Research integrity
Research integrity is a core requirement of this site, not an afterthought. It is encoded in how the site is built: statuses, disclosures, limitations and claim boundaries are first-class, and several are required by the content model rather than left optional.
Versioning & permanent identity
Every publication is versioned, and consolidated Research Objects carry a permanent identity-ledger id (for example TR-2026-0001) that is never reused. Version history is shown on the publication page so a reader can see what changed and when.
Methodology & contribution
Publications document their methods and, where relevant, author contributions using CRediT roles — conceptualisation, methodology, software, validation, investigation, data curation, writing, visualisation, supervision and funding acquisition. We do not fabricate author contributions.
Reproducibility
Where practical we share the setup, seeds, hardware, software versions and artefacts needed to interpret and reproduce a result. Evaluations state their kind — validation, test, private, out-of-distribution — so a result is never mistaken for a public leaderboard.
What we never fabricate
We do not invent authorship, results, citations, DOIs, venues, peer review, ethics approval or performance numbers. Preprints and non-peer-reviewed material are labelled as such, and metrics appear only where verified, with dataset and version context.
Correcting the record
When we are wrong we correct openly and, where appropriate, publish a correction or retraction notice rather than quietly editing the record. See the corrections and retractions policy for how this works.