--- name: eurosys-review-process description: Use when reasoning about how EuroSys decides papers — the two independent yearly rounds feeding one April conference, HotCRP double-blind reviewing, the accept/revise/reject outcome triple, the same-season resubmission ban after rejection, and community structures like the Shadow PC that shape who reviews. --- # EuroSys Review Process Use this to model the decision machinery around a EuroSys submission. Facts below are the 2026/2027 cycles as rendered on 2026-07-08 from the EuroSys CFPs and conference pages; per-round mechanics (rebuttal windows, review counts, online PC meeting format) are set inside each HotCRP instance — 待核实 there. ## One conference, two juries Each EuroSys edition accepts papers through two temporally separate rounds — for 2027: spring (papers May 14, 2026 → notification August 21, 2026) and fall (papers September 24, 2026 → notification January 29, 2027). Both cohorts are published in the same proceedings and presented at the same April conference. Strategically this means: - Your competition is round-local; the PC calibrates within the round it reads. - A spring reject cannot simply "try again in the fall": the CFP bars a rejected paper from returning **until the same round of the next edition** — a full year, not one deadline. - A revision offer, by contrast, explicitly bridges rounds: revise-and-resubmit at the *subsequent* deadline, judged against the issued condition list. ## The outcome triple | Outcome | What it means | Author's next move | |---|---|---| | Accept | Into the edition's proceedings | Camera-ready + artifact evaluation | | One-shot revision | Conditional path to the next round | Condition-by-condition execution plan | | Reject | One-year same-season ban | Substantial rework, or a different venue | "One-shot" is literal: the revision is evaluated once, against the conditions, by a process designed for continuity with the original reviews. There is no second revision offer downstream of the first. ## Who is reading - A double-blind HotCRP process: reviewers do not see authors, and conflicts declared at abstract registration drive assignments. - The PC spans the breadth of European and international systems research — kernels, distributed systems, storage, cloud runtimes, virtualization, systems security, and increasingly systems-for-ML. Expect at least one reviewer whose home sub-area differs from yours; the paper's framing must carry them. - EuroSys has repeatedly run a **Shadow PC** (e.g., `2026.eurosys.org/shadow-pc.html`), where early-career researchers review real submissions in parallel as training. Practical consequence: write for a rigorous but possibly junior reader, and expect the community's reviewing norms to be unusually explicit and well-documented. ## What moves a PC discussion here 1. **Verifiable evaluation fairness** — tuned baselines, named versions, disclosed hardware — because the deepest systems-review instinct is to distrust a benchmark table. 2. **A mechanism the reviewer can retell** — if the design section cannot be summarized in three sentences at the PC meeting, no champion can defend it. 3. **Claims scoped to evidence** — a paper that says where it loses earns the credibility spent by every paper that doesn't. 4. **Deployment plausibility** — EuroSys values systems that could run outside the lab; a purely synthetic setting needs an explicit bridge to practice. ## A round from the author's seat Between the paper gate and notification (roughly fourteen weeks in the 2027 spring round: May 14 → August 21), the author-visible surface is nearly silent. What happens behind it, in the standard HotCRP shape: 1. Bidding and assignment off the registered topics and conflicts. 2. First-round reviews; possibly a cut before additional reviews are commissioned on papers still alive. 3. Rebuttal or author-reviewer interaction if the round runs one (待核实 in the round's HotCRP announcements). 4. Online PC discussion converging each paper into one of the three bins, with revision-condition lists drafted for the middle bin. 5. Notification, reviews, and — for the middle bin — the condition list. The author-controllable inputs are all front-loaded: the registered abstract (bidding quality), the conflicts (assignment sanity), and the paper itself. Mid-round, the only lever is the response phase, if any. ## Conduct boundaries - No reviewer contact outside the platform, no identity fishing, no parallel submission of the same work to another archival venue. - Public talks and preprints during review are governed by the round's anonymity wording — 待核实 before scheduling a seminar on the submitted work under its submission title. ## Reading your reviews structurally ```text For each review, extract: score-driver: the sentence that explains the score (rarely the summary) fixability: bounded experiment/rewrite -> revision-compatible identity-level objection -> reject-compatible round-awareness: does the reviewer propose revision conditions? Then ask: across reviewers, is there a coherent, enumerable condition list? yes -> argue toward the revision bin in the rebuttal (if the round has one) no -> maximize the accept case on the strongest review's terms ``` ## Output format ```text [Round position] spring / fall, and days to its next gate [Outcome forecast] accept / revision / reject, with the driving signal [Condition-list coherence] [Ban exposure] [待核实] ```