← Files Research Writing GuardrailsARCHIVED FILE
skills/compag-elsevier-paper-editing/references/style-audit.md
1.88 KB · Oct 2, 2026 · 00:36 UTC
# COMPAG-style manuscript audit ## Precision Replace vague terms such as "better", "robust", "effective", "consistent", "adaptive", or "efficient" with the measured quantity, operating condition, or mechanism whenever possible. ## Evidence A statement should be traceable to at least one of the following - a definition or equation - implementation details - a figure or table - an ablation - a controlled comparison - a diagnostic measurement When that trace is missing, write the statement as design intent or remove it. ## Information density Delete - repeated motivation already established earlier - repeated definitions - generic transition sentences that add no technical content - restatements of table values without interpretation - unsupported adjectives - defensive qualification that is unnecessary for accuracy ## Logic Prefer forward progression problem → need → mechanism → intermediate effect → supported consequence Use parallel structure when the technical problems are independent. ## Punctuation and emphasis Do not use colon punctuation in ordinary manuscript prose or headings. Do not use italic emphasis. Avoid em dashes. Retain hyphens only where standard English or technical nomenclature requires them, such as nonholonomic, full-horizon when used as an established compound modifier, q-conditioned if that is the frozen term, numerical ranges, and identifiers. ## Tense Maintain one tense policy within each manuscript function. - Persistent method definitions and algorithm behavior are usually stated in the present tense. - Completed experimental procedures may use the past tense. - Results may use past or present tense according to the manuscript's established convention, but remain consistent within a section. - Established facts and figure/table contents are normally stated in the present tense. Do not change tense mechanically when it would alter meaning.
SHA-256: 262e5badddbb078bf2af621edaeb54925490147c9038bfe26f626bdec9e06e9a