Five editorial principles
Primary sources first
Technical claims link to the paper of record, normally through a DOI. Reviews provide orientation but do not replace the underlying result.
Publication status stays visible
Journal articles and arXiv preprints occupy separate lanes. We never describe a deposit as peer-reviewed merely because it looks formal.
Geometry is not hardware
We label whether a result is a metric, theorem, matter model, numerical evolution, analogue, measurement or propulsion demonstration.
Constraints and openings get equal scrutiny
A negative result is not dogma; a positive proposal is not a breakthrough until its assumptions and full stress–energy survive inspection.
Corrections are part of the product
Specific corrections with a DOI, equation, dataset or reproducible calculation receive priority.
How sources are classified
| Lane | Minimum condition | How we describe it |
|---|---|---|
| Peer-reviewed | Published in a refereed journal or proceedings venue with a verifiable record | Published result; still open to criticism and correction |
| arXiv preprint | Public deposit with a stable identifier and relevant technical content | Not peer-reviewed; provisional |
| Numerical/code result | Methods and outputs are inspectable; reproducibility noted where available | Evidence about a model, not automatically about nature |
| Experiment | Defined observable, controls, uncertainty and enough method to reproduce | Preliminary until independently replicated |
| Excluded | Predatory venue, unrelated keyword match, missing provenance or marketing-only claim | Not used as evidence |
Classification is not a truth score. Peer review filters some errors but does not guarantee correctness. A preprint can be excellent; its status still must be stated accurately.
Review workflow
Resolve the record
Verify title, authors, venue, year and DOI against the publisher or Crossref record.
Read the claim in context
Identify the metric class, assumptions, observer, speed regime and whether the conclusion is local or global.
Check competing literature
Link rebuttals, limitations and follow-up work next to the proposal they qualify.
Assign evidentiary language
Use “shows,” “constructs,” “claims,” “suggests” and “detects” according to what the source actually did.
Record the review date
Static guides display their review date. The live feed updates separately and retains explicit source labels.
Corrections policy
Send the page, the sentence that should change, the proposed correction and the strongest source supporting it. We prioritize factual errors, broken DOI mappings, publication-status mistakes and omitted peer-reviewed counterevidence.
Useful correction
Specific location, replacement wording, primary source and explanation of the technical consequence.
Insufficient correction
Appeal to popularity, institutional prestige, press coverage or an uncited assertion.
Substantive corrections change the page; they are not silently ignored because they weaken a preferred narrative.
Independence and anonymity
The site is not affiliated with a space agency, university, company or advocacy organization. Its maintainers remain outside the byline by design. This removes an easy authority signal, so the standard must be stricter: claims should survive the DOI trail without relying on identity.
No mailing list, advertising profile or analytics tracker is attached to these guides. Contact messages are stored privately for review and are never published automatically. The text is licensed CC BY 4.0 so it can be quoted, corrected and reused with attribution.
Correct the ledger
Point to the exact claim and bring the strongest source. Especially when we are wrong.