User contributions

Append-only log of contributions by the human author (Florian Krebs) to F(AI)²R. Each entry is mirrored in doc/provenance.ttl as a fair2r:HumanContribution entity per the Contribution extension to the schema; this file is the human-readable shadow.

Why this file exists

The verification ladder grades claims; the logbook records sessions; the provenance graph records activities. None of those, on its own, makes the human author's contribution legible at inspection time. Reviewers, panel discussants, and (especially) funding bodies asking "what did the human actually do here?" deserve an answer that is structured and self-updating rather than a heroic narrative.

This log answers that question with a typed contribution per intervention. The AI's contributions are mirrored at doc/ai-contributions.md (forthcoming) so that the partition the Author's Note describes (paper/sections/authors-note.tex, What surprised me about the cooperation) is empirically auditable rather than asserted.

Distribution by type

Contribution types

Each entry carries one of:

Schema

## YYYY-MM-DD --- short title
*Type:* <one of the above>
*Leverage:* high | medium | low
*Triggered:* <description of subsequent work, ideally with file paths>
*Artefacts touched:* <list>
*Provenance IRI:* fair2r:hc-<slug>

The Provenance IRI matches the corresponding fair2r:HumanContribution entity in doc/provenance.ttl.

Back-fill (reconstructed from git log, the conversation transcript, and doc/logbook.md)

The entries below were reconstructed in one batch on 2026-05-06 from the available evidence. Future sessions append in real time.

2026-05-07 — Rule shape: transcript preservation (first time honoured)

Type: rule-shape Leverage: high Triggered: User asked: "can you keep a full transcript of our chat in a separate document?". Closes a long-standing gap between practice 1 of the eight integrated practices (transcript preservation, prescribed since the bootstrap) and the actual state of the repository (doc/transcripts/ did not exist; the evolution chronology calls this "Pipeline as target, not yet as practice."). Created doc/transcripts/ plus a README and a session file 2026-05-07-session.md. Schema growth: added fair2r:Transcript class to doc/provenance.ttl so transcripts bind as typed entities. The discipline is now active for subsequent sessions; an active session without a transcript file is a defect the FAIR aligner can check for. Artefacts touched: doc/transcripts/README.md (new), doc/transcripts/2026-05-07-session.md (new), doc/provenance.ttl (schema + new entities), doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-transcript-preservation

2026-05-07 — Corrective: poster CI compile fix + README consistency pass

Type: corrective-intervention Leverage: medium Triggered: User flagged "poster link not working" and "update readme as well and keep consistent". (1) Poster CI: slides/poster-A0.tex \includegraphics calls switched to plain filenames + \graphicspath fallback chain; .github/workflows/build-slides.yml gains a Stage figures for the poster step that copies the three needed PNGs into slides/ before latex-action runs; .gitignore extended for the staged copies. (2) README updated to mention the poster as the fifth primary artefact, the mobile-friendly site, the Get-started page, the transcripts directory, the schema-growth class list, and the no-parentless-claim invariant; layout tree brought forward. Artefacts touched: README.md, .gitignore, slides/poster-A0.tex, .github/workflows/build-slides.yml, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-readme-consistency-pass

2026-05-07 — Corrective: mobile-friendly Pages site

Type: corrective-intervention Leverage: medium Triggered: User direction "make the pages mobile friendly". CSS-only fix in site/static/style.css: responsive defaults (img, video, iframe max-width: 100%; pre overflow-x: auto); phone breakpoint at 720 px (tighter padding, header stacks, nav wraps, hero H1 26 px, body headings stepped down, CTAs stack full-width, tables become horizontally scrollable); narrow-phone breakpoint at 380 px. Existing 900 px and 720 px breakpoints preserved. No template change, no JS. Artefacts touched: site/static/style.css, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-make-pages-mobile-friendly

2026-05-08 — Rule shape: portrait poster density fix + scrutinizer rule

Type: rule-shape Leverage: high Triggered: User screenshot showed the portrait poster filled only ~50 % of the page; user said "also update the scrutinizer to account for that". Two-step contribution: (a) fix the poster (font 25pt → 30pt, figures to full linewidth, +2 blocks in centre column; rendered fill verified at 98 %); (b) promote the catch into permanent rules in agents/layout-scrutinizer.md under You do (conference poster): vertical fill ≥ 80 %, column balance ≤ 15 %. The rule-shape leverage is high because the same density-defect class will recur whenever content sized for one orientation lands in another; the scrutinizer now catches it without rediscovery. Artefacts touched: slides/poster-A0.tex, site/static/figures/poster-A0.png, .github/workflows/build-slides.yml, agents/layout-scrutinizer.md, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-fix-poster-density-and-scrutinizer-rule

2026-05-08 — Corrective: provenance graph hygiene pass

Type: corrective-intervention Leverage: medium Triggered: User direction "validate provenance i think there are some unconnected entites". 9-query audit found: 1 orphan activity, 4 unused agent IRIs, 1 unused transcript, 2 untyped IRIs in our namespace, 28 ghost sources. Fixes: declared verif:lit-retrieved + verif:lit-read (overlooked rungs); typed ent:section-implementation (retired body role) + ent:section-discussion (removed-before-curation tombstone); connected act:invalidate-discussion-section via prov:wasInformedBy + prov:invalidated; bound 4 agent IRIs to their activities; bound the session transcript to its parent activity. The 28 ghost sources are documented expected behaviour; the 13 unused observations are hypothesis-stage by design. Artefacts touched: doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-provenance-graph-hygiene-pass

2026-05-08 — Rule shape: scrutinizer scope expanded; poster flipped to A0 portrait

Type: rule-shape Leverage: medium Triggered: User direction "layout scrutinizer should scrutinize slides and poster as well" + "poster should be portrait format not landscape". agents/layout-scrutinizer.md rewritten with explicit per-artefact scrutiny rules for paper / slide decks / poster / Pages site. Poster \documentclass flipped from landscape to portrait; PNG re-rendered; landscape→portrait references propagated to CLAUDE.md, README.md, and doc/provenance.ttl. Artefacts touched: agents/layout-scrutinizer.md, slides/poster-A0.tex, site/static/figures/poster-A0.png, CLAUDE.md, README.md, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-expand-layout-scrutinizer-and-flip-poster

2026-05-07 — Structural: conference poster as fifth primary artefact

Type: structural-decision Leverage: high Triggered: User direction "add another artifact to the build chain a conference poster/Infographic. consider distilling the process to its essentials. how would that change page count for paper", followed by "go" against a two-step plan. Step (a) landed: poster scaffolded (slides/poster-A0.tex, tikzposter A0 landscape, DLR-CD overrides on the Simple theme); Makefile poster target; CI workflow extended; rolling release publishes poster-A0.pdf alongside the decks; schema growth (fair2r:Poster class); CLAUDE.md primary-artefact list 4 → 5; site CTA link added. The discipline of writing the poster is the page-count forcing function for step (b). Artefacts touched: slides/poster-A0.tex (new), slides/Makefile, slides/poster-A0.png (render), site/static/figures/poster-A0.png, site/index.md, .github/workflows/build-slides.yml, CLAUDE.md, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-scaffold-conference-poster

2026-05-07 — Corrective: hero PNG embedded on Pages Home

Type: corrective-intervention Leverage: low Triggered: User question "do we have an eyecatcher graphic is it also in the pages?". PDF carried hero on front matter; Pages site did not. Rendered paper/figures/hero.tex via the existing standalone wrapper to a 200 dpi PNG (after adding \usepackage{amsmath} to the wrapper); copied to site/static/figures/hero.png; embedded near the top of site/index.md between the lede and the CTA row. Artefacts touched: paper/figures/hero.standalone.tex, paper/figures/hero.png, site/static/figures/hero.png, site/index.md, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-eyecatcher-on-pages

2026-05-07 — Corrective: scrutinizer-pass fix-up (post PR #43)

