# Contributing to Kairos

Thank you for your interest in **Kairos / 周期天象图**.

Kairos is an open macro-cycle observatory. Contributions are welcome, especially source corrections, evidence contributions, alternative periodization proposals, research questions, and website improvements.

Kairos is evidence-gated. We welcome disagreement, but we do not accept unsupported macro claims.

## 1. Contribution Principles

Kairos follows these principles:

- probability weather map, not date oracle;
- phase / window judgment, not pseudo-precise prediction;
- evidence before narrative;
- source hardness matters;
- contested claims stay contested;
- Watchlist is not confirmed fact;
- K6 remains Candidate Watch unless broad evidence changes that status;
- public research briefs are curated, not raw internal workspace exports.

## 2. What You Can Contribute

Good contribution types:

- **Evidence contribution**: a source-backed fact or dataset relevant to a Kairos open question.
- **Source correction**: a broken link, wrong date, wrong publisher, wrong source type, or claim-source mismatch.
- **Alternative periodization**: a better K4 / K5 / K6 candidate boundary proposal with sources.
- **Counterevidence**: evidence that weakens or falsifies an existing Kairos watch item.
- **Research brief improvement**: clearer wording, stronger source boundary, better rejected overclaims.
- **Website feedback**: layout, mobile, accessibility, copy clarity, broken links, screenshots.
- **Documentation improvement**: README, roadmap, contribution guide, open questions.

## 3. What We Do Not Accept

Please do not submit:

- investment advice or trade recommendations;
- claims that AI is confirmed K6 without broad evidence;
- claims that K6 is confirmed because a market price moved;
- unsourced macro predictions;
- Google search result pages as sources;
- AI-generated summaries as evidence;
- copyrighted book excerpts or private notes;
- raw internal Evidence Pack style worktables;
- material that cannot be legally or ethically reused.

## 4. Evidence Requirements

For evidence contributions, include:

- source title;
- URL / DOI / dataset code / filing link;
- author / publisher;
- publication date or filing period;
- source type;
- claim supported;
- claim challenged;
- why the source is relevant;
- any limitation or conflicting evidence.

Preferred source types:

- official statistics;
- central bank / treasury / government sources;
- SEC filings or company reports;
- peer-reviewed or working papers;
- reputable institutional reports;
- historical primary sources;
- clearly identified secondary sources.

Less preferred:

- media summaries without source links;
- commentary;
- screenshots without source;
- social media posts;
- AI summaries.

## 5. Source Hardness

Use this simple scale:

- `primary`: official data, filings, original paper, direct report, historical document.
- `high-quality-secondary`: edited volume, reputable institutional synthesis, serious historical work.
- `secondary`: media, commentary, analyst summary, blog, non-primary synthesis.
- `raw-search`: search result or external model output that still needs verification.
- `not-evidence`: unsupported claim, broken link, search page, or source mismatch.

Kairos may use secondary sources for orientation, but major public claims should not rely only on weak secondary sources.

## 6. K6 Candidate Rules

K6 Candidate is manually reviewed, evidence-gated, and event-driven.

Do not submit a contribution that treats AI, energy, automation, biotech, or any other candidate as confirmed K6 without evidence across multiple dimensions:

- broad productivity diffusion;
- cross-sector adoption;
- capital expenditure and ROI;
- financing structure;
- sectoral profit migration;
- infrastructure constraints;
- historical comparison and counterevidence.

K6 review is not automatic and not news-flow driven.

## 7. Watchlist Rules

Watchlist items are observations, not confirmed facts.

A useful Watchlist contribution should say:

- what would strengthen the watch item;
- what would weaken it;
- what would kill the hypothesis;
- what sources support or challenge it.

## 8. Public Research Briefs

The `research/` folder contains curated public briefs.

These briefs are not raw internal Evidence Packs. They remove:

- private paths;
- internal workflow notes;
- external-model search traces;
- unresolved internal verification details;
- raw extracted reading notes.

If your contribution is accepted, maintainers may turn it into a curated public brief or add it to an existing brief.

## 9. Pull Requests

Before opening a pull request:

1. Check whether the change is code, documentation, research, or source correction.
2. If it changes a macro claim, include sources.
3. If it changes K6 / AI / Watchlist language, explain how the guardrails are preserved.
4. If it changes content, respect the split license.
5. Keep pull requests focused.

Maintainers may request that a research-heavy PR starts as an issue first.

## 10. Licensing

Kairos uses a split license model:

- website implementation code: MIT License;
- Kairos content, research framing, names, copy, regime language, Watchlist wording, and Evidence Ledger language: all rights reserved.

By contributing, you agree that your contribution may be incorporated into the public Kairos repository under the applicable project license boundaries.

Do not submit content you do not have the right to contribute.

## 11. Review Outcomes

Possible outcomes:

- accepted;
- accepted with edits;
- accepted as source correction only;
- moved to open question;
- labeled `contested`;
- labeled `needs-source`;
- rejected as overclaim;
- rejected as out of scope.

Kairos prefers cautious public claims over dramatic unsupported claims.
