# Itcs Topic Selection > Use when deciding whether a theoretical-computer-science result belongs at ITCS or should be routed to STOC, FOCS, SODA, CCC, ICALP, TCC, or a journal, and when distinguishing ITCS from its depth-first siblings by the conceptual-novelty test — is this a new question, model, or connection worth asking, rather than the deepest theorem of the year. - Skill: `thedixitjain/itcs-topic-selection` (Agent Skill) - Install (CLI): `npx skillmds add thedixitjain/itcs-topic-selection` - Raw SKILL.md: https://api.skillmd.com/api/skills/thedixitjain/itcs-topic-selection/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: thedixitjain (https://skillmd.com/u/thedixitjain) - Updated: 2026-09-08 - Page: https://skillmd.com/skills/thedixitjain/itcs-topic-selection --- # ITCS Topic Selection Decide the venue before drafting, and decide it on the axis ITCS actually selects on. ITCS — the **Innovations in Theoretical Computer Science** conference — was founded (as *Innovations in Computer Science*, 2010) to reward **conceptual novelty**: a new model, a new question, a surprising connection, a direction the field had not thought to take. Its sibling flagships STOC and FOCS reward the **deepest, most polished** result on questions the field already agrees are hard. A technically strong paper whose real virtue is that it *finally proves a known hard theorem* is respected and then, often, routed away from ITCS as "not innovative enough" — while a bold, imperfect, genuinely new idea can be accepted here that STOC/FOCS would reject for want of depth. ## The routing question that matters most For most theory papers the decisive question is not "is this good theory?" but **"does this paper give the community a new *idea*, or a stronger *result* on an existing idea?"** - If a reader's main takeaway is *"I hadn't thought to ask that / model it that way / connect those two areas"* — that is an ITCS signal. - If the main takeaway is *"they finally proved / improved / tightened X"* — that is a STOC/FOCS/SODA signal, however elegant. Both are valuable; ITCS is simply the venue whose PC is briefed to weight the first over the second. When your paper has both a new idea *and* deep technical results, you have a genuine choice, and the calendar and community fit break the tie. ## Sibling-venue routing table | Signal in your project | Better home | Why | |---|---|---| | A **new model, definition, question, or connection** with a first result showing it is alive | **ITCS** | The conceptual-novelty venue; "innovation" is the selection criterion, not a bonus | | The **deepest/tightest theorem of the year** on an established question | **STOC / FOCS** | Depth-first general theory flagships; polish and difficulty are rewarded | | A **new or improved algorithm / data structure**, discrete-algorithms flavor | **SODA** | Algorithms-centered; the ACM-SIAM discrete-algorithms community | | A **complexity** result centered on classes, lower bounds, circuit/proof complexity | **CCC** | Computational Complexity Conference; specialist depth | | A broad EATCS-scope algorithms/automata/logic result, single September-ish cycle | **ICALP** | Track A/B breadth; European TCS flagship | | A **cryptography** result with a formal model and reductions | **TCC / CRYPTO / Eurocrypt** | Crypto-specialist venues; TCC especially for foundational crypto | | A result too long, too incremental-over-your-own-prior, or needing full journal treatment | **A journal (JACM/SICOMP/ToC/...)** | No conference deadline or page pressure; simultaneous journal submission is even allowed alongside ITCS | ## Conceptual shapes ITCS rewards - **A new model or definition** — you hand the community an object it did not have (the non-malleable-codes lineage). The definition *is* the contribution. - **A new algorithmic or complexity notion** — a fresh way to say what "solving" means (the pseudodeterministic lineage), reorganizing questions people already cared about. - **A surprising connection** — tools from area X attack a problem in area Y that neither had linked (the "quantum money from knots" lineage). - **A new question at a seam** — a problem two mature areas could have posed together but never did (interactive proofs meet learning; verification meets distributed networks). - **A conceptual reframing or "manifesto-with-teeth"** — a position that changes what questions look worth asking, *backed by at least one theorem* so it is not a mere opinion. ## The three ITCS acceptance tests Sharpen a borderline verdict with three quick tests the PC will apply: - **New-question test:** after the abstract, does a reader think *"I want to work on that"* about a question they had not previously posed? If the reaction is instead *"impressive that they proved it,"* the paper is depth-shaped -> STOC/FOCS. - **Not-empty-not-everything test:** a new model is only interesting if it is neither trivial nor all-powerful. Do you have at least one crisp result (a separation, a surprising possibility, an impossibility) proving the model is alive? Without it, ITCS reads the paper as a definition in search of a result. - **Survives-the-improvement test:** if someone later proves a *far* stronger theorem inside your model, does your *framing* still get credit? If yes, the idea is the contribution (ITCS). If your paper's whole value is the current bound, route to a depth venue where the bound is the point. ## Evidence maturity, ITCS-style Fit is necessary but not sufficient — but note ITCS's maturity ladder runs *opposite* to an empirical venue's. A **preliminary, incomplete, bold** exploration of a genuinely new model is *more* ITCS-appropriate than a fully mopped-up version whose novelty has worn off. What ITCS will not accept is a new-model paper with **no result at all** (a definition with nothing proved) or a "new" question that is actually a known question in disguise. The failure mode here is not "too early" (as at STOC) but "not actually new" — do the related-work delta honestly first (see [`itcs-related-work`](../itcs-related-work/SKILL.md)). ## Cheap reconnaissance before committing ```text [Novelty] can you state your contribution as a NEW noun (a model, a notion, a connection)? -> yes, and it's checkably new in the literature = ITCS-shaped -> it's a better bound/algorithm on an existing noun = STOC/FOCS/SODA [Scope] scan the last 2-3 ITCS programs (dblp conf/innovations, LIPIcs) for kindred papers -> a cluster of "new model" papers in your area = a reviewer pool exists [Calendar] ITCS has ONE deadline, early September, notification November, conference January. Missed it? the next depth-venue deadline may beat waiting a year for ITCS. ``` ## Decision procedure ```text [Takeaway] is the reader's takeaway a new IDEA or a stronger RESULT? [Shape] model / notion / connection / seam-question / reframing -> ITCS lane [Alive] do you have >=1 crisp result showing the model is not empty and not everything? [Sibling] depth on a known question -> STOC/FOCS; algorithms -> SODA; complexity -> CCC; crypto model -> TCC; too long/incremental -> journal (journal submission is even allowed simultaneously with ITCS) [Verdict] ITCS / sibling venue / journal, with a one-line reason naming the new noun ``` Run this before the writing skills; a wrong venue decision wastes every later step. When the verdict is ITCS, continue with [`itcs-workflow`](../itcs-workflow/SKILL.md) for the single-cycle calendar and [`itcs-writing-style`](../itcs-writing-style/SKILL.md) for the innovation-first paper shape. --- **Source:** [`brycewang-stanford/Awesome-Journal-Skills`](https://github.com/brycewang-stanford/Awesome-Journal-Skills) → `ITCS-Skills/skills/itcs-topic-selection/SKILL.md`