Type: corrective-intervention Leverage: medium Triggered: User direction "run a scrutinizer pass". Three audit subagents (layout-scrutinizer, readability-reviewer, fair-aligner) returned punch lists against the focused-narrative trim state. Small / safe items applied here: 8 newly orphaned bib entries re-cited in prose (PRISMA / GRADE in pattern.tex Crosswalk; PROV-O in pattern.tex axes; Datasheets + Model Cards in eight-practices item 3; ACL ARR + ICLR + NeurIPS LLM policies in intro.tex venue-policy churn). Three orphan figure labels cross-referenced from prose. Conclusion forward-ref to sec:research-infrastructure repaired to point at appendix-g. Xennial paragraph split. Redundant "None of the eight is novel" in conclusion + closing roadmap in Author's Note removed. Trim-artefact paragraph in pattern.tex folded into "Extensions and cousins". Long-term archivability paragraph in limits-and-objections.tex dropped (re-trod earlier paragraph). paper/sections/self-reference.tex and paper/figures/eight-practices.tex deleted (orphan files retired in earlier passes). Bib Missing $ warning + slug-match heuristic in scripts/provenance_analysis.py deferred. Artefacts touched: paper/sections/conclusion.tex, paper/sections/authors-note.tex, paper/sections/intro.tex, paper/sections/pattern.tex, paper/sections/eight-practices.tex, paper/sections/limits-and-objections.tex, paper/sections/self-reference.tex (deleted), paper/figures/eight-practices.tex (deleted), doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-scrutinizer-pass-fix-up

2026-05-07 — Structural: focused-narrative trim pass + update pass

Type: structural-decision Leverage: high Triggered: Researcher direction "generate the focused version. afterwards complete update pass." against the seven-item triage filtered by narrative load. Five-beat spine (gap → position → eight integrated practices + pipeline → recursive case study + honest-about-limits → why now) was the filter. Cuts: background.tex and related.tex removed entirely (substrate already inline-cited; Cousins. paragraph folded into pattern.tex); sustainability.tex + objections.tex merged into limits-and-objections.tex (~683 w from a combined ~1170 w); pattern.tex §domain-ontologies + §research-infrastructure moved to appendix-g-pattern-extensions.tex; authors-note.tex rewritten ~1010 → 561 w; evolution.tex three insight paragraphs collapsed into one; provenance-analysis.tex trimmed 436 → 276 w. Net body 10466 → 6496 words (~38%); PDF 38 → 33 pages. Artefacts touched: paper/main.tex, paper/sections/authors-note.tex (rewritten), paper/sections/intro.tex (roadmap), paper/sections/position.tex (cross-refs), paper/sections/pattern.tex, paper/sections/eight-practices.tex, paper/sections/provenance-analysis.tex, paper/sections/evolution.tex, paper/sections/limits-and-objections.tex (new), paper/sections/appendix-g-pattern-extensions.tex (new), paper/sections/{background,related,sustainability,objections}.tex (removed), doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-trim-pass-focused-narrative

2026-05-07 — Structural: trim pass 2 — A–D from further-candidates list

Type: structural-decision Leverage: high Triggered: Researcher direction "do a to d" against the further-candidates list. (A) provenance-analysis.tex mechanics → new appendix-e-provenance-analysis.tex (body keeps the two figures + a summary paragraph + the "what the analysis is and is not" framing). (B) implementation.tex → appendix-f-implementation.tex (label retitled sec:appendix-implementation; cross-refs in intro.tex and position.tex updated). (C) statement-of-authorship.tex converted to \section* with \addcontentsline (back-matter). (D) self-reference.tex cut from body \input (preserved on disk; load-bearing claim survives in Author's Note + practice 5). Net body reclaim ~2.0 pp. Two prov:Invalidation records; two new appendix sections. Artefacts touched: paper/sections/provenance-analysis.tex, paper/sections/appendix-e-provenance-analysis.tex (new), paper/sections/appendix-f-implementation.tex (renamed from implementation.tex), paper/sections/statement-of-authorship.tex, paper/sections/eight-practices.tex, paper/sections/evolution.tex, paper/sections/appendix-d-evolution-chronology.tex, paper/sections/intro.tex, paper/sections/position.tex, paper/main.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-trim-pass-2-appendix-routing

2026-05-07 — Structural: trim pass — evolution chronology -> appendix; pattern §handback de-duplicated

Type: structural-decision Leverage: high Triggered: User picked items 1 and 2 from the seven-item complete-loop trim list ("2 should be an appendix, fix 1"): (1) pattern.tex §handback trimmed from a three-paragraph mirror-discipline restatement to a one-paragraph handback statement plus a forward reference to practice 4 of \S\ref{sec:eight}; (2) evolution.tex's 14 \paragraph chronology blocks moved to a new paper/sections/appendix-d-evolution-chronology.tex. Body keeps the digest table plus three load-bearing summary paragraphs (Researcher-as-corrector, Pipeline as target not yet as practice, Logbook as conscience). Net body reclaim ~1.6 pp toward the 10-pp budget. Artefacts touched: paper/sections/evolution.tex, paper/sections/appendix-d-evolution-chronology.tex (new), paper/sections/pattern.tex, paper/main.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-trim-evolution-and-pattern

2026-05-07 — Content prompt: getting-started guide + auto-sync from CLAUDE.md

Type: content-prompt Leverage: high Triggered: User asked: "to make process usable to other users can you write a getting started guide (especially for the page)? basically I want ppl to run a specific initialization prompt to set the process up. (keep this instructions always synced to process)". New page site/getting-started.md hands an adopter a single initialization prompt that points the LLM at this repository as the canonical reference; the Ground rules block on the page is auto-synced from CLAUDE.md by scripts/sync_getting_started.py, which the site builder invokes on every build. Wired into PAGES + NAV. Site now publishes 27 pages. Artefacts touched: site/getting-started.md (new), scripts/sync_getting_started.py (new), scripts/build_provenance_site.py, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-getting-started-page-and-sync

2026-05-07 — Observation: associative thinking as the cooperation multiplier

Type: observation Leverage: high Triggered: End-of-session prompt from the researcher: "how far does associative thinking capability help in leveraging llm cooperation benefits. I mean in the convo you often provide suggestions which lead me to other angles or ideas". Logged at hypothesis stage in doc/user-observations-log.md; not graduated to paper text yet. Reframes the model's value as producing near-miss candidates the researcher converts to usable angles rather than being right. Schema-growth hint flagged for the next case study (fair2r:cooperativeOriginType). Artefacts touched: doc/user-observations-log.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-obs-associative-thinking-as-multiplier

2026-05-07 — Corrective: complete-loop fix pass (cross-refs, voice, missing graph entity, palette, slides)

Type: corrective-intervention Leverage: medium Triggered: The complete-loop audit ensemble (five subagents) returned a punch list; this entry captures the small, safe fixes applied directly from the findings. Three broken cross-references repaired (sec:discussion, sec:fm-residual, sec:practice-disclosure). One voice-drift instance corrected (source-analyser → Source Analyzer). Missing src:ravi2024fair4ml graph entity added (FAIR4ML cited five times in prose but had no fair2r:Source triple). Palette comment in paper/figures/src/provenance-topology.py:27 realigned with paper/style/fair2r.sty (authoritative source). AI² grid finally wired into both slide decks (PR #33's claim that it "lands on the position frame" was unmet; now true). Page-budget cuts and Xennial-paragraph split deferred — those are taste calls. Artefacts touched: paper/sections/intro.tex, paper/sections/objections.tex, paper/sections/appendix-a-ladder.tex, paper/sections/failure-modes.tex, paper/figures/src/provenance-topology.py, slides/pitch-5min.tex, slides/conference-30min.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-complete-loop-fix-pass

2026-05-07 — Content prompt: summary table of process evolutions

Type: content-prompt Leverage: high Triggered: User asked for "a summary maybe a table what evolutions (rules, agents, ...) during the process of writing the paper occurred". Graduated to paper text as a 20-row longtable at the top of paper/sections/evolution.tex (Table~1) and mirrored as a Markdown table on doc/methodology.md (and therefore on the Pages site). Five evolution classes documented: rule, agent, schema, pipeline, manuscript. tabularx + longtable added to paper/style/fair2r.sty. New claim:evolution-summary-table born with parent activity per the no-parentless-claim rule. Artefacts touched: paper/sections/evolution.tex, paper/style/fair2r.sty, doc/methodology.md, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-evolution-summary-table

