@@ -13,8 +13,8 @@ We welcome contributions to StatsPAI! This document provides guidelines for cont
13135 . ** Testing** : Add test cases or improve test coverage
1414
1515> ** On credit** : every merged PR is credited in ` CONTRIBUTORS.md ` and
16- > the paper's Acknowledgments. Paper ** authorship** is separate and
17- > extended by invitation against objective criteria — see
16+ > the paper's Acknowledgments. Paper ** authorship** is separate —
17+ > extended by invitation at the maintainers' discretion. See
1818> [ Authorship & Acknowledgments] ( #-authorship--acknowledgments ) below.
1919
2020### Getting Started
@@ -216,54 +216,13 @@ credited. Specifically:
216216No invitation is needed. This is the default and applies to all merged
217217contributions — bug fixes, docs, tests, refactors, new estimators.
218218
219- ### Paper Authorship (by invitation, criteria-based )
219+ ### Paper Authorship (by invitation)
220220
221221** By default, code contributors are not co-authors of StatsPAI
222- papers.** The core development team extends authorship invitations to
223- contributors who meet ** at least one** of the following objective
224- criteria, consistent with
225- [ ICMJE] ( https://www.icmje.org/recommendations/browse/roles-and-responsibilities/defining-the-role-of-authors-and-contributors.html )
226- and [ JOSS] ( https://joss.readthedocs.io/en/latest/submitting.html#authorship )
227- authorship norms:
228-
229- 1 . ** New estimator family** — independently designed and implemented a
230- first-class public function family exposed through ` sp.* `
231- (e.g., a new ` sp.xxx ` or a new branch of an existing dispatcher
232- such as ` sp.synth(method=...) ` / ` sp.decompose(method=...) ` /
233- ` sp.dml(model=...) ` ), including reference-parity tests.
234- 2 . ** Reference alignment ownership** — led the numerical alignment of
235- a non-trivial module against Stata / R / published paper numbers
236- (added tests under ` tests/reference_parity/ ` or
237- ` tests/external_parity/ ` that exercise real external output).
238- 3 . ** Sustained core contribution** — ≥ 10 merged PRs ** and** ≥ 2,000
239- net lines of production code (excluding generated files, vendored
240- assets, and pure formatting) in core estimator modules (any module
241- listed under "因果 / 处理效应", "面板 / 结构", or "因果发现 / ML" in
242- ` CLAUDE.md ` ).
243- 4 . ** Paper writing** — drafted or substantially revised sections of
244- the manuscript, participated in response-to-reviewers, and approved
245- the final submitted version.
246-
247- Meeting a criterion makes a contributor ** eligible** for an
248- invitation; the core team issues the invitation in writing and the
249- contributor must accept explicitly. Invited authors are expected to
250- also satisfy standard ICMJE responsibilities: review and approve the
251- final manuscript, and be accountable for the accuracy of the portions
252- they contributed.
253-
254- ### What this means in practice
255-
256- - A one-off bug fix or docs PR → Acknowledgments + ` CONTRIBUTORS.md ` .
257- - Implementing a new estimator with parity tests → eligible for
258- authorship invitation under criterion (1) or (2).
259- - Large-scale code contribution across many PRs → eligible under
260- criterion (3).
261- - Contributions to paper text or reviewer response → eligible under
262- criterion (4).
263-
264- If you are actively working toward authorship eligibility, feel free
265- to open a discussion or email < brycew6m@stanford.edu > ahead of time so
266- scope and criteria can be agreed upon before the work is done.
222+ papers.** The core maintainers reserve the sole right to extend
223+ authorship invitations to contributors whose work they judge to be
224+ exceptional. There is no fixed rubric — the decision rests with the
225+ maintainers and is made case by case.
267226
268227## 📞 Getting Help
269228
0 commit comments