Key Takeaways — Chapter 14: Research Papers

The summary card. Use it to re-ground before the quiz, or to review weeks later.

The one idea

A research paper is an argument, not a description. It makes a specific claim, aimed at a specific gap in what the field knows, and defends it with exactly the evidence the claim requires — and no more. The test for every sentence is not "is this true and did I do it?" but "does this advance the argument?" A paper is not a reward for your effort; it's a case you make to a skeptic.

🚪 Threshold concept: A paper makes an argument; it does not catalog what you did. Before you cross this, more work done feels like a stronger paper, and you include things because you did them. After, the unit is "what I claim and the minimum evidence that proves it" — and you cut things that are true and cost you weeks because they don't serve the claim.

How publishing works (know the machine you're feeding)

• Journals = archival, slow, multi-round review, the permanent record (dominant in natural sciences, medicine, engineering).
• Conferences = one concentrated review round, hard deadlines, published in proceedings — often the primary venue in CS.
• Peer review = experts read your work adversarially before it's published. Flavors: single-blind (they know you), double-blind (anonymize your manuscript), open.
• Your real audience is the handful of reviewers who decide the paper's fate and the narrow community working on your exact problem — not "science" in the abstract.
• Find the venue before you write. Read its author guidelines and three recent papers; write to its conventions.

Join the conversation: the gap, the question, the hypothesis

• The gap = the specific thing the field doesn't yet know / can't do / got wrong. It's what makes your contribution necessary. "More research is needed" names no gap.
• The literature review's job is to establish the gap — to argue it, not summarize sources (preview of Ch 15).
• Research question = the specific, answerable question your study addresses.
• Hypothesis = a falsifiable prediction — a result must be able to refute it. "It will perform well" is not one; "it reduces p99 latency vs. LRU with no throughput penalty" is.

The hourglass (the shape of the argument)

The most common reviewer complaint — "significance unclear" — is almost always a missing bottom funnel: you reported the work but never widened back out to "so what?"

The abstract — the most-read, highest-leverage paragraph

Write four moves, labeled or not: (1) gap → (2) approach → (3) key result with a number → (4) significance. It's the hourglass in one paragraph.

• The fatal flaw is no result. "The results are promising" / "we present findings" describes that results exist without saying what they were. Always state the finding, with a number.
• The other killers: all background (gap should be 1–2 sentences) and overclaiming (it's the most-scrutinized paragraph — claim only what you can support).
• Write it early as a plan; perfect it last as a summary.

The pipeline and responding to reviewers

• Pipeline: submit → editorial screen (risk of desk reject) → peer review → decision → revise + respond → re-review → publish.
• "Major revision" is a victory, not a defeat — it means "we'll publish this if you address these concerns." Rejection is normal and not a verdict on you.
• Respond point-by-point: gracious, firm, every comment. 1. Address every comment — concede, disagree, or honestly decline, but never ignore. 2. Concede what's right, gladly — state exactly what changed and where. 3. Disagree with evidence, not attitude — "we see it differently because…," never "the reviewer misunderstood."
• Conceding valid points buys credibility for the points you defend.

Predatory journals (the urgency exploit)

Tells: unsolicited flattering invitations, guaranteed/absurdly fast acceptance (= no real review), aggressive vague fees, fake metrics, padded editorial boards, a name mimicking a famous journal. (A fee alone isn't the tell — legit open-access journals charge them; it's the fee plus no real review.) Vet every venue; a predatory publication is worse than none.

Themes this chapter surfaced: #1 writing-is-thinking (framing as argument is how you discover what your result means) · #5 structure-serves-the-reader (the hourglass, the abstract) · #2 audience-is-everything (reviewers as the real audience) · #4 revision-is-where-the-writing-happens (the response letter).

Threshold concept: A paper makes an argument; it does not catalog what you did.

Back to: Chapter 14 · Exercises · Quiz · Further Reading