Results Section Revision

Overview

Use this skill for late-stage Results revision when the science is mostly stable but the writing architecture is not. It is narrower than scientific-writing and manuscript-optimizer: the job here is to repair subsection titles, bridge paragraphs, paragraph openings, and local argumentative flow.

Use scientific-writing for general prose drafting or rewriting. Use manuscript-optimizer when the claim hierarchy, evidence chain, or figure logic is still unstable. Use this skill when the section is largely right but still reads like stitched figure captions rather than a controlled argument. Run anti-defensive-writing after it when Results carries audit, defensive or self-critical writing.

Quick Checks

Before revising any Results subsection, check:

1. Is the title selling the conclusion rather than the method?
2. Does the first sentence of each paragraph state that paragraph's point?
3. Are openings like we asked, we examined, or to test this overused?
4. Are abstract words masking a more specific noun?
5. Is the relationship to the previous paragraph explicit?
6. Is the paragraph only reporting a result, or advancing the argument?
7. Does the paragraph close by stating what changes in interpretation?
8. Does the subsection end in a way that naturally leads into the next one?
9. Are the first and last sentences of each subsection written for a broad peer, stating the question and what the result means in the reader's terms, while the body carries the technical detail? See the Readers section of paper-workflow's references/section-contracts.md.

Subsection Pattern

Each Results subsection should answer two reader questions in order:

1. Why is this analysis needed now?
2. What does it establish?

If the second question appears before the first, the subsection will feel abrupt.

The first paragraph should bridge from the previous subsection and state why the next analysis is necessary. Later paragraphs should each do one job: sharpen a finding, separate related findings, show a consequence for evaluation or prioritization, or explain what a comparison actually distinguishes.

Paragraph Openings

Prefer openings that either orient the reader or state a concrete result:

• The molecular branch first showed ...
• The same pattern was also evident in ...
• A further question was whether ...
• Cross-representation analyses showed ...

Avoid repeated procedural scaffolds:

• To test this ...
• We next asked ...
• We then examined ...

At the start of a new paragraph, avoid vague pronouns unless the referent is unmistakable. Replace this structure or that mismatch with the explicit noun phrase.

Style Rules

• One paragraph, one job.
• Use explicit contrasts early.
• Surface key quantitative anchors before interpretation.
• Keep the register controlled, direct, and specific.
• End each subsection with a sentence that changes the reader's expectation for the next subsection.

Common Failure Modes

• method-like subsection titles
• first paragraphs that announce a result without a rationale
• several paragraphs in a row with the same procedural opening
• abstract nouns that blur what is actually being compared
• summary sentences that restate the result but not its implication
• provenance qualifiers inside Results ("post hoc", "added after the original evaluation", "locked")
• implementation checks reported as findings ("verified", "confirmed to be orthogonal")
• setup, safeguard, caveat, control and exception chained before the result; state the setup in one sentence, then the result