Revision log: 05-maynard-risk-and-ai-map.md#
Revision of 26 September 2026 in response to fidelity.md, breadth.md and coherence.md. Each issue was checked against working/maynard/corpus/, manifest.json, book/FFTF-*.txt, concept-index.md and timeline.md before any edit. Modem Futura podcast posts were not opened or used, as the user asked. After editing, a script re-checked every quotation in the map against the corpus and book (all found; nested quotes use single marks), every date–slug pair against the manifest, and confirmed that no Modem Futura promo post is cited.
Outcomes: Fixed, Partly fixed (with reason), Not changed (with reason).
Fidelity review#
| # | Issue | Verification | Outcome |
|---|---|---|---|
| F1 | Frontier-AI paper’s analysis (incentive field, conversion channels) treated as his own shift | Confirmed: 2026-07-04 says Fable applied his work “in a way that hadn’t previously occurred to me” and calls the analysis “genuinely novel”; drafting credited to Claude; “incentive field” occurs only in 2026-07-16 | Fixed. §1 provenance now cites 2026-07-04 and lists the incentive-field analysis and conversion channels as possibly Fable’s. C10, T5, §7 change 6, tension 4 and lens D3 now lead with his own structural prose (2022-02-12, 2023-11-18, 2024-07-13) and treat the paper as formalisation [mixed]. “Explaining developers” rewritten as continuity, not a psychology-to-structure shift. |
| F2 | 2026-01-17 misdescribed as Claude-drafted; larger AI credit omitted | Confirmed: post is his account; Claude drafted the arXiv paper and developed the vigilance-bypass mechanisms; 2026-01-10 was “primarily based on my own thinking” | Fixed. §1 entry rewritten; cognitive-Trojan-horse thesis anchored on 2026-01-10; §5.8 row notes the paper’s mechanism account is mixed; [mixed] tag removed from his own 2026-01-17 reflection (tension 13, T8). |
| F3 | “One of the scariest things” quoted without counterweight | Confirmed: sentence continues “the potential is profound”, followed by “neither an AI optimist nor an AI pessimist” | Fixed. Full “stuck between” framing in §2 and §7 change 8; removed from T3. Own-prose “worries me — a lot” (2026-05-10) now leads. |
| F4 | Formation rated Core from 2024 on misread sources | Confirmed: 2024-10-20 does not discuss formation; 2026-06-10 means formation of learning outputs; only explicit statement is the lecture | Fixed. Row re-rated “Rising (2026)”, 2026-06-10 dropped, 2024-01-01 and 2026-07-19 given as precursors, lecture marked as the only explicit statement. |
| F5 | Mapmaker coinages presented as his concepts | Confirmed: “reverse formation”, “validation gap”, “second-order evolutionary mismatch” have zero corpus hits (“evolutionary mismatch” alone is in 2026-01-10 as a borrowed term) | Fixed. Added a † convention for mapmaker labels (§1); “reverse formation” now appears only as a flagged label beside his wording; “validation gap†” given his wording; “second-order” removed. |
| F6 | Changes-of-mind table mixes inferred and signalled shifts; brand-nano misquoted; stochastic-parrot row misread | Confirmed on all counts (2018-02-21 “brand-nano”, no “helped build”; 2025-01-05 and 2025-02-23 show no reversal) | Fixed. Table split into “Signalled by him” and “Inferred (my reading)”; brand-nano row recast as a sceptical insider’s reassessment with correct spelling; stochastic-parrot row replaced by a “not a change of mind” note. |
| F7 | AGI “irrelevant” to his risk map | Confirmed: scoped to “this conversation” about loss of control; “All of those might happen” | Fixed in §5.6 and tension 14. |
