FAIR Press exists to make one change stick: that published science is as reproducible as its data allow. Where results can be re-run, they are — before the article earns its reproduction badge, and by anyone after. Where privacy stands in the way, we go as far as we can — anonymised or synthetic twins — and still publish the work, clearly marked for what can and cannot be independently reproduced. The name is the feature list: fair, along three axes.
Fair to the taxpayer
Most research is publicly funded. It should come back to the public open and re-runnable — reusable evidence, not paywalled claims. Everything we publish is open access under CC-BY, with code and data alongside wherever they can be shared, and results re-executed before publication whenever the data allow. Reproducibility is our default and our strongest signal — never a barrier that turns away sound work bound by privacy.
Fair to authors
Authors get a fast, transparent decision, judged on soundness, not prestige or fashion — rigorous replications and null findings are welcomed on equal terms. Authors keep copyright. Review is AI-assisted and fully disclosed, and a named human editor signs every decision: you will know who read your work, and you can hold someone accountable for the verdict. Submission takes minutes, not an afternoon of form-filling — the system reads your manuscript and fills the forms; you verify.
Fair to researchers the fee systems leave behind
Good, reproducible science must never fail because someone cannot afford a publication fee. The big publishers do run waiver programmes — we do not claim to have invented fairness — but the documented gap is the middle-income countries, where a percentage discount on a very large fee still prices researchers out. Our machinery runs at near-zero marginal cost by construction, so our baseline fee can be low for everyone and waived where needed: we close that gap as normal pricing, not selective charity, with need-blind waivers and a decision process that never sees payment status.
How we get there
- Verification built from experience. The AI in our review is not a generic model let loose — we define it in detail: the rules, the instructions, the system itself, an agentic suite that runs our own review skills and encodes two decades of editor-in-chief practice. It checks statistics, code, reproducibility and references — checks most journals cannot staff — running to millions of tokens of analysis (often $8–20 of computation) per submission, with every AI artifact disclosed and labelled.
- Feedback you can use. A minor or major revision comes with the full three-reviewer panel's reports, free — the substance authors need to improve the work. A desk rejection comes with a short letter developing the one decisive issue; the complete multi-reviewer report is then available to purchase for a small fixed fee — not to make money, but because the analysis behind it genuinely runs to millions of tokens.
- A human signature on every decision. Automation drafts; a named editor decides and answers for it.
- Openness as preservation. CC-BY plus author-retained copyright plus independent deposits mean published work never depends on us to survive.
The first title, the Journal of Reproducible Statistics, is open for submissions. Further journals follow the same system — same standards, own voice — as demand proves out. The commitments that bind every FAIR Press journal are on the policies page.