2026-05-07 — New claim: senior-researcher Xennial-like bridging responsibility

Type: content-prompt Leverage: high Triggered: User proposed: "we as senior researchers who know the way how things were done and why they were done that have a responsibility to help build bridges to the unthinkable of the past (a similar situation as xennials)". Graduated to a new paragraph The bridge to what is now unthinkable. in the Conclusion (between The longer arc. and the closing emphasis line). New bib entry shafer2014xennials (lit-retrieved). New claim:senior-researcher-bridge born with an explicit parent activity per the no-parentless-claim rule from the 2026-05-07 curator pass. Artefacts touched: paper/sections/conclusion.tex, paper/references.bib, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-senior-researcher-bridge-claim

2026-05-07 — Rule shape: expand illustration toolset; pin DLR-CD constraint

Type: rule-shape Leverage: medium Triggered: User flagged that "Mermaid, matplotlib, and other publication-ready tools can also be used by the illustrator. if unsure ask" and, on a follow-up, "but keep dlr design system in mind". agents/illustration.md updated: the "TikZ-only" implicit rule is retired; the prompt now records the full toolset (matplotlib, seaborn, plotly, scienceplots, cmocean, colorcet, proplot, TikZ, PGFPlots, svgwrite, drawsvg, schemdraw, matplotlib-scalebar, diagrams, graphviz, networkx, mermaid-py, gridspec, plotnine, patchworklib, bokeh, ipywidgets, plus molecular / image-processing libraries for sub-domain instantiations). The DLR Corporate Design (Arial / Helvetica fallback, dlrBlau1 #00658B accent, mid-grey neutrals, square corners only, hairline rules at 0.4pt, no shadows / glows / gradients / 3D, no emoji, no clip-art) is now a binding section above the toolset list and explicitly applies to every tool — default tool theming counts as an unfinished figure. Artefacts touched: agents/illustration.md, doc/logbook.md, doc/user-contributions.md, doc/provenance.ttl. Provenance IRI: fair2r:hc-illustration-toolset-and-dlr-cd

2026-05-07 — Verification pass: 16 sources advanced + 3 bib corrections applied

Type: rule-shape Leverage: high Triggered: User requested "verification pass". A source-analyzer subagent walked the reading queue, advancing 16 fair2r:Source entities from lit-retrieved to ai-confirmed with verbatim quoted snippets in doc/sources.md. Three paywalled-but-load- bearing sources (vannoorden2023chatgpt, neurips_llm_policy, iclr_llm_policy) were escalated into the new doc/sources-needing-institutional-access.md for the human author to retrieve via DLR institutional access. Four bib corrections flagged; three applied in this commit (eisen year 2018→2020 and "Peer Review:" title prefix; kobak title revision; curdt full author list); the fourth (vannoorden DOI) was already fixed in PR #11. Artefacts touched: doc/sources.md, doc/provenance.ttl, doc/sources-needing-institutional-access.md (new), paper/references.bib, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-verification-pass-and-bib-corrections

2026-05-06 — Bootstrap request: initialise repo for F(AI)²R

Type: content-prompt Leverage: high Triggered: Initial 37-file commit; the entire scaffolding (paper/, doc/, agents/, root metadata). Artefacts touched: whole tree. Provenance IRI: fair2r:hc-bootstrap

2026-05-06 — Primary-artefact consistency invariant

Type: rule-shape Leverage: high Triggered: Primary-artefact consistency (binding) block added to every prompt under agents/; orchestrator named ultimate custodian; provenance-curator named keystone; FAIR aligner surfaces desync as audit fail. Artefacts touched: CLAUDE.md, doc/methodology.md, agents/*.md, doc/provenance.ttl, doc/logbook.md. Provenance IRI: fair2r:hc-consistency-invariant

2026-05-06 — Site + paper-build CI

Type: content-prompt Leverage: high Triggered: scripts/build_provenance_site.py, .github/workflows/pages.yml, .github/workflows/build-paper.yml. Artefacts touched: scripts/, .github/workflows/, site/. Provenance IRI: fair2r:hc-ci-site

2026-05-06 — Chapter-per-file rule for .tex

Type: rule-shape Leverage: medium Triggered: File granularity (binding) block in agents/scientific-writer.md; paper/main.tex refactored as a thin assembler; one \section{} per file under paper/sections/. Artefacts touched: agents/scientific-writer.md, paper/main.tex, paper/sections/. Provenance IRI: fair2r:hc-chapter-rule

2026-05-06 — "This paper, generated by this process" section

Type: content-prompt Leverage: medium Triggered: paper/sections/self-reference.tex. Artefacts touched: paper/sections/self-reference.tex, paper/main.tex. Provenance IRI: fair2r:hc-self-reference-section

2026-05-06 — "Process evolution during cooperative writing"

Type: content-prompt Leverage: high Triggered: paper/sections/evolution.tex (the recursive case study of the cooperation that produced the paper). Artefacts touched: paper/sections/evolution.tex. Provenance IRI: fair2r:hc-evolution-section

2026-05-06 — Short-form pivot (~10 pages of body)

Type: structural-decision Leverage: high Triggered: paper/main-condensed.tex retired; every section file compactly rewritten; agents/condenser.md repurposed as page-budget enforcer; paper/Makefile gains pages target with PAGE_BUDGET=10. Artefacts touched: paper/, agents/condenser.md, agents/scientific-writer.md. Provenance IRI: fair2r:hc-short-form-pivot

2026-05-06 — DLR Corporate Design integration

Type: structural-decision Leverage: high Triggered: site/static/dlr/ (vendored tokens, fonts, logos); paper/style/fair2r.sty; site CSS rewritten on the DLR token palette; imprint added to every metadata surface. Artefacts touched: site/static/dlr/, paper/style/fair2r.sty, site/static/style.css, CITATION.cff, codemeta.json, .zenodo.json, README.md, paper/main.tex. Provenance IRI: fair2r:hc-dlr-design

2026-05-06 — "Add a section on provenance analysis"

Type: content-prompt Leverage: high Triggered: paper/sections/provenance-analysis.tex, scripts/provenance_analysis.py, four auto-generated LaTeX fragments (rung distribution, section coverage, source strength, recent activities). Artefacts touched: paper/sections/provenance-analysis.tex, scripts/provenance_analysis.py, paper/Makefile, .github/workflows/build-paper.yml. Provenance IRI: fair2r:hc-provenance-analysis-section

2026-05-06 — Generate illustrations + interactive graph viewer

Type: content-prompt Leverage: medium Triggered: paper/figures/pipeline.tex (TikZ ten-stage pipeline figure); site/provenance-explorer.md (vis-network interactive node- link viewer); JSON exporter in scripts/build_provenance_site.py. Artefacts touched: paper/figures/, site/provenance-explorer.md, scripts/build_provenance_site.py. Provenance IRI: fair2r:hc-illustrations

2026-05-06 — Splashy homepage; PDF link prominent

Type: content-prompt Leverage: medium Triggered: site/index.md rewritten with a hero CTA-row, callout, three-card grid; the GitHub Pages URL promoted to a top-of-README section in two passes ("the URL is missing" → promote to dedicated section; "I mean the pages" → confirm Pages URL not repo URL). Artefacts touched: site/index.md, README.md, site/static/style.css. Provenance IRI: fair2r:hc-splashy-homepage

2026-05-06 — Fix CI; verify build runs

Type: corrective-intervention Leverage: high Triggered: paper/style/fair2r.sty rewritten (five bugs fixed: slash-prefixed \ProvidesPackage, broken xkeyval machinery, missing colortbl, fragile sansmath, risky quote redefinition); .github/workflows/pages.yml Pages-specific steps gated on main. Artefacts touched: paper/style/fair2r.sty, .github/workflows/pages.yml, .github/workflows/build-paper.yml. Provenance IRI: fair2r:hc-ci-fix

