brycewang-stanford ECAI Topic Selection
Routing is the highest-leverage decision, made before writing. ECAI is EurAI's general-AI flagship — broad by design, spanning symbolic reasoning, KR, planning, search, multi-agent systems, ML, and applications. That breadth is a feature and a trap: many projects could go to ECAI, so the real question is whether ECAI is the best-matched home for your contribution and calendar, and whether it belongs in the main track or the co-located PAIS applications track.
Is it ECAI-shaped?
ECAI wants a contribution a general AI audience recognizes as advancing how systems reason, plan, learn, or decide — with evidence proportional to the claim (a proof, or a fair comparison). Apply two tests:
- Model-swap test. If you replaced the underlying model/solver with another, would a lasting AI lesson remain? If yes → a genuine AI contribution (ECAI-shaped). If no → likely a pure-ML paper wearing an AI title (route to an ML venue).
- General-audience test. Could a reviewer from a different AI subfield see why the contribution matters? ECAI's broad pool rewards work that travels beyond one niche.
ECAI vs the sibling flagships
| If your contribution is... | Prefer | Because |
|---|---|---|
| Broad AI (symbolic + learning), European flagship, general audience | ECAI | EurAI's general-AI venue; breadth is the point |
| The same, but you want the international flagship / its calendar | IJCAI | In 2026 they are the same event (joint IJCAI-ECAI); in other years, different calendars/organizers |
| The same, but on the AAAI calendar / North-American flagship | AAAI | General AI too; different body, dates, template |
| Fundamentally about multi-agent interaction, mechanisms, negotiation | AAMAS | Dedicated MAS pool and reviewers |
| Pure knowledge representation / reasoning theory | KR | Specialist KR audience and depth |
| A benchmark-driven ML advance with no symbolic/decision lesson | NeurIPS/ICML | Pure-ML venues reward the leaderboard framing ECAI does not |
| A deployed, real-world AI application | ECAI → PAIS | PAIS is ECAI's prestigious-applications track |
Because these venues overlap, decide by best-matched reviewer pool + calendar + community pull, not by prestige alone. A paper matched to a specialist pool (AAMAS/KR) can fare better there than in ECAI's broad pool.
The 2026 twist: ECAI is joint with IJCAI
For this cycle, ECAI 2026 = IJCAI-ECAI 2026 (Bremen, 15-21 Aug 2026). So "ECAI vs IJCAI" is not
a choice in 2026 — it is one submission to the joint conference, following IJCAI's process,
template (ijcai.sty), 7+2 page budget, and open-access IJCAI proceedings. The standalone-ECAI
FAIA/ecai.cls/PAIS identity returns in a non-joint year. Confirm which regime your target edition
is in before routing (ecai-submission, ecai-camera-ready).
Main track vs PAIS
Route to PAIS (Prestigious Applications of Intelligent Systems, co-located with a standalone ECAI) when the contribution is a credible real-world deployment whose value is the application, not a new general mechanism:
- Main track: the contribution is a new technique, guarantee, or understanding that generalizes.
- PAIS: the contribution is a system that works in the real world — the lesson is the deployment, the constraints handled, the impact demonstrated.
PAIS has its own call and page budget (ECAI 2024: 7+1) and rewards impact and real conditions over
benchmark abstractions (ecai-experiments). Note PAIS is a standalone-ECAI fixture; verify
whether an applied track exists in a joint IJCAI-ECAI year (special tracks like AI4Tech, AI &
Robotics, AI & Health, AI & Social Good may serve that role — 待核实 per cycle).
Calendar-driven routing
- ECAI/IJCAI/AAAI run on different calendars (except the joint years). Route partly by which
deadline your work will actually be ready for — a rushed ECAI submission with a gap has no
revision round to save it (
ecai-review-process). - The two-deadline front (abstract, then paper) means you commit the title/abstract a week before the PDF; do not register an ECAI abstract you are not confident you can deliver.
Red flags that you are at the wrong venue
- The lesson evaporates under the model-swap test → ML venue.
- Only a specialist could care → AAMAS/KR.
- The value is a deployment, not a mechanism → PAIS.
- You are choosing ECAI over IJCAI in 2026 → they are the same event this year.
Output format
[Contribution shape] technique/guarantee / empirical understanding / multi-agent / KR theory / deployed system
[Model-swap test] AI lesson survives model swap? yes/no
[Venue] ECAI (main) / ECAI-PAIS / IJCAI / AAAI / AAMAS / KR / ML venue — with the deciding reason
[2026 regime] standalone ECAI or joint IJCAI-ECAI (same event as IJCAI this year)
[Calendar fit] ready for this deadline given no revision round? yes/no