside a structure that can vouch for them.

I do not have an institutional email address. I am not currently affiliated with a university — I purposefully stepped out of that realm in order to investigate my field through a different lens. As a first-time contributor to scholarly literature, my publication history is nonexistent by database standards; the clinical experience and intellectual work behind the submission are not. I am a practitioner who built a tool because the existing tools were failing the people I serve, and I tried to put it somewhere clinicians could find it. By every emerging standard of researcher identity verification, I am indistinguishable from noise.

Lloyd acknowledges this risk. He writes that verification systems must offer multiple routes to trust, and that early-career, unaffiliated, and technically limited researchers need equivalent pathways. My case underscores how urgent that qualifier is, because the direction of travel right now is toward systems that will make my experience at PsyArXiv not an aberration, but the default. If publication history becomes a trust marker, then people without publication history or institutional affiliation are by definition untrusted, regardless of their credentials, clinical experience, or the quality of the work they are trying to contribute.

The fraud these systems are designed to catch is real — and the exclusion they produce is also real, and it is not random. It clusters around practitioners rather than academics, clinicians rather than researchers, people whose training crosses disciplinary boundaries, and independent scholars working outside the university system. These are not the populations producing fraudulent research. They are the populations most likely to see what institutional insiders cannot, precisely because they are working at the edges of the field rather than at its center.

A Different Gate

After the PsyArXiv rejection, I submitted the same paper to SocArXiv, the social sciences preprint server that shares PsyArXiv’s OSF infrastructure. SocArXiv accepted it. The paper now has a citable home, and the practical problem is solved.

I raise this not as vindication, but as evidence that the gate is a choice. Two preprint servers on the same platform made different decisions about what to evaluate at the point of entry. One checked my publication record and stopped there. The other evaluated whether the submission was recognizably scholarly and appropriate to the repository. Both are trying to filter out noise. They produce different outcomes for the same submission, and the difference is not inevitable.

This matters because it shows that the tradeoff between integrity and inclusion is not fixed. It is possible to build systems that take fraud seriously without defaulting to pedigree as the primary filter. SocArXiv is a functioning example of a server that does just this.

What the Field Loses

The cost of the current approach is invisible by design, because the people it excludes are the people the system never sees. When a practitioner-researcher submits a paper and is rejected on credentialing grounds without the work being read, that rejection does not appear in any dataset, or generate a retraction or a correction. It does not register as a problem the system needs to solve, because the system was built to produce exactly this outcome. The work simply does not enter the literature, and the field continues without knowing what it missed.

I am not suggesting that practitioner-researchers should be exempt from scrutiny, rather that the people who design these systems should take seriously the possibility that a fraud-prevention infrastructure built entirely around institutional markers of legitimacy will, as a predictable and systematic byproduct, exclude the independent and cross-disciplinary voices the academy most needs. Not because those voices are always right — being right has never been a precondition to publication. But a field that can only hear from people already inside has limited its own capacity to see what it is missing.

The solutions to fraud being built right now need to be tested not only against the question of whether they catch bad actors — but also whether they catch the practitioner in West Virginia sharing a novel screening tool. If the system can’t tell the difference between those two cases, it’s solving for a different problem — who belongs — and calling it quality control.

Author’s note: This essay was drafted with the assistance of Claude, Anthropic’s AI. The argument, clinical experience, and professional judgment are mine; the drafting process was collaborative. This piece is also under peer review at the Journal of Reproductive and Infant Psychology and available as a preprint on SocArXiv (https://doi.org/10.31235/osf.io/bse4j_v3).