2026-05-06 — Vendored DRAFT-watermarked PDFs to reduce CI runs

Type: rule-shape Leverage: medium Triggered: \faiarDraft=true default in paper/main.tex; rolling latest-draft GitHub Release published by the build-paper workflow on pushes to main via softprops/action-gh-release. Artefacts touched: paper/main.tex, paper/style/fair2r.sty, .github/workflows/build-paper.yml, README.md. Provenance IRI: fair2r:hc-draft-pdf-release

2026-05-06 — Source-research mandate; Consensus / Scholar default

Type: rule-shape Leverage: high Triggered: Source-analyser subagent dispatched; doc/sources.md populated with topic-organised, ladder-graded entries; agents/source-analyzer.md and doc/research-protocol.md updated to mandate Consensus / Scholar as the default scientific-search tool. Artefacts touched: agents/source-analyzer.md, doc/research-protocol.md, doc/sources.md, paper/references.bib. Provenance IRI: fair2r:hc-source-workflow

2026-05-06 — Add scientific repositories + paywall escalation

Type: rule-shape Leverage: medium Triggered: arXiv, OpenAlex, Crossref, PubMed/EuropePMC, IEEE Xplore, Springer Link, ACM Digital Library, ScienceDirect added to the source workflow; institutional-access escalation channel introduced (a paper that cannot be read on the open web is requested via a structured log that the human author can fulfil through DLR institutional access). Artefacts touched: agents/source-analyzer.md. Provenance IRI: fair2r:hc-scientific-repos

2026-05-06 — Cache-bust the deployed Pages site

Type: corrective-intervention Leverage: low Triggered: SHA-1-keyed ?v= versioning on style.css and provenance.json; soft Cache-Control meta; build-version footer. Artefacts touched: scripts/build_provenance_site.py, site/provenance-explorer.md. Provenance IRI: fair2r:hc-cache-bust

2026-05-06 — Mermaid renderer fix for provenance-graph.md

Type: corrective-intervention Leverage: low Triggered: Three GitHub-Mermaid syntactic issues repaired (subgraph title with parentheses; node labels with colons; edge labels with colons); companion HTML-escaping fix in the site builder. Artefacts touched: doc/provenance-graph.md, scripts/build_provenance_site.py. Provenance IRI: fair2r:hc-mermaid-fix

2026-05-06 — Author's Note carried over from Obscurity-Is-Dead

Type: structural-decision Leverage: high Triggered: paper/sections/authors-note.tex adapted from the source paper's front-matter "Author's Note — Advice for reading"; positions F(AI)²R as a position paper; folds in the "What surprised me about the cooperation" finding. Artefacts touched: paper/sections/authors-note.tex, paper/main.tex. Provenance IRI: fair2r:hc-authors-note

2026-05-06 — Position-paper reframing

Type: structural-decision Leverage: high Triggered: paper/sections/position.tex (five numbered claims); paper/sections/objections.tex (six anticipated objections with charitable responses); paper/sections/statement-of-authorship.tex; paper/sections/ discussion.tex retired; paper/sections/conclusion.tex sharpened. Artefacts touched: paper/sections/, paper/main.tex. Provenance IRI: fair2r:hc-position-reframe

2026-05-06 — "What's my contribution?" — the question that became this file

Type: responsibility-uptake Leverage: high Triggered: The model's perspective in chat became the seed for the What surprised me about the cooperation paragraph in the Author's Note; the user's follow-up "yes add it" promoted the question and the answer into the published paper text; this contributions log and the companion observations log were created so the partition is empirically auditable rather than asserted. Artefacts touched: paper/sections/authors-note.tex, doc/user-contributions.md, doc/user-observations-log.md, doc/provenance.ttl (Contribution class). Provenance IRI: fair2r:hc-contribution-question

2026-05-06 — Contribution-tracking rule (this file's existence)

Type: rule-shape Leverage: high Triggered: This file; companion observations log; fair2r:Contribution / HumanContribution / AIContribution / MetaContribution extension to the PROV-O schema; binding clause in CLAUDE.md and in agents/orchestrator.md requiring that every material contribution be logged here as a side-effect of the primary-artefact consistency invariant. Artefacts touched: doc/user-contributions.md, doc/user-observations-log.md, doc/provenance.ttl, CLAUDE.md, agents/orchestrator.md. Provenance IRI: fair2r:hc-contribution-rule

2026-05-07 — Claim: F(AI)²R relates to model checking / automatic theorem proving

Type: content-prompt Leverage: high Triggered: New paragraph in §11 (Related Work) titled "Formal methods: model checking and theorem proving." The paragraph maps F(AI)²R machinery onto formal-methods vocabulary (conjecture / proof-state / proof-checker / invariants / prover-checker separation), names two future-work imports the analogy suggests (counter-example witnesses on retracted claims; lemma libraries citable by IRI), and hedges with "the analogy is partial". Two new bib entries: clarke2009modelchecking (CACM 2009 Turing lecture) and klein2009sel4 (SOSP 2009 verified microkernel). Artefacts touched: paper/sections/related.tex, paper/references.bib, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-formal-methods-relationship

2026-05-07 — Fix Beamer conference deck: enumitem + Beamer enumerate-template conflict

Type: corrective-intervention Leverage: high Triggered: User pasted the conference-30min.log tail from the failing build-slides run. Real error: TeX capacity exceeded, sorry [grouping levels=255]. \labelenumi -> { \labelenumi } — a recursive macro definition caused by loading enumitem alongside Beamer's \setbeamertemplate{enumerate item} in slides/style/fair2r-beamer.sty. Removed \usepackage{enumitem} from slides/conference-30min.tex and stripped [leftmargin=...,itemsep=...] from all three \begin{enumerate} calls. The pitch deck never used enumerate, which is why it compiled all along. Artefacts touched: slides/conference-30min.tex, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-fix-beamer-enumitem-conflict

2026-05-07 — Reconcile provenance + new eight-practices figure + scrutiny pass

Type: rule-shape Leverage: medium Triggered: (a) Deleted paper/sections/discussion.tex and paper/sections/acknowledgements.tex (both marked superseded after the position-paper reframe; unreferenced); the provenance.ttl activity act:rev-reconcile-and-illustrate records a prov:Invalidation of ent:section-discussion. (b) New paper/figures/eight-practices.tex (4x2 grid colour-coded by failure-mode family) wired into paper/sections/eight-practices.tex. (c) Scrutiny of all section files for voice issues; the Author's Note's direct address is preserved as a documented genre exception (claim:authors-note-voice-exception); agents/scientific-writer.md updated to make the carve-out explicit; self-reference.tex's closing aphorism reworded to remove a second-person "you". Artefacts touched: paper/sections/eight-practices.tex, paper/sections/self-reference.tex, paper/figures/eight-practices.tex (new), paper/sections/discussion.tex (deleted), paper/sections/acknowledgements.tex (deleted), agents/scientific-writer.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-reconcile-illustrate-scrutinize

2026-05-07 — Claim: traditional publishing under stress; journal-as-distribution in decline

Type: content-prompt Leverage: high Triggered: New third paragraph in §1 of paper/sections/intro.tex titled "The journal as distribution channel may be in decline.". The paragraph escalates the §1 framing from "LLMs cause failure modes per paper" to "the journal-as-distribution format itself is under structural strain"; cites preprint / open-access / overlay- journal alternatives plus a Nature piece on AI-fabricated submissions; hedges with "we do not claim the journal is dead". Three new bib entries: eisen2018preprints, tennant2016open, conroy2023sleuths (all lit-retrieved, TODO-VERIFY at lit-read). Artefacts touched: paper/sections/intro.tex, paper/references.bib, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-journal-decline

