venue_writing_styles.md
80 lines · skills/venue-templates/references/venue_writing_styles.md at 5f2bc03
Writing Across Venues
Reviewed: 2026-10-01. These are drafting choices, not official acceptance criteria or mandatory structures. Follow the exact journal/track instructions in journals_formatting.md and conferences_formatting.md.
Match audience and contribution
| Audience | What to make explicit | Specialist reference |
|---|---|---|
| Multidisciplinary science | The scientific problem and supported implications outside the specialty | nature_science_style.md |
| Biological research | The question, evidence for mechanism, resources, and alternatives | cell_press_style.md |
| Clinical research | Population, design, estimand, uncertainty, harms, and clinical relevance | medical_journal_styles.md |
| ML and vision research | Contribution, assumptions, evaluation conditions, and reproducibility | ml_conference_style.md |
| NLP, HCI, mining/IR | Contribution appropriate to the track, evaluation, participants/data, and implications | cs_conference_style.md |
Venue choice should not change the underlying evidence. Do not upgrade an association to a causal mechanism, remove uncertainty to sound confident, or invent benchmark wins to imitate a conference abstract.
Raw lines 7 to 20Abstract and summary
Resolve the article type before choosing headings or numerical limits. Nature's referenced Article summary differs from Cell's citation-free Summary. Medical journals have different structured headings; a publisher-wide IMRaD formula is insufficient. Cell's eTOC/In Brief is a separate short summary from its Summary, and Highlights have character limits.
For a broad audience, use fewer unexplained terms and explain why the finding matters. For specialists, preserve the technical conditions required to assess the claim. Neither choice requires removing essential methods or quantitative results. Replace stock adjectives with the actual supported contribution.
Raw lines 21 to 33Structure
Choose sections that expose the reasoning and evidence. An empirical paper may use Introduction, Methods, Results, and Discussion; a theory paper may place assumptions, theorem, proof, and implications at its center. A qualitative HCI paper may organize findings by themes with an explicit analytic method.
Contribution bullets, conceptual figures, separate related-work sections, and paragraph counts are optional conventions unless the exact call requires them. Use the space needed for evaluation rather than filling a fixed template quota.
Raw lines 34 to 44Evidence language
- Use “we observed” for the measured result and distinguish the interpretation.
- Use causal language only when the design and stated identification assumptions support it; neither sample size nor journal prestige grants causality.
- Report uncertainty and failed or null findings alongside positive findings.
- Make the scope explicit: population, task, measurement, setting, and time horizon.
- Avoid “similar,” “equivalent,” or “safe” based only on non-significant tests.
Adapting a manuscript
- Check the target's overlap, prior-publication, and simultaneous-submission policy.
- Preserve the research question, data provenance, analyses, and claim strength.
- Restructure for the new audience and page/word budget.
- Keep evidence essential to evaluation in the main text; use supplements as allowed.
- Recheck citation style, reporting checklists, declarations, and anonymization.
- Inspect the rendered result and submission preview.
Changing venue is not permission to recycle a published paper without disclosure, conceal a related submission, or invent additional experiments.
Raw lines 54 to 65Voice and graphics
Prefer direct sentences with a clear subject. First person and passive voice can both be appropriate; there is no publisher-wide requirement that IEEE prose be passive or all medical prose avoid first person. Follow an actual house rule if one applies.
Use graphics to explain evidence or the argument, not as decoration. A conceptual model should show which links are established and which are hypotheses. Figures must remain readable, quantitatively honest, and understandable without relying on color alone.
The bundled examples use placeholders and fictional drafting situations. They are unexecuted writing demonstrations and must not be cited as research evidence. See reviewer_expectations.md for evidence-focused self-review and rebuttals.