SIGMETRICS Related Work
Use this to audit novelty and eligibility. SIGMETRICS reviewers are close to the
performance-evaluation literature and expect to see where your paper sits relative to the nearest
prior model, bound, or measurement — stated as a delta, not a list. Reopen the current call for
the simultaneous-submission and prior-publication rules (a paper under one-shot revision counts as
under submission) before advising authors.
Positioning checks
- Separate the analytic/measurement novelty from the engineering effort. What is new: a tighter
bound, a more general model, a policy that provably beats a known one, a measurement of a system
nobody had characterized, or a learning algorithm with a new guarantee?
- Cover the performance-evaluation lanes (see the table), not just the papers nearest your
method. A bibliography missing the obvious queueing-theory predecessor or the prior measurement of
the same system reads as unaware.
- Write delta-first. Each closely related paper gets one sentence naming what it did and one
naming what you do differently — a tighter bound, a weaker assumption, a broader policy class, a
larger/newer measurement — not a summary.
- Preserve double-anonymity. Cite your own prior work in the third person and never link
reviewers to an identity-revealing preprint, system page, or repository (Operational Systems Track
excepted).
- Declare overlap with any prior conference/workshop version or concurrent submission; do not
re-submit archival work as new.
Performance-evaluation literature lanes
| Lane |
Typical venues |
What SIGMETRICS reviewers check |
| Core performance evaluation |
SIGMETRICS/POMACS, Performance Evaluation, QUESTA |
Whether the nearest model/bound/measurement is compared or distinguished |
| Systems (when you claim a systems payoff) |
NSDI, OSDI, SIGCOMM, ATC |
Whether the system you improve/measure is credited and fairly baselined |
| Measurement |
IMC, PAM, INFOCOM |
Whether prior measurements of the same system/workload are engaged |
| Learning (Learning track) |
NeurIPS, ICML, COLT |
Whether the learning-theoretic predecessor (regret bounds, algorithms) is cited to its origin |
| Networking/queueing journals |
IEEE/ACM TON, QUESTA, Stochastic Models |
Whether deeper journal-length analyses of the model are engaged |
A bibliography that cites only your own subarea tells a reviewer the delta may be smaller than
claimed; one that reaches the neighboring theory, systems, and measurement venues signals command of
the field.
Delta-first positioning vignette
Suppose the paper proves a tail-latency bound for a rank-based scheduler. Its nearest neighbors: a
prior analysis of a single age-based policy (one policy, mean latency), a general scheduling
framework (broad class, but no tail bound), and a measurement study of the target system (data, no
policy analysis). The novelty sentence should name all three contrasts — a tail bound where the
single-policy analysis gave only mean, a provable tail guarantee where the framework gave none,
and a policy with analysis where the measurement gave only characterization.
Concurrent and prior-version judgment calls
[Concurrent arXiv work] cite neutrally, state the technical difference (tighter bound? weaker
assumption? newer measurement?), avoid unverifiable priority claims;
keep the citation double-anonymous
[Your workshop version] usually non-archival and citable, but confirm against the current call
wording and phrase so anonymity survives
[Prior short version] declare the overlap and state what the full paper adds (proofs, validation)
[Paper under one-shot revision] it is under submission to SIGMETRICS -- do not submit it elsewhere
before withdrawing
Eligibility red flags
- Substantial text/result overlap with a published paper by the same authors (self-plagiarism risk).
- A "new" analysis that re-derives a known bound without a tighter result or weaker assumption.
- Citations exclusively to non-performance-evaluation venues, signaling the paper may be a systems
or learning paper rerouted without reframing.