2026-05-07 — Reconcile: main is default; Pages source = GitHub Actions

Type: rule-shape Leverage: high Triggered: User switched the GitHub default branch to main and set the Pages source to "GitHub Actions". Docs reconciled: CLAUDE.md branch-policy rewritten + new "Pages source" subsection; agents/orchestrator.md refusal-conditions updated; doc/methodology.md gains a "Branch and deployment policy" subsection. Artefacts touched: CLAUDE.md, agents/orchestrator.md, doc/methodology.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-reconcile-default-branch-and-pages-source

2026-05-07 — Verify CI; fix the conference-deck Beamer compile failure

Type: corrective-intervention Leverage: high Triggered: Inspection of PR #19's check_runs surfaced two build-slides.yml failures at the "Compile conference deck" step (exit code 12); logs auth-protected. Four defensive fixes in slides/style/fair2r-beamer.sty: removed \\renewcommand{\\familydefault}, simplified \\sectiondivider (no group-local background canvas), reduced helvet scaling to 0.95, wrapped frametitle in a minipage. Companion fix in slides/conference-30min.tex: single-line subtitle. Artefacts touched: slides/style/fair2r-beamer.sty, slides/conference-30min.tex, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-fix-beamer-conference-build

2026-05-06 — Pages staleness: defensive fixes + forced deploy

Type: corrective-intervention Leverage: low Triggered: scripts/build_provenance_site.py now writes _site/.nojekyll; the page template's hero section gains a visible build-tag printing the cache-bust hash; touching the script + CSS matches pages.yml's path filter so the merge triggers a fresh deploy and busts the CDN cache. Artefacts touched: scripts/build_provenance_site.py, site/static/style.css, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-pages-staleness-fix

2026-05-06 — Beamer-only: drop Marp, ship Beamer slide decks

Type: structural-decision Leverage: high Triggered: Switched the slide pipeline from Marp to Beamer at researcher request ("only ship beamer"). Removed slides/pitch-5min.md, slides/conference-30min.md, slides/static/dlr/ (vendored Marp theme + assets), and scripts/build_slides.py (the Marp asset-inlining preprocessor). Added slides/style/fair2r-beamer.sty (Beamer companion to paper/style/fair2r.sty); two Beamer decks slides/pitch-5min.tex and slides/conference-30min.tex; rewrote slides/Makefile to drive latexmk; rewrote .github/workflows/build-slides.yml to use xu-cheng/latex-action exactly like build-paper.yml. Updated agents/presentation.md to be Beamer-only (with explicit refusal to reintroduce a Marp / HTML / PPTX path); CLAUDE.md and doc/methodology.md primary-artefact list now names .tex sources; README and site/index.md link the Beamer PDFs only (no PPTX downloads). Artefacts touched: slides/, agents/presentation.md, .github/workflows/build-slides.yml, CLAUDE.md, doc/methodology.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md, README.md, site/index.md. Provenance IRI: fair2r:hc-beamer-only-slides

2026-05-06 — Fix slide rendering (DLR theme) + trigger Pages rebuild

Type: corrective-intervention Leverage: high Triggered: New scripts/build_slides.py ports the marp-dlr framework's scripts/run-marp.mjs asset-inlining preprocessor to Python: the DLR theme's url('./assets/...') references are replaced with base64 data URIs before Marp ever sees them, so the title and section-divider background plates plus the DLR logos actually load in headless Chromium. slides/Makefile and .github/workflows/build-slides.yml now invoke the Python preprocessor instead of calling Marp directly. site/index.md gains a one-line note about the preprocessor, which has the side effect of triggering pages.yml (whose path filter the slide-only commits did not match), so the deploy refreshes and busts the CDN cache that made the site look "back to the old page". Artefacts touched: scripts/build_slides.py, slides/Makefile, .github/workflows/build-slides.yml, site/index.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md, doc/user-observations-log.md. Provenance IRI: fair2r:hc-fix-slide-rendering

2026-05-06 — Two slide decks + presentation agent + slides as fourth primary artefact

Type: structural-decision Leverage: high Triggered: New slides/ tree with the vendored marp-dlr theme under slides/static/dlr/; two source decks (slides/pitch-5min.md for the 5-minute pitch and slides/conference-30min.md for the 25 + 5 conference talk); slides/Makefile (make all, make pitch, make conference, make watch); new .github/workflows/build-slides.yml workflow that renders both decks to PDF + PPTX + HTML on every push and publishes the rendered files to a rolling latest-draft-slides GitHub Release on main; new agents/presentation.md as the slide-deck custodian; slides promoted to fourth primary artefact in CLAUDE.md and doc/methodology.md. README and site/index.md gain prominent download links to both decks (PDF and PPTX). Artefacts touched: slides/, .github/workflows/build-slides.yml, agents/presentation.md, CLAUDE.md, doc/methodology.md, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md, README.md, site/index.md. Provenance IRI: fair2r:hc-slide-decks

2026-05-06 — "yes" — promote the longer-arc paragraph into the Conclusion

Type: structural-decision Leverage: medium Triggered: New "The longer arc" paragraph in paper/sections/conclusion.tex restating claim:future-research-infrastructure with reference back to §3.6 and the Helmholtz cross-centre graph (curdt2025hmc). The position-paper closer now ends on the forward-looking note before the closing aphorism. Artefacts touched: paper/sections/conclusion.tex, doc/provenance.ttl, doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-promote-longer-arc

2026-05-06 — Forward-looking pair: bioinformatics precedent + research-infrastructure vision

Type: content-prompt Leverage: high Triggered: New §3.6 Looking forward: provenance graphs as research infrastructure in paper/sections/pattern.tex. The researcher posed two paired claims: (a) "future research success might depend a lot more on provenance models for hypothesis development, graph interconnectivity (Helmholtz graph supercharged) and clever experimental validation", and (b) "bioinformatics was driven there earlier by the at-the-time obscene amounts of data leading to semantic tagging becoming the norm to identify genes". Both graduated to paper text. The bioinformatics-precedent argument is anchored on Ashburner et al. 2000 (Gene Ontology, Nature Genetics); the Helmholtz cross-centre graph reuses curdt2025hmc. The marketing-verb "supercharged" was rephrased to "fully-instrumented" to fit the DLR voice rules. Two new fair2r:Claim entries (bioinformatics-precedent, future-research-infrastructure), both at human-confirmed. Artefacts touched: paper/sections/pattern.tex, paper/references.bib, doc/provenance.ttl, doc/user-contributions.md. Provenance IRI: fair2r:hc-research-infrastructure

2026-05-06 — "ontologies might actually help" — domain ontologies as schema extension

Type: content-prompt Leverage: high Triggered: New §3.5 "Extensibility through domain ontologies" in paper/sections/pattern.tex; three new bib entries (janowicz2019sosa for SOSA/SSN, rijgersberg2013om for OM-2, qudt for QUDT); new claim:domain-ontologies-extension in the graph at human-confirmed. The German fragment "konkrete standardisierte Methode" carried into the prose as a load-bearing technical term in the source-author voice. Artefacts touched: paper/sections/pattern.tex, paper/references.bib, doc/provenance.ttl, doc/user-contributions.md. Provenance IRI: fair2r:hc-domain-ontologies

2026-05-06 — "link reviewer-side AI policies + reproducibility-crisis literature"

Type: content-prompt Leverage: medium Triggered: Five new bib entries: ioannidis2005, pineau2021reproducibility, neurips_llm_policy, iclr_llm_policy, aclrr_llm_policy. Citations inserted in paper/sections/intro.tex (reproducibility baseline) and paper/sections/background.tex (reviewer-AI policies). Two new fair2r:Claim entries. Artefacts touched: paper/sections/intro.tex, paper/sections/background.tex, paper/references.bib, doc/provenance.ttl. Provenance IRI: fair2r:hc-gap-fill-reproducibility-and-policies

2026-05-06 — "do we have a list of reading tasks for the researcher?"

