Eurosys Review Process

---
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] <enumerable fixes or identity-level objections>
[Ban exposure] <date this paper may legally return if rejected>
[待核实] <this round's rebuttal window, review counts, PC-meeting form>
```