Verbatim As Primary Evidence

1. Render Nathan's literal utterance first. Open the reflection with the verbatim quote (or a labelled verbatim block), then analyse. The quote is the anchor; the analysis hangs off it. 2. Preserve typos and characteristic phrasing on first capture. Add a one-line justification if the typo carries the signal ("the instinct is the message, not the spelling"), but do not silently correct. Future pages will link back to the original spelling. 3. Preserve override clauses and conditional phrases as load-bearing. When a quote includes "unless," "when...," "if...," or "or..." clauses that mark the rule's escape channel, render them in full and flag them as load-bearing in the surrounding prose. 4. When the same kind of moment recurs across multiple iterations, preserve the trajectory, not just the latest quote. The trajectory is part of the signal — the user's escalating frustration or their reframing across iterations is data, not noise. 5. Anti-pattern: paraphrasing into fluent reflection prose. The trap is writing "Nathan pushed for legibility across four rounds" — fluent, descriptive, and informative-sounding — when "Round 1: 'four charts. Three we...'" would carry the actual direction of the iteration. The fluent version is easier to read; the verbatim version is what future pages will cite.

The third surface pass over the four in-scope reflections reproduces the prior read exactly: all four reflections lead with a verbatim quote (or block of quotes) and explicitly justify the verbatim discipline inside the reflection itself. The pattern's three load- bearing failure modes — typo carry, override-clause carry, trajectory carry — each have at least one example in this cycle's reflections, and the principle-extraction pair would not be citeable without them. The page's coverage is the most stable of any in the ≥3-reflection set: the reflections are not just cited, they are evidence-of-the-discipline. No new page warranted.

— the trajectory-as-evidence instance. The reflection renders the user's verbatim words across twelve loops as the primary artefact ("the verbatim trajectory of their own words across one session"). It would have been easier — and lossy — to write "the user kept reporting failure across many iterations." The verbatim trajectory surfaces the meta-state (frustration, calibration, suspicion) that a single sentence would have flattened. The reflection could not have made its case about the stacking-bugs arc without showing the exact phrases.