conext-related-work

CoNEXT Related Work

Position the paper for a networking reviewer who knows the field's flagships cold. At CoNEXT, related

work is not a literature dump; it is the delta argument — what your contribution adds over the

closest prior systems and measurements — written so a double-anonymous reviewer can locate your

paper in the networking landscape in one read.

Cover the lanes a networking reviewer expects

A CoNEXT reviewer will mentally check whether you engage the right bodies of work:

• The flagship conferences: SIGCOMM, NSDI, and CoNEXT itself for systems and architecture.
• Measurement: IMC (and PAM) for internet/system measurement methodology and findings.
• Performance/modeling: SIGMETRICS for analysis and evaluation methodology.
• Wireless/mobile: MobiCom, MobiSys, and related venues when relevant.
• The journal and letters record: IEEE/ACM ToN, CCR (SIGCOMM Computer Communication Review), and

now PACMNET itself.

Missing the lane your paper lives in — e.g., a measurement paper that ignores IMC methodology, or a

data-plane paper that ignores the recent SIGCOMM/NSDI switch-hardware line — reads as unfamiliarity

and invites a change-list item in a one-shot revision.

Delta-first positioning

Write each comparison as a delta sentence, not a summary:

Prior work X [SIGCOMM'YY] measures/builds <what> but <the gap on the real target>;
we <the specific advance> and show it on <the real platform/trace>.

• Lead with the closest prior work, not the oldest or most famous.
• Make the delta concrete and testable: a new mechanism, a measured phenomenon prior work missed, a

platform prior work did not evaluate on, a scale prior work did not reach.

• Avoid "to the best of our knowledge, we are the first" unless it is literally true and checkable;

a measurement reviewer will know a counterexample.

Keep self-citation anonymous

CoNEXT is double-anonymous, so your own prior work is a leak surface:

• Cite your prior papers in the third person ("Prior work [12] proposed..."), never "our

previous system."

• Do not reveal a system lineage that identifies the group (e.g., "the successor to OurSys").
• If a comparison genuinely requires your non-anonymous artifact, use the anonymized location and

describe it without owner-identifying detail.

Positioning against CoNEXT's own recent programs

Because CoNEXT runs two cycles a year and publishes in PACMNET, the venue's own recent output moves

fast:

• Scan the last two CoNEXT programs and recent PACMNET issues for work in your subarea; a

paper that ignores a directly relevant CoNEXT/PACMNET paper from last cycle looks careless.

• If your work extends a CoNEXT paper, state the delta precisely — reviewers may include that

paper's authors.

Structure and length

• Related work can be its own section or woven into the introduction and design sections; CoNEXT

tolerates both, but the delta must be visible early regardless.

• Under the acmart budget, references are unlimited — so cite generously, but keep the prose tight;

a wall of one-line summaries wastes body pages you need for evidence.

Failure modes

| Failure | Why it hurts at CoNEXT | Fix |

|---|---|---|

| Literature list with no delta | Reviewer cannot place your contribution | Rewrite each entry as a delta sentence |

| Missing the closest prior system/measurement | Reads as unfamiliarity; revision risk | Lead with the nearest work and its gap |

| First-person self-citation | Breaks double-anonymity | Third-person, no lineage reveal |

| Ignoring the right lane (IMC/NSDI/SIGMETRICS) | Signals you are a visitor | Engage the venue where your method lives |

| Over-claiming "first" | A reviewer knows a counterexample | Scope the novelty to the checkable delta |

Output format

[Lanes covered] SIGCOMM / NSDI / IMC / SIGMETRICS / MobiCom / ToN-CCR-PACMNET as relevant
[Closest prior work] <papers + the specific gap each leaves on the real target>
[Delta sentences] <one per closest work>
[Anonymity] third-person self-citation; no system-lineage leak
[Recency] last two CoNEXT programs + recent PACMNET issues checked? yes/no