Type: content-prompt Leverage: high Triggered: New scripts/build_reading_queue.py that produces doc/reading-queue.md from the graph plus the bib. Entries sorted by load-bearing weight (number of fair2r:Claim entries that depend on the source via prov:wasDerivedFrom). Wired into paper/Makefile (make provenance regenerates both analysis fragments and the queue) and into .github/workflows/build-paper.yml. Surfaced on the public site nav as "Reading queue". Artefacts touched: scripts/build_reading_queue.py, doc/reading-queue.md, paper/Makefile, .github/workflows/build-paper.yml, scripts/build_provenance_site.py. Provenance IRI: fair2r:hc-reading-queue

2026-05-06 — "Its still a bit hard to read" — more illustrations

Type: content-prompt Leverage: medium Triggered: Three TikZ figures added on the DLR token palette: the verification-ladder FSM, the F(AI)²R axes 4× 2 grid, the accelerator-vs-blender coupling-rule partition. Each replaces a paragraph that was carrying load by prose alone. Artefacts touched: paper/figures/ladder-fsm.tex, paper/figures/axes.tex, paper/figures/coupling-rule.tex, paper/sections/pattern.tex, paper/sections/authors-note.tex, doc/provenance.ttl. Provenance IRI: fair2r:hc-readability-figures

2026-05-06 — "yes, also perform another pass on the context data"

Type: content-prompt Leverage: high Triggered: The accelerator-or-blender observation graduated to paper text in paper/sections/authors-note.tex as the coupling rule extending What surprised me about the cooperation; corresponding fair2r:MetaContribution hc:obs-accelerator-or-blender flipped from promotedToPaper="pending" to "yes". A third-pass source-research subagent was dispatched in parallel to (a) verify the existing bib entries that remain at lit-retrieved, (b) build new sources for the manifold-vs-extrapolation framing, distributed-cognition philosophical grounding, prompt-engineering literature, and HMC / NFDI4Ing position papers, and (c) run a context-data pass: orphan fair2r:HumanContribution checks, dangling-cite check across paper/sections/*.tex, section-graph mismatch checks, and unactioned-Next: todos in doc/logbook.md. Artefacts touched: paper/sections/authors-note.tex, doc/provenance.ttl, doc/user-contributions.md, paper/references.bib (via the dispatched subagent), doc/sources.md (via the dispatched subagent). Provenance IRI: fair2r:hc-third-pass-research

Aggregate count (back-filled at 2026-05-06)

These are reconstructions; the count is a lower bound on the actual number of contributions, since some sessions made several at once.

2026-05-07 — Add-illustrations pass: hero graphic + two supporting figures

Type: content-prompt Leverage: medium Triggered: Researcher reported the paper "felt a bit hard to read" and asked for a hero graphic plus two to three supporting illustrations. The illustration subagent added one hero (paper/figures/hero.tex plus a hero.standalone.tex wrapper) wired into paper/main.tex front matter and slides/pitch-5min.tex title slide; one writer/reader dual-loop figure (paper/figures/dual-loop.tex) placed in paper/sections/intro.tex; one failure-mode coverage heat-strip (paper/figures/failure-mode-coverage.tex) placed in paper/sections/failure-modes.tex below Table 1. Each figure ships with a .figspec.md per the illustration agent's output format. No new TikZ libraries required. Provenance updated: act:add-illustrations-pass, ai:add-illustrations-pass (first fair2r:AIContribution recorded in the graph), and matching ent:figure-* entities; rdflib re-parse: 1531 -> 1591 triples (+60). Artefacts touched: paper/figures/hero.tex, paper/figures/hero.standalone.tex, paper/figures/dual-loop.tex, paper/figures/failure-mode-coverage.tex, the three matching .figspec.md files, paper/main.tex, paper/sections/intro.tex, paper/sections/failure-modes.tex, slides/pitch-5min.tex, doc/provenance.ttl, doc/logbook.md. Provenance IRI: fair2r:hc-add-illustrations-prompt (matching AI mirror: fair2r:ai-add-illustrations-pass)

2026-05-07 — Verification pass 2: 19 further sources advanced + 4 bib corrections flagged

Type: rule-shape Leverage: high Triggered: User dispatched a second source-analyzer subagent run after the first verification pass landed (PR #11). The subagent walked the 22 remaining lit-retrieved entries (skipping the 3 already on the institutional-access list) and advanced 19 of them to ai-confirmed, each with a verbatim quoted snippet in the new fifth-pass log of doc/sources.md: clarke2009modelchecking, klein2009sel4, alemohammad2023mad, shumailov2024collapse, chen2023drift, luccioni2024power, patterson2021carbon, strubell2019energy, li2023thirsty, birhane2022values, thorp2023chatgpt, sadasivan2023reliably, reynolds2021prompt, liu2023prompt, anderson2024homogenization, kuteeva2024diversity, clark1998extended, clark2025extending, hutchins1995cognition. Direct WebFetch returned 403 at every publisher domain in this pass; all retrievals went through the Exa web_fetch MCP tool, with two web_search_exa assists for OUP Analysis and OUP Applied Linguistics. Four bib corrections flagged for the orchestrator to apply: clark2025extending DOI / volume / pages; kuteeva2024diversity volume / number / pages; liu2023prompt article number; li2023thirsty final-version title. Result: rdflib re-parse 1591 → 1607 triples (+16); rung distribution 51 ai-confirmed, 3 lit-retrieved (all on the institutional-access list), 21 human-confirmed, 10 source-vendored, 2 needs-research. Artefacts touched: doc/sources.md, doc/provenance.ttl, doc/reading-queue.md (regenerated), doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-verification-pass-2

2026-05-07 — Scoping prompt: provenance verification programme

Type: content-prompt Leverage: medium Triggered: User dispatched the research-protocol subagent to author a position-paper-style scoping document at doc/research/provenance-verification.md answering the question "how could the F(AI)²R provenance graph be verified or analyzed?". The doc covers three orthogonal axes (structural / SHACL, semantic / SPARQL, formal / model checking and theorem proving), surveys what is in place today (scripts/provenance_analysis.py), proposes a six-item 12-month verification programme with one-line success criteria, and is explicit about what verification cannot solve (form vs. truth). Three runnable SPARQL examples are inlined and were verified against the live graph before being pasted in. The doc was registered in the public site nav (one PAGES entry plus one NAV entry in scripts/build_provenance_site.py). Artefacts touched: doc/research/provenance-verification.md (new), doc/provenance.ttl (+30 triples; new ent:doc-provenance-verification, act:write-provenance-verification-scoping, hc:provenance-verification-prompt), scripts/build_provenance_site.py, doc/logbook.md, doc/user-contributions.md. Provenance IRI: hc:provenance-verification-prompt

2026-05-07 — Illustrations pass 2: ladder-fsm to Mermaid, rung-distribution figure added, dual-loop retired

Type: content-prompt Leverage: medium Triggered: Researcher prompt: "rerun illustrator with new info — Mermaid, matplotlib, and other publication-ready tools can also be used". The illustration subagent re-ran on the full paper, with no privilege given to the prior TikZ-only pass, and published an audit verdict: keep the six TikZ figures whose argument is shape (hero, axes, pipeline, coupling-rule, eight-practices, failure-mode-coverage); replace the verification-ladder FSM with a Mermaid stateDiagram-v2 source (paper/figures/ladder-fsm.mmd, rendered to .pdf and .png via mmdc) so the manuscript and the Pages site can share one source-of-truth artefact; retire the dual-loop figure that had duplicated the hero on the same opening page; add a matplotlib rung-distribution figure (paper/figures/src/rung-distribution.py, deterministic, DLR palette, rendered to .pdf and .png) as a one-glance companion to the auto-generated rung table in paper/sections/provenance-analysis.tex. Conference deck updated in two frames (verification-ladder frame; rung-distribution frame) to use the rendered PDFs. Provenance updated: act:add-illustrations-pass-2 (informed by the first illustrations activity), ai:add-illustrations-pass-2, hc:rerun-illustrations-prompt, two new ent:figure-* entities (rung-distribution, ladder-fsm-mermaid with prov:wasDerivedFrom ent:figure-ladder-fsm), and prov:wasInvalidatedBy triples on the prior ent:figure-ladder-fsm and ent:figure-dual-loop. rdflib re-parse: 1607 -> 1686 triples (+79). Net figure count: 8 (was 8). Tools used: TikZ (6 figures), Mermaid (1 figure), matplotlib (1 figure). Artefacts touched: paper/figures/ladder-fsm.mmd, paper/figures/ladder-fsm.pdf, paper/figures/ladder-fsm.png, paper/figures/ladder-fsm.tex (now a thin shim around the rendered PDF), paper/figures/ladder-fsm.figspec.md, paper/figures/rung-distribution.pdf, paper/figures/rung-distribution.png, paper/figures/rung-distribution.figspec.md, paper/figures/src/rung-distribution.py, paper/sections/intro.tex (dual-loop input removed), paper/sections/provenance-analysis.tex, slides/conference-30min.tex (two frames updated), paper/figures/dual-loop.tex (deleted), paper/figures/dual-loop.figspec.md (deleted), doc/provenance.ttl, doc/logbook.md. Provenance IRI: fair2r:hc-rerun-illustrations-prompt (matching AI mirror: fair2r:ai-add-illustrations-pass-2)

