PerCom Workflow
Use this as the project-management skill for a PerCom submission. Replace every date with the
current official timetable and work backwards from the HotCRP registration cutoff — which
precedes the paper deadline by about a week. PerCom runs a single annual deadline feeding one
review round with a bounded rebuttal and an early-rejection gate; there is no journal-style
revision, so plan for the paper to be complete at submission.
PerCom is an IEEE Computer Society conference: it has no standing editor-in-chief and no APC
in the open-access sense. Rotating leadership is the per-edition General and TPC (Program) Co-Chairs
(PerCom 2027 full roster 待核实 as of 2026-07-09), and the cost model is IEEE registration with
at least one author presenting in person.
Milestones
- Venue fit: confirm PerCom over ACM UbiComp/IMWUT, MobiCom, SenSys, or IPSN, and confirm the
paper is ready this cycle (
percom-topic-selection).
- Data collection lock: IRB/consent secured, subjects recruited, sensing protocol frozen — the
long pole for a human-subjects paper.
- Evidence lock: freeze claims, cross-subject splits, baselines, metrics, limitations, and the
dataset contents. Nothing can be added in the rebuttal.
- Registration: submit real title, abstract, authors, topics by the earlier registration
deadline (PerCom 2027: 4 Sep 2026).
- Submission: upload the anonymized PDF and the open-data plan (PerCom 2027: 11 Sep 2026 AoE).
- Early-rejection gate: notification (~20 Nov 2026) either early-rejects or invites a rebuttal.
- Rebuttal (if invited): answer explicit reviewer questions from existing evidence, anonymously
(~26 Nov 2026).
- Decision: accept/reject (~18 Dec 2026); archive reviews, rebuttal, and every submitted
version.
- Acceptance: prepare the IEEE Xplore camera-ready (PDF eXpress, eCopyright, ORCID), deposit the
de-identified dataset, register, and present in person.
Backward plan from the submission deadline
| Weeks out (heuristic) |
Human-subjects sensing milestone |
| 16+ |
IRB/consent approved; recruitment and data collection begun (the long pole) |
| 12 |
Data collection complete; labeling protocol run with annotator agreement |
| 10 |
Claims fixed; cross-subject (LOSO) splits and baselines chosen |
| 8 |
Experiments complete; F1/event-level results with per-subject spread computed |
| 6 |
Analysis done; limitations drafted alongside results; dataset de-identified |
| 4 |
Full draft in the IEEEtran 9+1 template; artifact assembled and anonymized |
| 3 |
Internal mock review by a ubicomp reader ("is this within-subject only?") |
| 2 |
Limitations hardened, related-work delta sharpened, 9-page budget met |
| 1 |
Double-blind sweep on PDF and dataset link; open-data plan final |
| 0 |
Registration by its earlier cutoff, then paper + dataset link on HotCRP |
These offsets are planning heuristics only — anchor every one to the current Important Dates page,
never to a previous cycle's calendar. Note the 16-week IRB/collection pole: human-subjects data
cannot be rushed in the final month.
Cycle reality: one shot per year, gated early
PerCom runs one annual research deadline. As of 2026-07-09, PerCom 2026 (Pisa) has concluded, so the
live next target is PerCom 2027 (Goa; deadline 11 Sep 2026). Two planning consequences:
- There is no revision cycle. A paper that is not complete — especially one lacking cross-subject
evidence — cannot be rescued in the rebuttal; it will likely be early-rejected. Finish the
evaluation before submitting.
- Missing the annual deadline costs roughly a year at PerCom; ACM IMWUT has rolling
quarterly deadlines and a revise cycle, so a paper that needs another round of experiments may be
better routed there than held for a year — factor that into routing rather than idling.
Failure modes by stage
- IRB or recruitment slipping pushes data collection into the final month and forces a
within-subject-only evaluation — the classic PerCom early-reject in the making.
- Discovering at week 3 that the split was within-subject is fatal with no time to recollect;
design the LOSO split at week 10.
- Leaving the anonymized dataset to the final week is how a testbed name, a re-identifying photo,
or an owner string leaks identity under double-blind review.
- Skipping the mock review forfeits the one chance to hear "your evaluation is within-subject"
from a friend instead of a reviewer — and there is no revision to fix it.
- Treating the rebuttal as a revision and planning to "add the experiment then" — the process
neither expects nor permits it.
Coordination notes
- Assign one owner for IRB/consent and the dataset, and another for the double-blind anonymity
sweep; shared ownership is how both slip.
- Archive the exact submitted PDF, dataset, and (later) rebuttal — the discussion must quote them
precisely and stay anonymous.