| F8 | 2026-08-30 quotes spliced; “failing” overstated | Confirmed: footnote 7 anchors to “may not be up to the task” and ends “But there’s always hope”; footnote 6 softens “guardians of the past” | Fixed. Tension 10 rewritten with both qualifiers, retagged [he says so]; “describes as failing” removed. |
| F9 | Pippard’s ladder dated 2023; 2023-05-04 misdescribed as a Future Rising excerpt | Confirmed from 2023-05-04 and 2024-08-18 | Fixed. §1 source list corrected (adds 2020-10-30, 2024-05-19, 2024-09-08); tipping-points row and T6 now date the idea 2018 draft → 2020 book → 2023 revival. |
| F10 | Citation slips in §5.8 rows | Confirmed: “easy button” only in 2026-09-24; “cognitive surrender” not in 2026-05-10; “constitutive resonance” only in 2026-05-21 | Fixed. Both rows re-sourced; Shaw and Nave credited for “cognitive surrender”. |
| F11 | “16 posts in eleven years” | Confirmed: posts run 2018-12-13 to 2026-08-02 | Fixed (“about 16 posts between 2018 and 2026”). |
| F12 | Hedges dropped (“childish irresponsibility”; “plans just on the off chance”; manipulation reaffirmed “every year since 2023”) | Confirmed in 2024-05-21, 2025-04-06, 2025-07-06 | Fixed. Restored “sometimes seems” and “AI companies like OpenAI” (T7, Appendix A); “plans” → thinking about responsible innovation “just on the off chance”; C14 now says restated explicitly in 2023 and 2025. |
| F13 | §8.2 overstates the strictness of the 2018 plausibility rule | Confirmed: FFTF p.281 calls the razor never “more than an aid” and leaves “the door open” | Fixed. Added to C7, tension 2 (“a change of emphasis, not a reversal”) and §7 change 3; §5.5 Occam row updated. |
| F14a | “who owns you” is FFTF p.146 | Confirmed | Fixed. |
| F14b | “a bedrock of reality” is a passing aside in 2019-09-04 | Confirmed | Fixed. Deepfakes row led by 2024-02-25; 2019-09-04 marked “passing”. |
| F14c | Artisanal-intellectual row omits 2025-02-16 | Confirmed; the essay body was drafted with Deep Research | Fixed (added as “his framing” in the new §5.10 knowledge row). |
| F14d | Deepfake letter dated 2025–26 | Confirmed as 2024-02-25 | Fixed (tension 12). |
| F14e | “Nature doesn’t care” capitalised | Confirmed lower-case mid-sentence | Fixed (C13). |
| F14f | Bare dates shared by two posts (2025-01-07, 2023-04-24, 2023-10-30) | Confirmed; a script also found 2016-01-11 and 2023-04-04 shared | Fixed. §1 conventions now say which post a bare date means; the second post on those dates is always named (e.g. the 2023-04-04 Substack launch post); the 2025-01-07 Modem Futura promo is excluded. |
Breadth and proportion review#
| # | Issue | Verification | Outcome |
|---|---|---|---|
| B1 | One recent mixed-provenance lecture sets the endpoint of most arcs | Confirmed (30 citations; sole source in several places) | Partly fixed. Own prose now leads wherever it exists (inevitability 2025-08-31/2026-05-21; “categorical error” 2025-03-15; AGI 2026-04-11; binary 2024-03-31; reversals 2020-10-15, 2024-02-25, 2025-11-19); lecture removed from C14, T3, T10, §5.4, §5.10 and tension 10; remaining uses marked “single source” or paired with his own prose. Occurrences of the date fall from 30 to 24, most now in provenance notes, paired citations or explicitly scoped claims; the cap of 10–12 was not reached because several claims (publics’ role, informed speculation, “I was wrong”, the AGI scoping) have no other source and are needed for accuracy. |
| B2 | Education and learning missing as a territory | Confirmed (87-post thread; concept-index Family 11 dropped) | Fixed. New territory row, C17, §5.10 (11 rows), T10 thread, §2 paragraph, open question, lenses E3–E4, Appendix A T10 sources. |
