research-paper-craft

research-paper-craft — writing a SOTA scientific paper

A paper is a UI for an idea. Almost no one reads it linearly. A reader skims title → abstract → figures → conclusions, then decides whether to read the body. So the paper must work at every altitude — the title alone, the abstract alone, the figures alone — and reward whoever goes deeper. Optimize for the skim and the deep read. (Readers gauge relevance from title+abstract and skim figures before committing — Royal Society 2024; Editage.)

The two-move process (do these in order)

1. Outline first (Whitesides' method). Before any prose, write the outline as the figures and tables in final order, each with a one-sentence claim. The paper IS that figure sequence plus the argument connecting them. If the outline doesn't tell a story, prose won't save it. (Whitesides, Adv. Mater. 2004.)
2. Write backwards. Decide the take-home messages (the 1–3 novel claims) → write Results around the figures → Methods → sharpen Discussion → Introduction (the funnel down to your gap) → Abstract → Title last. You cannot write a tight intro or abstract until you know exactly what you are claiming. (the "writing backwards" strategy, Mar. Life Sci. Technol. 2021 / PMC10077155.)

The narrative spine

Every strong paper is one story: broad context → the specific GAP (the tension) → your question → your answer → the so-what. The gap is the hook — name what is missing, wrong, or unknown, and make the reader feel it before you resolve it. Beginning–middle–end; problem-posed-then-solved. Narrative-style papers measurably engage readers more (PMC10491847). Write the one-sentence version first: "We show that X, which matters because Y." If you can't, you're not ready to write.

Per-section playbook

The engagement / keeping-interesting layer

The reviewer / editor lens

Modern SOTA (2024-era)

Workflow checklist

Sources