Output format
[Current stage] idea / IRB+collection / evidence / writing / registration / submitted / rebuttal / accepted
[Next official deadline] <date and source, or unknown>
[Critical path] <three tasks that determine readiness -- IRB/collection is usually one>
[Risk register] <IRB / cross-subject evaluation / 9-page budget / anonymity / dataset / presentation>
[Owner map] <task -> person or role>
1---2name: percom-workflow3description: Use when planning an IEEE PerCom project timeline from venue fit through paper registration, the September submission, the early-rejection gate, the bounded rebuttal, the IEEE Xplore camera-ready, and in-person presentation, with backward-planning offsets for a human-subjects sensing paper and honest handling of the single-annual-deadline cycle.4---56# PerCom Workflow78Use this as the project-management skill for a PerCom submission. Replace every date with the9current official timetable and work backwards from the HotCRP **registration** cutoff — which10precedes the paper deadline by about a week. PerCom runs a **single annual deadline** feeding one11review round with a **bounded rebuttal** and an **early-rejection gate**; there is no journal-style12revision, so plan for the paper to be complete at submission.1314PerCom is an IEEE Computer Society conference: it has **no standing editor-in-chief** and **no APC**15in the open-access sense. Rotating leadership is the per-edition General and TPC (Program) Co-Chairs16(PerCom 2027 full roster **待核实** as of 2026-07-09), and the cost model is IEEE registration with17at least one author presenting **in person**.1819## Milestones2021- **Venue fit:** confirm PerCom over ACM UbiComp/IMWUT, MobiCom, SenSys, or IPSN, and confirm the22 paper is ready this cycle (`percom-topic-selection`).23- **Data collection lock:** IRB/consent secured, subjects recruited, sensing protocol frozen — the24 long pole for a human-subjects paper.25- **Evidence lock:** freeze claims, cross-subject splits, baselines, metrics, limitations, and the26 dataset contents. Nothing can be added in the rebuttal.27- **Registration:** submit real title, abstract, authors, topics by the earlier registration28 deadline (PerCom 2027: 4 Sep 2026).29- **Submission:** upload the anonymized PDF and the open-data plan (PerCom 2027: 11 Sep 2026 AoE).30- **Early-rejection gate:** notification (~20 Nov 2026) either early-rejects or invites a rebuttal.31- **Rebuttal (if invited):** answer explicit reviewer questions from existing evidence, anonymously32 (~26 Nov 2026).33- **Decision:** accept/reject (~18 Dec 2026); archive reviews, rebuttal, and every submitted34 version.35- **Acceptance:** prepare the IEEE Xplore camera-ready (PDF eXpress, eCopyright, ORCID), deposit the36 de-identified dataset, register, and present in person.3738## Backward plan from the submission deadline3940| Weeks out (heuristic) | Human-subjects sensing milestone |41|---|---|42| 16+ | IRB/consent approved; recruitment and data collection begun (the long pole) |43| 12 | Data collection complete; labeling protocol run with annotator agreement |44| 10 | Claims fixed; cross-subject (LOSO) splits and baselines chosen |45| 8 | Experiments complete; F1/event-level results with per-subject spread computed |46| 6 | Analysis done; limitations drafted alongside results; dataset de-identified |47| 4 | Full draft in the IEEEtran 9+1 template; artifact assembled and anonymized |48| 3 | Internal mock review by a ubicomp reader ("is this within-subject only?") |49| 2 | Limitations hardened, related-work delta sharpened, 9-page budget met |50| 1 | Double-blind sweep on PDF and dataset link; open-data plan final |51| 0 | Registration by its earlier cutoff, then paper + dataset link on HotCRP |5253These offsets are planning heuristics only — anchor every one to the current Important Dates page,54never to a previous cycle's calendar. Note the **16-week IRB/collection pole**: human-subjects data55cannot be rushed in the final month.5657## Cycle reality: one shot per year, gated early5859PerCom runs one annual research deadline. As of 2026-07-09, PerCom 2026 (Pisa) has concluded, so the60live next target is **PerCom 2027** (Goa; deadline 11 Sep 2026). Two planning consequences:6162- **There is no revision cycle.** A paper that is not complete — especially one lacking cross-subject63 evidence — cannot be rescued in the rebuttal; it will likely be early-rejected. Finish the64 evaluation before submitting.65- **Missing the annual deadline costs roughly a year** at PerCom; ACM **IMWUT** has rolling66 quarterly deadlines and a revise cycle, so a paper that needs another round of experiments may be67 better routed there than held for a year — factor that into routing rather than idling.6869## Failure modes by stage7071- **IRB or recruitment slipping** pushes data collection into the final month and forces a72 within-subject-only evaluation — the classic PerCom early-reject in the making.73- **Discovering at week 3 that the split was within-subject** is fatal with no time to recollect;74 design the LOSO split at week 10.75- **Leaving the anonymized dataset to the final week** is how a testbed name, a re-identifying photo,76 or an owner string leaks identity under double-blind review.77- **Skipping the mock review** forfeits the one chance to hear "your evaluation is within-subject"78 from a friend instead of a reviewer — and there is no revision to fix it.79- **Treating the rebuttal as a revision** and planning to "add the experiment then" — the process80 neither expects nor permits it.8182## Coordination notes8384- Assign one owner for IRB/consent and the dataset, and another for the double-blind anonymity85 sweep; shared ownership is how both slip.86- Archive the exact submitted PDF, dataset, and (later) rebuttal — the discussion must quote them87 precisely and stay anonymous.8889## Output format9091```text92[Current stage] idea / IRB+collection / evidence / writing / registration / submitted / rebuttal / accepted93[Next official deadline] <date and source, or unknown>94[Critical path] <three tasks that determine readiness -- IRB/collection is usually one>95[Risk register] <IRB / cross-subject evaluation / 9-page budget / anonymity / dataset / presentation>96[Owner map] <task -> person or role>97```
brycewang-stanford