Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› ecta-writing-style
ecta-writing-style
GitHub用于计量经济学论文后期的风格润色，强制实施简洁、正式且包含编号定理的 House Style。优化行文与符号一致性，但不修正证明或模拟结果，需在模型稳定后使用。
Trigger Scenarios
引言动机冗长未尽早陈述结果
假设和定义混杂在正文中未编号
符号不一致、重载或未先定义后使用
文风过于叙事化而非正式学术风格
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill ecta-writing-style -g -y
SKILL.md
Frontmatter
{
"name": "ecta-writing-style",
"description": "Use as a late-stage polish for an Econometrica manuscript to enforce the terse, formal, theorem-numbered house style and clean mathematical exposition. Polishes prose and notation; it does not fix the proof (use ecta-theory-model) or the simulations (use ecta-robustness)."
}
Writing Style (ecta-writing-style)
When to trigger
- The introduction spends pages on motivation before stating any result
- Assumptions and definitions are buried in prose instead of numbered and set off
- Notation is inconsistent, overloaded, or introduced after first use
- The draft reads like an applied paper (long narrative) rather than a formal one
Trigger this late, after the theorems and simulations are stable. Do not polish prose over an unstable proof.
The Econometrica register
Econometrica prose is terse, precise, and formal. The reader is a specialist. Value is
signaled by the result and its proof, not by rhetorical motivation. The journal's house
style is theorem-numbered LaTeX (the de facto standard is amsthm/amsmath), and the
45-page main-text limit (incl. references and appendices; font ≥12pt, line spacing ≥1.5,
margins ≥1.25in) actively rewards economy — verbose motivation costs pages you need for the
proof architecture. This is a sharper formal register than the narrative style that suits
AER / QJE / JPE / REStud. Concretely:
- State the result early. The introduction should reach the contribution — ideally an informal statement of the main theorem — quickly, then explain why it is hard and how the proof works. Avoid a long throat-clearing literature tour up front.
- Definitions and assumptions are displayed, numbered objects, not asides in a sentence. A reader should be able to find Assumption 2 instantly.
- Every symbol is defined before use, used consistently, and not reused for two things. Keep a private notation table to enforce this.
- Say what is proven and what is assumed. Distinguish clearly between an assumption, a claim proven in the paper, and a claim cited from elsewhere.
- Economy. Cut sentences that restate the math in words without adding interpretation. But do give the one sentence of economic / inferential interpretation each key result deserves.
Mathematical exposition
- Theorem environments (Definition / Assumption / Lemma / Proposition / Theorem / Corollary) numbered consistently; cross-reference by number.
- Proof sketches in the body, full proofs in the Supplemental Material when long (the body's 45 pages are scarce) — but the body must convey the architecture and the hard step.
- Display long equations; number only those referenced later.
- Interpretation paragraphs. After a theorem, one short paragraph: what it means, what is surprising, what assumption is doing the work. This is where the economics lives.
Structure of a methods/theory paper
A common, referee-friendly skeleton:
- Introduction (problem, informal main result, relation to closest prior work, roadmap)
- Model / setup and assumptions (numbered)
- Main results (theorems, with proof sketches)
- Monte Carlo / numerical illustration
- (If applicable) empirical application
- Conclusion (brief)
- Appendix / Supplemental Material (full proofs, additional results, Monte Carlo detail)
Checklist
- Contribution / informal main theorem stated within the first ~2 pages
- Definitions and assumptions are numbered, displayed objects
- Every symbol defined before use; no symbol overloaded
- Assumed vs. proven vs. cited is unambiguous throughout
- Each main result followed by one short interpretation paragraph
- Theorem numbering consistent; cross-references by number, not "above/below"
- Full proofs in the Supplemental Material; body conveys architecture + hard step
- Body fits the 45-page limit (incl. references and appendices); secondary material moved to Supplemental Material
- Prose trimmed of restating-the-math-in-words filler
Anti-patterns
- Three pages of motivation before the first formal statement
- Assumptions stated inline in a paragraph so a referee cannot find them
- The same letter used for a parameter and an index in different sections
- "It is well known that ..." doing the work of a citation or a proof
- Verbose narrative that mimics an applied paper's storytelling
- A theorem with no interpretation paragraph — the reader is left to guess the point
- Cross-references like "the theorem above" that break when the paper is reorganized
Output format
【Result stated early?】yes / no — currently first appears p...
【Assumptions numbered/displayed?】yes/no
【Notation conflicts】[...]
【Interpretation paragraphs present?】yes/no — missing after: [...]
【Register】formal-terse / too narrative — fixes: [...]
【Next step】ecta-replication-package
Version History
- 1839142 Current 2026-07-05 12:52
