brycewang-stanford EDBT Workflow
Use this as the project-management skill for an EDBT submission. EDBT's defining rhythm is the multiple-cycle rolling model (three cycles a year since 2022), so the first planning act is not "hit the deadline" but "choose the cycle my evidence is ready for." Replace every date with the current host-site timetable; the per-cycle author-feedback and revision windows are decided per edition.
EDBT is a conference whose proceedings are open access on OpenProceedings.org (CC-BY-NC-ND, no APC): it has no standing editor-in-chief, the rotating leadership is the per-edition General and Program Chairs appointed with the EDBT Association, and the host site moves each year (Barcelona 2025, Tampere 2026). Re-check the current host page rather than carrying a name or URL forward.
The cycle structure (one cycle, end to end)
Each EDBT cycle runs the same five-step arc (EDBT 2022 timeline; reverify dates per edition):
[1] Submission upload PDF + artifact to Microsoft CMT by the cycle deadline
[2] Author feedback phase a short window to respond to initial reviews
[3] Notification #1 accept / REVISE / reject
[4] Revised submission if REVISE: upload the revised paper + change letter
[5] Notification #2 accept / reject on the revision
The revise branch is a genuine in-cycle revise-and-resubmit re-read by the same reviewers — plan for it, do not treat it as a formality. Papers accepted in the earlier cycles present at that year's conference; papers accepted in the last cycle roll to the next edition.
Choosing a cycle
| Situation | Cycle choice |
|---|---|
| Evidence complete, artifact built, writing solid | The nearest cycle — do not wait a full third of a year for polish |
| One key experiment still running | The next cycle — a rushed submission burns a revise chance you could earn |
| You want to present at this specific edition | An earlier cycle of that year; the last cycle rolls forward |
| Just rejected from an EDBT track in this format | None for 12 months — the resubmission ban applies (edbt-submission) |
Milestones
- Venue fit + shape: confirm EDBT over SIGMOD/VLDB/ICDE/ICDT and pick Regular / Experiments-&-
Analysis / Vision (
edbt-topic-selection). - Evidence lock: freeze workloads, datasets, baselines, measurements, and the artifact contents.
- Cycle choice: commit to a cycle and back-plan from its deadline.
- Submission: upload the PDF and artifact to Microsoft CMT.
- Author feedback: answer the initial reviews concisely in the feedback window
(
edbt-author-response). - Revise (if issued): execute the revised paper plus a point-by-point change letter within the in-cycle window, re-read by the original reviewers.
- Decision: archive reviews, responses, and every submitted version.
- Acceptance: prepare the OpenProceedings camera-ready, finalize the artifact and its DOI archive, register, and present at the edition your cycle feeds.
Backward plan from a cycle deadline
| Weeks out (heuristic) | Database-systems milestone |
|---|---|
| 10+ | Problem and claims fixed; workloads, datasets, and baselines chosen |
| 8 | System/prototype stable; measurement harness and provenance logging in place |
| 6 | Full evaluation run; results and scope drafted alongside them |
| 4 | Full draft in the host/OpenProceedings template; artifact assembled |
| 3 | Internal mock review by a DB-systems reader; baseline fairness challenged |
| 2 | Scope hardened, related-work delta sharpened, page budget met for the shape |
| 1 | Artifact re-run clean; availability/reproducibility statement final |
| 0 | CMT fields complete; PDF + artifact uploaded before the cycle cutoff |
These offsets are planning heuristics only — anchor every one to the current host-site timetable, never to a previous cycle's calendar.
Cycle-hopping: the honest calendar reality
- A revise you cannot finish in the in-cycle window is not automatically portable to the next cycle — the revision is scored within the cycle. Budget the revision window like a hard deadline rather than assuming a rollover.
- Missing a cycle costs roughly a third of a year, not a whole one — EDBT's biggest scheduling advantage over single-deadline venues. But a rejected paper still owes the 12-month same-format resubmission ban, so a rushed reject is expensive.
- The cycle-to-conference roll is real: a last-cycle acceptance presents at the following edition. If presenting at a specific city/year matters (travel, graduation, funding), target an earlier cycle of that year.
Failure modes by stage
- Submitting into an early cycle with evidence still moving forfeits the revise opportunity you could have earned with two more weeks of measurement.
- Treating the revise as a formality and under-budgeting the in-cycle window turns a winnable revise-and-resubmit into a reject.
- Skipping the mock review forfeits the one chance to hear "your baseline is unfair" or "this scale is unrealistic" from a friend instead of a reviewer.
- Assuming the last cycle presents this year — it rolls to the next edition; plan travel accordingly.
Coordination notes
- Assign one owner for the artifact/reproducibility package and another for the measurement harness; shared ownership is how provenance slips.
- Archive the exact submitted PDF and artifact — the revision round and the change letter must quote them precisely.
Output format
[Current stage] idea / evidence / cycle-chosen / submitted / author-feedback / revise / accepted
[Target cycle] cycle N, deadline <date and source>, presents at <edition/year>
[Critical path] <three tasks that determine readiness>
[Revise risk] <can the likely revision be finished inside the in-cycle window?>
[Owner map] <task -> person or role>