2026-05-07 — Rerun illustrator with new toolset (pass 3)

Type: content-prompt Leverage: medium Triggered: Researcher prompt: "run the illustrator again with the broader toolset now pinned in agents/illustration.md — decide per-figure whether to refine, replace, or add, capped at 10 total figures; DLR Corporate Design held binding for every tool". The illustration subagent re-ran on the full paper. Audit verdict: keep all eight prior figures (hero, axes, pipeline, coupling-rule, eight-practices, failure-mode-coverage in TikZ; ladder-fsm in Mermaid; rung-distribution in matplotlib) — their renderings are already DLR-CD-compliant and no default theming has leaked through. Add one new figure: a provenance-topology preview (paper/figures/provenance-topology.{pdf,png}, source paper/figures/src/provenance-topology.py), a node-link drawing of the F(AI)²R provenance schema with five PROV-O lanes (Agents / Plans / Activities / Entities / Claims) hand-laid for honest schema-shape but each node's n=<count> badge read at render time from doc/provenance.ttl via rdflib. Wired into paper/sections/provenance-analysis.tex at the section opening, so the reader sees the shape of the graph before reading the audit numbers drawn from it. AI-only nodes carry a // hatch in addition to the soft-blue fill (greyscale-legible). Provenance updated: act:add-illustrations-pass-3 (informed by act:add-illustrations-pass-2), ai:add-illustrations-pass-3, hc:rerun-illustrations-pass-3-prompt, one new ent:figure-provenance-topology. rdflib re-parse: 1703 -> 1736 triples (+33). Net figure count: 9 (was 8; cap is 10). Tools used (cumulative): TikZ (6 figures), Mermaid (1 figure), matplotlib (2 figures). latexmk/pdflatex are not present in the current environment, so make -C paper pdf could not be exercised; the figure render was verified independently by running the script. Artefacts touched: paper/figures/src/provenance-topology.py (new), paper/figures/provenance-topology.pdf (new), paper/figures/provenance-topology.png (new), paper/figures/provenance-topology.figspec.md (new), paper/sections/provenance-analysis.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-rerun-illustrations-pass-3-prompt (matching AI mirror: fair2r:ai-add-illustrations-pass-3)

2026-05-07 — Rule shape: run provenance-curator repair pass with reasons + WIP topology-page transparency

Type: rule-shape Leverage: high Triggered: Researcher prompt: dispatch the provenance-curator subagent to repair the 8 claims missing prov:wasGeneratedBy that the public topology page advertises (Worked verification example, §2), and while doing so, capture, for each defect, a one-sentence reason: "we can learn for our collaborative process". Plus a transparency requirement: the topology page must show both states (BEFORE / AFTER) so a reader sees the audit working, not a sanitised after-image. The provenance-curator subagent: (a) added the 8 missing prov:wasGeneratedBy edges, attaching three to existing act:rev-* activities and minting five synthetic act:meta-cooperation-<date>-<slug> activities for the claims whose landing commits had no act:rev-* parent; (b) recorded a per-claim repair activity carrying fair2r:repairs (new schema property, domain prov:Activity, range fair2r:Claim) plus an rdfs:comment literal explaining why the edge was missing; (c) rewrote the Worked verification example as an honest two-state walkthrough (WIP disclaimer, BEFORE table preserved, AFTER table added, "Why those edges were missing" subsection summarising the dominant pattern, forward-looking next-likely-defect line); (d) appended the cooperative-process observation "Why audited defects are pedagogical, not embarrassing" to doc/user-observations-log.md (hypothesis stage; recommends a no claim without a parent activity refusal rule for agents/provenance-curator.md); (e) appended a logbook entry and this contribution entry. rdflib re-parse: 1736 → 1906 triples (+170). Conformance query against the post-repair graph: 0 rows. Artefacts touched: doc/provenance.ttl, doc/provenance-graph.md, doc/logbook.md, doc/user-contributions.md, doc/user-observations-log.md. Provenance IRI: fair2r:hc-prov-curator-repair-pass-2026-05-07 (matching AI mirror: fair2r:ai-prov-curator-repair-pass-2026-05-07)

2026-05-07 — Content prompt: still waiting for new illustrations

Type: content-prompt Leverage: medium Triggered: User signalled that PR #31 (one new figure: provenance-topology) had been too modest a return on the broader illustration toolset and asked the illustration subagent to deliver substantially more new illustrations under the 10-figure cap. Pass-4 retired eight-practices and rung-distribution and added three new figures: ladder-populations (verification ladder FSM with rung counts as node areas), ai-squared-grid (2x2 writer-vs-reader matrix), and contribution-histogram (human-author contributions by type, read from this file). Artefacts touched: paper/figures/ai-squared-grid.tex, paper/figures/ai-squared-grid.figspec.md, paper/figures/ladder-populations.{pdf,png}, paper/figures/ladder-populations.figspec.md, paper/figures/src/ladder-populations.py, paper/figures/contribution-histogram.{pdf,png}, paper/figures/contribution-histogram.figspec.md, paper/figures/src/contribution-histogram.py, paper/sections/pattern.tex, paper/sections/provenance-analysis.tex, paper/sections/authors-note.tex, paper/sections/eight-practices.tex (retired figure include), doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: fair2r:hc-rerun-illustrations-pass-4-prompt (matching AI mirror: fair2r:ai-add-illustrations-pass-4)

2026-05-07 — "complete the loop": close the one outstanding lit-retrieved source

Type: responsibility-uptake Leverage: medium Triggered: User dispatched a third source-analyzer subagent run to close the loop opened by PR #33, which had added a single new lit-retrieved source (shafer2014xennials, the GOOD Magazine piece popularising the Xennial bridge-generation framing). The "complete the loop" framing is the corrective-loop counterpart to the larger fifth-pass (verification-pass-2) sweep: rather than let one new lit-retrieved entry sit on the queue, the user required it be verified and either advanced or escalated within the same session it was added. Result: source advanced to ai-confirmed, no new escalations, one bib correction flagged (author was wrong — "Stankorb and Oelbaum", not "Shafer" — title was wrong; bibkey kept stable for citation continuity). The contribution is responsibility-uptake because the prompt is about closing an audit gap rather than producing new content; the human author is putting their name to the discipline of "no lit-retrieved source sits on the queue past the session that introduces it". Artefacts touched: doc/sources.md, paper/references.bib, doc/provenance.ttl, doc/reading-queue.md (regenerated), doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:hc-complete-loop-prompt-2026-05-07 (matching AI mirror: fair2r:ac-verification-pass-3-2026-05-07; parent activity: fair2r:act-meta-cooperation-2026-05-07-verification-pass-3)