Output format
[Eligibility] clear / needs declaration / risky
[Lanes covered] <performance-eval / systems / measurement / learning / journals>
[Nearest 3 works] <work -> one-line delta (tighter bound / weaker assumption / broader class / newer data)>
[Archival-overlap risk] <none / declare: what>
[Novelty sentence] <SIGMETRICS-ready contribution contrast against the nearest prior work>
1---2name: sigmetrics-related-work3description: Use when positioning an ACM SIGMETRICS submission against the performance-evaluation literature across SIGMETRICS/POMACS, Performance Evaluation, QUESTA, TON, and the systems/learning/measurement neighbors (NSDI/OSDI, IMC, NeurIPS/ICML), writing delta-first contrast rather than a citation catalog, keeping self-citations double-anonymous, and handling concurrent and prior-version overlap.4---56# SIGMETRICS Related Work78Use this to audit novelty and eligibility. SIGMETRICS reviewers are close to the9performance-evaluation literature and expect to see where your paper sits relative to the nearest10prior model, bound, or measurement — stated as a **delta**, not a list. Reopen the current call for11the simultaneous-submission and prior-publication rules (a paper under one-shot revision counts as12under submission) before advising authors.1314## Positioning checks1516- **Separate the analytic/measurement novelty from the engineering effort.** What is new: a tighter17 bound, a more general model, a policy that provably beats a known one, a measurement of a system18 nobody had characterized, or a learning algorithm with a new guarantee?19- **Cover the performance-evaluation lanes** (see the table), not just the papers nearest your20 method. A bibliography missing the obvious queueing-theory predecessor or the prior measurement of21 the same system reads as unaware.22- **Write delta-first.** Each closely related paper gets one sentence naming what it did and one23 naming what you do differently — a tighter bound, a weaker assumption, a broader policy class, a24 larger/newer measurement — not a summary.25- **Preserve double-anonymity.** Cite your own prior work in the third person and never link26 reviewers to an identity-revealing preprint, system page, or repository (Operational Systems Track27 excepted).28- **Declare overlap** with any prior conference/workshop version or concurrent submission; do not29 re-submit archival work as new.3031## Performance-evaluation literature lanes3233| Lane | Typical venues | What SIGMETRICS reviewers check |34|---|---|---|35| Core performance evaluation | SIGMETRICS/POMACS, Performance Evaluation, QUESTA | Whether the nearest model/bound/measurement is compared or distinguished |36| Systems (when you claim a systems payoff) | NSDI, OSDI, SIGCOMM, ATC | Whether the system you improve/measure is credited and fairly baselined |37| Measurement | IMC, PAM, INFOCOM | Whether prior measurements of the same system/workload are engaged |38| Learning (Learning track) | NeurIPS, ICML, COLT | Whether the learning-theoretic predecessor (regret bounds, algorithms) is cited to its origin |39| Networking/queueing journals | IEEE/ACM TON, QUESTA, Stochastic Models | Whether deeper journal-length analyses of the model are engaged |4041A bibliography that cites only your own subarea tells a reviewer the delta may be smaller than42claimed; one that reaches the neighboring theory, systems, and measurement venues signals command of43the field.4445## Delta-first positioning vignette4647Suppose the paper proves a tail-latency bound for a rank-based scheduler. Its nearest neighbors: a48prior analysis of a single age-based policy (one policy, mean latency), a general scheduling49framework (broad class, but no tail bound), and a measurement study of the target system (data, no50policy analysis). The novelty sentence should name all three contrasts — a **tail** bound where the51single-policy analysis gave only mean, a **provable tail guarantee** where the framework gave none,52and a **policy with analysis** where the measurement gave only characterization.5354## Concurrent and prior-version judgment calls5556```text57[Concurrent arXiv work] cite neutrally, state the technical difference (tighter bound? weaker58 assumption? newer measurement?), avoid unverifiable priority claims;59 keep the citation double-anonymous60[Your workshop version] usually non-archival and citable, but confirm against the current call61 wording and phrase so anonymity survives62[Prior short version] declare the overlap and state what the full paper adds (proofs, validation)63[Paper under one-shot revision] it is under submission to SIGMETRICS -- do not submit it elsewhere64 before withdrawing65```6667## Eligibility red flags6869- Substantial text/result overlap with a published paper by the same authors (self-plagiarism risk).70- A "new" analysis that re-derives a known bound without a tighter result or weaker assumption.71- Citations exclusively to non-performance-evaluation venues, signaling the paper may be a systems72 or learning paper rerouted without reframing.7374## Output format7576```text77[Eligibility] clear / needs declaration / risky78[Lanes covered] <performance-eval / systems / measurement / learning / journals>79[Nearest 3 works] <work -> one-line delta (tighter bound / weaker assumption / broader class / newer data)>80[Archival-overlap risk] <none / declare: what>81[Novelty sentence] <SIGMETRICS-ready contribution contrast against the nearest prior work>82```
brycewang-stanford