Dantian · Editorial
CORRECTIONS. stated ex ante, held accountable ex post.
How Dantian research is verified, cited, and corrected.
Standards for original research
- First-person evidence over secondary citation. When a claim is made, the grounding should come from direct experience or primary data (markets pulled live, transactions observed, code shipped). Secondary citations support context, not load-bearing claims.
- Hypotheses stated ex ante. Predictions are written with falsification criteria inside the same piece. Every research article that makes a quantitative claim includes a "what would prove this wrong" section.
- Named sources only. Every external fact is linked to a primary source with a publish date. If the source is behind a paywall or a login, the piece says so.
- No anonymous composites. Research is signed. If multiple contributors are involved, all names appear on the byline.
Guest-contributor vetting
Guest contributions are invited from named domain experts with an existing public body of work. Before publication:
- The contributor's credentials are verified against at least two independent sources (LinkedIn + an institutional page, or a published book + academic profile).
- Prior work is reviewed for citation integrity; prolific contributors with a history of unattributed claims are not invited.
- The guest piece carries a visible byline, a credentials block, and an explicit "Guest Contribution" tag in the piece body. Editorial line stays with Dantian.
- Guest contributors do not receive editorial say over unrelated Dantian research.
AI use disclosure
Dantian research is written by humans. Language models are used as research assistants, copy editors, and drafting partners. They do not author claims. Every sentence, hypothesis, and citation in a research piece has been reviewed by the named author and is their responsibility.
Correction process
When an error is identified:
- The error is corrected inline in the original article.
- A [Correction] footnote is appended at the bottom of the piece describing what was changed, when, and why.
- The article's
dateModifiedis updated in both the visible header and the schema.orgdateModifiedfield. - Corrections affecting a research claim (not a typo) also bump the methodology version where applicable.
Silent rewrites of substantive claims are not permitted. If something material was wrong, the piece says so.
Data + methodology transparency
For the Dantian AGI Index and any future live-data products:
- Raw sources are listed with live weights on the public page.
- The composite formula is published and versioned (v0.1 → v0.2 → v0.3 …).
- A "What would prove this wrong" section states falsification criteria for the whole approach, not just individual data points.
- Prior values are archived intact on every version bump rather than silently overwritten.
- The JSON feed is available under Creative Commons Attribution 4.0 for anyone to audit.
How to report an issue
Reach out via LinkedIn with the article URL and a description of the error. Substantive corrections usually ship within 72 hours; methodology bumps take up to two weeks.
← Home