2026-05-07 — Verification pass 3 (AI mirror): close out shafer2014xennials

Type: corrective-intervention Leverage: medium Triggered: Source-analyzer subagent run (claude-opus-4-7[1m] under the source-analyzer prompt) executed the complete-the-loop prompt: walked the one outstanding lit-retrieved source introduced since the fifth pass (shafer2014xennials), fetched the GOOD Magazine article via the Exa web_fetch MCP tool (direct WebFetch returned 403 against good.is), captured a verbatim quoted snippet in the new sixth-pass log of doc/sources.md, advanced the source from lit-retrieved to ai-confirmed, and flagged a bib correction (the article is by Stankorb and Oelbaum, not "Shafer"; title is "Reasonable People Disagree about the Post-Gen X, Pre-Millennial Generation", not "Generation Xennial"). Bibkey kept stable for citation continuity. Re-parsed the graph via rdflib (2014 → 2016 triples, +2 net) and regenerated doc/reading-queue.md (51 entries, 3 lit-retrieved on the institutional-access list, 47 ai-confirmed, 1 no-rung). Logged this entry as the AI mirror per the no-parentless-claim rule: parent activity act:meta-cooperation-2026-05-07-verification-pass-3. Artefacts touched: doc/sources.md, paper/references.bib, doc/provenance.ttl, doc/reading-queue.md (regenerated), doc/user-contributions.md, doc/logbook.md. Provenance IRI: fair2r:ac-verification-pass-3-2026-05-07

HC-2026-05-13 — Integrate Ara, Vaswani, Croissant; venue research

Type: content-prompt + responsibility-uptake Leverage: high Date: 2026-05-13 Researcher direction: "Compare to https://arxiv.org/abs/2604.24658, integrate. Historical: since https://arxiv.org/abs/1706.03762 things have changed. Croissant https://docs.mlcommons.org/croissant/docs/croissant-spec.html. Potential venue for f-(ai)2-r: https://www.rd-alliance.org/plenaries/p27/ what format would you suggest? Also research other venues sort by chance of acceptance and impact." Outcome: Three external works named-and-positioned (Liu 2026 Ara as adjacent maximalist proposal F(AI)²R declines; Vaswani 2017 as historical pivot at the end of intro.tex; Croissant 1.0 as adjacent vocabulary that composes rather than competes, in limits-and-objections.tex + cross-ref from appendix-g-pattern-extensions.tex). Three new bib entries, three new fair2r:Sources, three new fair2r:Claims with prov:wasGeneratedBy back to act:integrate-external-2026-05-13. Venue shortlist refreshed in doc/submission-plan.md, ranked by (acceptance × impact) with fit as tie-break; top three are arXiv, RIO Journal, and RDA P27 (BoF + poster). For RDA P27 specifically, the recommended format is a BoF + poster + A4 handout with WG-chair outreach this week since formal sessions CfP closed in Nov 2025. Artefacts touched: paper/sections/intro.tex, paper/sections/limits-and-objections.tex, paper/sections/appendix-g-pattern-extensions.tex, paper/references.bib, doc/provenance.ttl, doc/submission-plan.md, doc/logbook.md, doc/user-contributions.md, plus slide-scrutinizer dimensional fixes in slides/pitch-5min.tex, slides/conference-30min.tex, slides/poster-A0.tex and vendored paper/main.pdf. Provenance IRI: hc:integrate-external-2026-05-13

2026-05-17 — Conviction: true novelty is a human quality LLMs do not match yet

Type: content-prompt Leverage: high Triggered: New paragraph "What I still believe the work is for" in paper/sections/authors-note.tex (after the coupling-rule paragraph, before Tokens out, tokens in.). The researcher supplied the substantive belief and its careful hedge (yet, not never); the model drafted it into the Author's Note voice and tied it to the existing "would the abstraction have been thinkable without the model?" diagnostic. Artefacts touched: paper/sections/authors-note.tex, doc/provenance.ttl (claim + activity + contribution + agent:claude-sonnet-4-6 declared), doc/logbook.md, doc/user-contributions.md. Provenance IRI: hc:authors-note-novelty-conviction

2026-05-17 — RDA P27 Ideas Pitch draft + OECD AI-in-Science source

Type: content-prompt Leverage: medium Triggered: doc/rda-p27-ideas-pitch.md (fill-in-ready P27 Ideas Pitch submission draft, thematic area = Engineering and technology per the author's selection); OECD report added to the source pool — paper/references.bib (oecd2023aiscience), doc/provenance.ttl (src:oecd2023aiscience + a prov:wasDerivedFrom edge on claim:equity-not-neutral + activity + contribution), \cite in paper/sections/limits-and-objections.tex Equity paragraph, new doc/sources.md §13 entry. Artefacts touched: doc/rda-p27-ideas-pitch.md, paper/references.bib, paper/sections/limits-and-objections.tex, doc/provenance.ttl, doc/sources.md, doc/logbook.md, doc/user-contributions.md. Provenance IRI: hc:add-oecd-aiscience-source

2026-05-18 — Seventh source-analyzer verification pass

Type: responsibility-uptake Leverage: medium Triggered: a source-analyzer subagent run with a live-web research loop over the known-defect list and the recently-added, never-human-checked sources. Found and corrected the lagging src:vannoorden2023chatgpt graph mirror (wrong DOI d41586-023-03907-5 -> d41586-023-03930-6, both creators added, full title, rung lit-retrieved -> ai-confirmed); aligned the src:benjelloun2024croissant graph title to the cited spec @techreport; updated the iclr_llm_policy bib url to the confirmed stable policy page and added an icmje2023 note for the January-2026 revision; re-confirmed ravi2024fair4ml, oecd2023aiscience, liu2026ara, vaswani2017attention against live records with no change. The AI side of this corrective contribution is mirrored in doc/provenance.ttl as ai:verification-pass-7-2026-05-18. Artefacts touched: paper/references.bib, doc/provenance.ttl, doc/sources.md, doc/logbook.md, doc/user-contributions.md, doc/reading-queue*.md (regenerated). Provenance IRI: hc:verification-pass-7-prompt-2026-05-18

2026-05-18 — Slide-deck reconciliation with the post-split manuscript

Type: content-prompt Leverage: medium Triggered: a presentation-agent reconciliation pass after a session that split the appendices into a standalone companion technical report, integrated three external works (Vaswani 2017, Liu 2026 Ara, Benjelloun 2024 Croissant) and OECD 2023, and added the Author's Note conviction "What I still believe the work is for". All three slide artefacts brought back into consistency: companion + A0 poster added to the pitch/poster artefact lists; the conference "Recursive attestation" frame's stale numbers corrected against the live repo (1300+→2500+ triples, 50→59 bib entries, 45+→55 reading-queue sources, precise site page-count replaced with honest qualitative wording) and the ~22 pp paper / ~14 pp companion split folded in; Vaswani as the historical pivot in conference §1 and the pitch problem frame; a new conference §8 frame for Ara and Croissant framed exactly as paper/sections/limits-and-objections.tex frames them; OECD as the policy-body equity anchor; the novelty conviction (with its yet hedge intact) surfaced as the closing message on both Thank-you frames and the poster. No new slide-only claim. The AI side of this content contribution is mirrored in doc/provenance.ttl as ai:reconcile-slides-2026-05-18 (agent:claude-sonnet-4-6). Artefacts touched: slides/pitch-5min.tex, slides/conference-30min.tex, slides/poster-A0.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: hc:reconcile-slides-2026-05-18

2026-05-18 — Slide visual where it helps the message (adjacent-proposals diagram)

Type: content-prompt Leverage: low Triggered: Standing direction to use a non-photorealistic visual on slides where it strengthens the message. Confirmed no image-generation connector exists; built paper/figures/adjacent-proposals.tex (deterministic TikZ, DLR Corporate Design) and converted the conference Adjacent proposals and adjacent vocabularies frame from two dense bullets to the positioning diagram + one-line caption. Judged the rest of the 2026-05-18 reconciliation changes textual by nature and added no figures elsewhere (no-bloat). Artefacts touched: paper/figures/adjacent-proposals.tex (new), slides/conference-30min.tex, doc/provenance.ttl, doc/logbook.md, doc/user-contributions.md. Provenance IRI: hc:adjacent-proposals-figure-2026-05-18 (AI mirror ai:adjacent-proposals-figure-2026-05-18)