imc-writing-style
GitHub指导ACM IMC测量论文写作，强调首段呈现核心发现、确保观测点与方法可审计、处理纵向数据与混淆因素、论证局限性而非罗列，并早期撰写伦理部分。
Trigger Scenarios
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill imc-writing-style -g -y
SKILL.md
Frontmatter
{
"name": "imc-writing-style",
"description": "Use when drafting or revising the prose of an ACM IMC measurement paper, covering the first-page measurement finding, making vantage points and methodology auditable, longitudinal framing for a moving Internet, arguing limitations rather than reciting them, writing the Ethics section early, and acmart page-budget discipline."
}
IMC Writing Style
Use this while drafting the body. IMC reviewers are measurement experts who read for what you learned about the real Internet and whether the data support it. The organizing principle is that the paper is a measurement contribution: the finding leads, the vantage points and method are auditable, and the limitations and ethics are visible from the start — not appended.
The first-page arc
Put the measurement finding on the first page, in this shape:
- A question about the real Internet a practitioner or researcher recognizes ("how reachable / how deployed / how secure / how does X behave in the wild?").
- Why current knowledge is inadequate — prior work measured too few vantage points, a single snapshot, a testbed instead of deployment, or a biased input.
- The measurement and its vantage points — what you measured, from where, over what period.
- What the data show — the headline finding, stated as a measured result with uncertainty.
- What it means — for operators, protocol designers, or the community, plus what you release.
Do not lead with a system you built or a tool's speed. If the first page is a systems win, the
paper reads as a re-routed NSDI/SIGCOMM submission (imc-topic-selection).
Make vantage points and method auditable
A measurement claim is only as good as the reader's ability to judge the measurement:
- Name the vantage points: locations, ASes, probe types, and how many. Summarize in the body, detail in the appendix.
- Argue representativeness explicitly: why these vantage points support the claim, and where they do not. Quantify coverage bias rather than hoping no one asks.
- Date everything: measurement windows, cadence, and target-list capture dates. The Internet moves; undated results are unfalsifiable.
- Show the method, not just the tool name: enough that a reader could re-measure and judge soundness.
Longitudinal and confound awareness
- Distinguish a snapshot from a trend. If you claim behavior over time, show the window and its stability; if you have a snapshot, scope the claim to it.
- Name confounds a moving Internet introduces — diurnal patterns, churn, load-balancing, CDN effects, measurement-vantage bias — and instrument to bound them.
Limitations that argue, not recite
A good IMC limitations discussion is an argument about validity, engaged where the results live:
- For each headline result, state the threat to its validity (coverage, ground truth, temporal, measurement bias) and how you bounded it.
- Prefer "we observed X from these vantage points; it may not hold for networks unlike them, which we quantify in §Y" over a generic "our study has limitations."
- Do not quarantine all caveats into one closing paragraph.
Write the Ethics section early
The Ethics section is a required gate, not a closing courtesy (imc-submission). Drafting it
early forces you to design safe active measurement, obtain IRB determinations, and plan disclosure
before collecting data — none of which can be retrofitted. Reference it from the introduction so
the reviewer sees the posture immediately.
Page-budget discipline (acmart)
IMC's acmart budget (IMC 2025: up to 13 pages full / 6 pages short of text+figures, plus an
unlimited appendix) is generous but finite for the body. Recover space by cutting redundancy,
not by exiling decision-critical evidence to the appendix (imc-supplementary). Every figure earns
its space by supporting a claim; a vantage-point map that clarifies coverage is worth more than a
decorative architecture diagram.
Common prose failures at IMC
| Failure | Why it hurts | Fix |
|---|---|---|
| System/tool leads the first page | Reads as a systems re-route | Lead with the measurement finding |
| Vantage points unnamed or unjustified | Reader cannot judge validity | Name them; argue representativeness |
| Undated, single-snapshot claims | Unfalsifiable on a moving Internet | Date everything; scope or extend temporally |
| Limitations recited generically | No validity argument | Tie each threat to a result and bound it |
| Ethics as an afterthought | Fails the gate | Write and reference it early |
Output format
[First-page arc] question -> inadequacy -> measurement/vantage points -> finding -> meaning? present/gaps
[Vantage points] named, dated, representativeness argued? yes/no
[Temporal framing] snapshot vs. trend stated correctly? yes/no
[Limitations] argued per-result and bounded? yes/no
[Ethics] section drafted early and referenced up front? yes/no
[Body edits] <ordered list to strengthen the measurement narrative>
Version History
- 9f86f09 Current 2026-07-19 16:01
