Support
Help & support
How a review works, what the rating means, and what stays private. If your question is not here, write to the editorial office - we usually reply the same business day.
Running a review
The main thing this site does: read a paper closely and tell you what does not hold up.
What does a review actually check?
Eight dimensions of rigor, across sixteen separate checks. The statistics are recomputed rather than read back to you; citations are resolved against the sources they claim; figures are screened for integrity problems; data and code links are followed to see whether they resolve; and every consequential claim is tested against the evidence offered for it. You get concrete findings, each ranked by how much it matters, with the quote it came from. Run one, or read reviews of papers you already know.
What can I give it?
A DOI or PMCID, a manuscript pasted as text, or a PDF, DOC or DOCX up to 50 MB. The paste box counts down against its limit as you type. Uploads need an account; an identifier or pasted text does not. A paper with no full text available reaches the engine as an abstract only, which limits what can honestly be checked - the report says so when that happens.
How many reviews do I get?
One free review per 24 hours, whether or not you are signed in. The next one unlocks 24 hours after your last, not at midnight. Signed in, a run that fails does not count against it; signed out it can, because the free look is counted when the paper is accepted rather than when the report lands - one more reason to make an account.
How long does it take?
Usually one to three minutes. You do not have to wait on the page: the report continues without you and appears under your reports on the review page when it lands.
Do I need an account?
Not for your first review. You get the same engine, the same checks and the same report as a signed-in author, at its own private link. An account adds file upload, your report history, the ability to publish a record, and Formal Expert Review - each of which needs an owner who can stand behind it.
My review failed, or came back thin.
Almost always the full text could not be reached, so there was little to read. Run it again on the manuscript itself rather than the identifier: paste the text, or upload the PDF if you have an account. Whether the failed run cost you the day’s review depends on whether you were signed in - see How many reviews do I get? above. If it keeps failing, send us the identifier at editorial@alpha1science.com.
Reading the report
What each number is, and - as important - what it is not.
What do the stars mean?
Every paper starts at five stars and loses them to what is found: a serious problem costs a full star, a medium one half, a minor one a quarter. Five is exemplary, four strong, three adequate, two weak, one critical. It never falls below one. The itemised arithmetic is one click away, under How this rating was calculated - a rating you cannot audit is one you should not trust.
Critical and high-priority - what is the difference?
How much the finding changes what you should believe. A critical finding undermines a headline result: a statistic that does not recompute, a claim the data cannot carry. A high-priority one is a real problem to fix that does not by itself overturn the paper. Both are listed with the evidence they rest on, so you can disagree with us specifically rather than generally.
Why does it say some checks did not run?
Because they did not, and you are owed that plainly. A check with nothing to work on - no figures to screen, no registry entry to reconcile - or one that could not reach a source reports itself as such, and is never quietly counted as clean. That is why the report carries both a Checks line, counting what ran against what applied to this paper, and a Not verified in this report list naming what did not. A clean report with four checks missing is a weaker result than a clean report with none, and it should look like one.
What is "model-reported confidence"?
The model’s own assessment of its verdicts. It is not a calibrated probability and we do not present it as one. How often the engine is actually wrong is published, per version, on the Kaimen Ledger.
Is this peer review?
No. It is a fast, model-assisted rigor check, and it is deliberately not sold as a substitute for expert judgement. When you need named human experts on the record, that is Formal Expert Review.
Publishing, privacy and the record
Nothing you run becomes public unless you decide it should.
Who can see my report?
You, and our administrators. A review is unlisted, at a link nobody else can guess, until you publish it yourself - publishing is a separate, explicit step with its own confirmation. Administrators can open an unpublished report because somebody has to be able to answer a support request and investigate a bad run; editors and reviewers cannot.
What happens if I publish it?
The report gets a public URL anyone can read and is listed on the Rigor Showcase. The paper’s title, its authors and every finding go public, including the critical ones. It becomes a citable, tamper-evident record that others can verify.
Can I take a published record back?
Yes. Make private on the report removes it from the showcase and takes its public page down straight away. Anything already copied or cited elsewhere is, of course, out of our hands.
Formal Expert Review
When a model-assisted check is not enough and you need people who will sign their name.
How does it work?
You pay up front. We deliver at least two independent reviews from paid, domain experts within 21 days, or you get your money back. You keep copyright and can submit anywhere. The fee is shown before you commit, and the full terms are in the service delivery policy.
How do I request one?
From a finished report, use Request expert review; it carries the review you already have across, so the experts start from it rather than from nothing. Or begin at Formal Expert Review.
Your profile and account
A profile is optional. It is what makes your reviewed work findable under your name.
How do I claim my researcher profile?
Sign in, then go to /onboard to find and claim your OpenAlex or ORCID profile. Publications, citations and co-authors come across automatically.
How do I edit my profile?
Open your profile and use Edit profile in the header. Every section is editable in place.
Where do my degrees and affiliation dates come from?
Your verified ORCID record. You can refresh from ORCID, or edit the rows in Affiliations & Education directly.
Something on my profile is wrong and I cannot change it.
Most fields are owner-editable from the profile page. For anything that came from upstream (OpenAlex, ORCID, NIH RePORTER), write to editorial@alpha1science.com and we will correct it.
How do I delete my account?
Email editorial@alpha1science.com from your sign-in address. We reply within two business days and confirm before anything is deleted.
Read further
Still stuck? Talk to a human.
The editorial office handles reviews, expert review, payments, profiles and account changes. Send links, screenshots, and what you expected to happen.