Bottom line
Editorial decision letter, V0–V5 verification grade, multi-axis scorecard, and a journal-readiness gate tied to the public rubric.
Pith Review
Find what can break your paper before submission.
Upload a manuscript or arXiv ID and get a completed review ticket: decision letter, load-bearing claim ledger, resolver-backed reference checks, and a verification grade. The report is a computation: versioned, re-runnable, and yours to publish or keep private.
$5 per instrument run · usually 3–6 min · private by default · payment never improves the assessment · adverse recommendations are not refunded. Rubric and hard floors are public at /journal/policy.
01What you get
Claims under load
Claim ledger and inventory for every load-bearing statement: type, evidence, status, axioms, and circularity audit.
Integrity checks
Resolver-backed reference integrity first, then AI-artifact flags and overlap candidates from the bibliography and prior-art scan.
What to fix
Stable R-items you can answer one by one, plus major and minor technical comments separated from the decision letter.
Fit & program
Journal-fit tiers, programmatic bottleneck and downstream unlocks: venue fit guidance, never an acceptance probability.
Disagreement & reassessment
Three independent referee passes with synthesizer adjudication, plus an author-response loop that re-runs assessment on your reply.
sign in to request review
- sign in required
- connect ORCID, Apple, or X
- verify an email address
02How it runs
- Review first. This page creates the review ticket. It does not directly submit the paper for a public pre-print record.
- Private by default. The report is visible only to the requester while it is queued, running, and after completion unless the owner explicitly shares it.
- Pre-print submission later. Once the ticket is complete, the owner can choose whether to make it public and enter pre-print eligibility scoring. Directly-submitted pre-prints that publish receive a Crossref DOI pointing at their permanent Pith article page.
03Recent public reviews
04The grade ladder
| Grade | What it means | Tier |
|---|---|---|
| V0 | Prose only; no formal artifacts supplied. | ≙ T0 |
| V1 | Theorem statements or formal names linked; build not checked. | ≙ T0 |
| V2 | Lean files supplied; the review did not reproduce the build. | ≙ T3 |
| V3 | Lean build reproduced and zero-sorry status checked. | ≙ T4 |
| V4 | Lean artifacts plus data or code reproduced. | ≙ T4 + T2 |
| V5 | Independent formal audit passed. | ≙ T4 |
Trust and limitations
- We audit; we never ghostwrite. Pith does not edit your manuscript or draft rebuttals for you.
- The fee buys the instrument run, never the verdict. Payment cannot improve a score; rejections are not refunded.
- Overlap candidates are not plagiarism findings. They flag prior-art overlap for author review.
- AI-artifact flags are not authorship detection. They catch concrete unchecked-generation defects, not stylometry.
- Journal fit is not an acceptance forecast. Fit tiers describe venue alignment, not submission odds.
- How this instrument can be wrong: referee-model sections can miss context, overstate confidence, or fail on scanned PDFs; resolver checks only cover extracted bibliography entries; claim status notes mix deterministic and model judgment. Every section labels its method where possible.