itcs-supplementary
GitHub指导ITCS论文内容在正文、附录和外部全版本间的分配策略。核心原则是：所有影响录用判断的关键证明必须包含在提交的PDF中，禁止将决定性内容仅置于arXiv等外部版本，以确保评审人能独立验证。
Trigger Scenarios
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill itcs-supplementary -g -y
SKILL.md
Frontmatter
{
"name": "itcs-supplementary",
"description": "Use when splitting an ITCS paper's content between the body, the appendix, and the arXiv\/ECCC\/ePrint full version — deciding by what a reviewer must verify, so no central proof lives only in a preprint the PC is not obligated to read, while keeping the first-10-pages merits window clean."
}
ITCS Supplementary
ITCS gives you three places to put content — the body, the appendix (part of the submitted PDF), and the full version on arXiv/ECCC/ePrint — and the split is governed by one rule: anything a reviewer must read to judge the paper lives inside the submitted PDF. Because ITCS has no hard page limit and no rebuttal, the appendix is generous but load-bearing: you cannot promise "details in the full version" for a proof that decides acceptance, and you cannot answer later if a reviewer cannot follow. This skill decides what goes where.
The governing principle: decision-criticality
Sort every piece of content by whether a reviewer needs it to decide:
- Body (first ~10 pages): everything needed to grasp the contribution and judge its
novelty — the model, the motivation, the main theorem statements, proof ideas, and the
novelty delta. This is the merits window the PC actually reads (see
itcs-writing-style). - Appendix (same PDF): the complete proofs of central claims, full definitions, and any technical development too long for the body. This is not optional reading — it is where the "complete proofs of all central claims" requirement is satisfied.
- Full version (arXiv/ECCC/ePrint): the permanent, exhaustive record — extended discussion, additional corollaries, alternate proofs, and the same central proofs mirrored. It may contain more, but nothing decision-critical may live only here.
The rule that trips people up
No central proof may live only in the external full version. The PC is not obligated to open your arXiv posting. A reviewer who reaches "the proof of our main theorem is in the full version [ref]" and stops there will — correctly — treat the claim as unproved in the submission. Put every load-bearing proof in the submitted PDF's appendix.
Deferring to the full version is acceptable only for genuinely peripheral content: a fourth corollary, a longer historical discussion, an alternate proof of an already-proved result.
Use the no-page-limit generously but honestly
ITCS's lack of a hard limit is a gift with a catch:
- Do put full proofs in the appendix rather than compressing them into unreadability. A dense, unverifiable body proof is worse than a clear appendix proof with a body sketch.
- Do not pad the merits window. The first ~10 pages are judged on clarity of the idea; extra length there dilutes the innovation. Push technical bulk to the appendix, not into the intro.
- Signpost precisely. Every deferred proof gets an explicit "proof in Appendix C" and every appendix section says which body claim it discharges. A reviewer should never hunt.
Content-placement table
| Content | Body (~first 10pp) | Appendix (same PDF) | Full version only |
|---|---|---|---|
| Model & definitions | Yes (core ones) | Full/edge-case variants | — |
| Main theorem statements | Yes | — | — |
| Proof ideas / sketches | Yes (a paragraph each) | — | — |
| Complete proofs of central claims | Idea only | Yes (required) | Mirror allowed, not sole home |
| Novelty delta / related work | Yes (the deltas) | Extended comparison | — |
| Routine calculations | Reference | Yes | — |
| Peripheral corollaries, alt proofs | — | Optional | Yes |
| Finite separating objects (graphs/codes) | Small ones inline | Larger ones | — |
Anonymity and the full version
Posting a full version is encouraged under lightweight double-blind, but keep the split
consistent with the anonymity strategy (see itcs-submission):
- The submitted PDF (body + appendix) omits author names, affiliations, and emails.
- The arXiv/ECCC/ePrint full version may be non-anonymous — that is permitted and normal.
- Do not let a cross-reference in the submission point at the named preprint in a way that reads as a deliberate identity reveal; a neutral "a full version is available" is enough.
Output format
[ITCS content-split status] clean / decision-critical-content-misplaced
[Merits window] model+motivation+results+deltas within ~10 pages, no padding? yes/no
[Appendix] complete proofs of ALL central claims present in the submitted PDF? yes/no
[Full-version leak] any load-bearing proof deferred ONLY to arXiv/ECCC/ePrint? list them
[Signposting] every deferred proof cross-referenced to its appendix section? yes/no
[Fix queue] <ordered; move any decision-critical content into the PDF first>
Version History
- 9f86f09 Current 2026-07-19 16:06
