paper-writing

# Paper writing Stage 07 turns approved stage summaries into a manuscript. The failure mode is not bad grammar — it is a draft that reports what was done instead of arguing one claim. This skill is the correction for that. `reference.md` in this directory is the long-form treatment: reviewer reading order, the seven sentence-level principles from Gopen and Swan, mathematical notation habits, figure design, and a pre-submission checklist. Read it when a specific problem below is the one you have. ## Start here: the one-sentence contribution Before drafting anything, write the paper's claim as one sentence: - "We prove that X converges under assumption Y." - "We show that method A improves B by 15% on benchmark C." - "We identify failure mode D and propose mechanism E that removes it." If you cannot write it, the framing has not converged, and no amount of prose will fix that. Go back to the Stage 02 hypothesis manifest and the Stage 06 analysis and find out which claim the evidence actually supports. Writing around a claim the results do not support is the most expensive mistake available at this stage. Every section then serves that one claim. Related work, experiments and discussion support it; they are not independent mini-papers. ## The abstract, in five sentences 1. What is the general problem area, in one line a non-specialist follows. 2. What is the specific gap or failure this paper addresses. 3. What did you do — the method, named and characterised. 4. What d...