| B3 | Being human under-built; “flourishing” absent | Confirmed (2025-01-30 prompt text; 2026-07-10; 2026-08-16; 2026-05-21) | Fixed. New C1 (thriving), §2 purpose paragraph, rewritten being-human paragraph, §5.9 expanded (flourishing, “normal” vs “human”, intelligence ≠ goodness, learning to be human), T11 thread; being human, justice and universities added to “What stays constant”. |
| B4 | Risk perception, moral panic, techlash missing | Confirmed (2016-03-12, 2024-02-18, 2024-08-07, 2024-09-01, 2025-06-01, 2026-07-10) | Fixed. §5.1 row, C9 rewritten, T1 paragraph, §3 connection, lens D6. |
| B5 | Non-AI emerging-technology risk work, especially neurotech, missing | Confirmed (quotes in 2016-03-31, 2020-08-28, 2020-10-15, 2024-09-18, 2024-11-17 verified; “mental monoculture” is Gordon and Seth’s term) | Fixed. §2 sentence, §3 connection and territory row, T2 “proving ground” paragraph, §5.8 neurotech row upgraded to “Recurring (2016–2024; the bridge to AI)”, §5.3 indentured-servitude row. |
| B6 | Centrality labels contradict rubric and favour 2026 | Confirmed | Fixed. New “Rising” grade; Trojan horse, relational technology, formation, constitutive resonance and what-we-do model re-graded Rising; catastrophe and mundane-but-serious re-graded Occasional; hazard vs risk raised to Core; several others lowered; Core now about two dozen; §4 retitled “with firmness and trajectory”. |
| B7 | Structural account of developers dated to 2026 | Confirmed (2022-02-12, 2023-11-18, 2024-03-21, 2024-07-13) | Fixed (see F1). |
| B8 | Ethics-to-risk reframe under-weighted | Confirmed (2019-11-01, 2021-08-03, 2023-04-04, 2024-12-17) | Fixed. §5.2 row, sentence in C3 and §2. |
| B9 | Democracy, misinformation, cohesion, drain of agency, x-risk ideology missing | Confirmed (2023-05-31, 2023-07-25, 2024-01-17, 2024-04-28, 2024-08-25, 2024-11-24, 2024-07-21) | Fixed. Three §5.7 rows, C15, T3, §2. |
| B10 | Moral status of AI dropped | Confirmed (2023-08-18, 2023-08-23, 2023-09-28, 2024-06-16) | Fixed. §5.6 row and a clause in tension 14 (noting it does not recur in 2025–26). |
| B11 | Chemical/nano carry-overs thin | Confirmed | Fixed. Dated “What he carries from chemicals and nanomaterials” list in T2; non-linear dose–response added to C4 and §5.1. |
| B12 | “No quantitative treatment” overstated; comparative risk dropped | Confirmed (2023-11-09 “back-of-the-envelope calculations”) | Fixed. Gap reworded; §5.1 row added. |
| B13 | 2019–22 and Future Rising ideas under-used | Confirmed (2024-09-08 quotes are his own compiled book quotations) | Fixed for “stealing the futures of others”, “early warnings”, fallible science (new §5.5 row), trustworthiness (engagement row), researchers and hype. Latency added only as a co-written note, since 2019-08-13 is weaker evidence. |
| B14 | AI and the future of knowledge treated only as provenance | Confirmed | Fixed. §5.10 “Knowledge, expertise and validation” row; T10. |
| B15 | Lenses omit care, capacity-building, perception; lens 9 on a one-off | Confirmed | Fixed. Lenses E3, E4, D6 added; D3 re-sourced to 2024-07-13 first. |
| B16 | Modem Futura promo counted in timeline figures | Confirmed (2024-11-06, medium relevance) | Fixed in the map (§1: 279 posts; 2024 = 64). timeline.md itself not edited, as it is outside this task’s target file. |
Coherence review#
| # | Issue | Verification | Outcome |
|---|---|---|---|
| C1 | No single architecture; diagram contradicts prose | Confirmed; his self-descriptions (2026-07-10, 2026-08-16, 2026-09-24 intro) put navigation and flourishing on top | Fixed. §3 now states one architecture (purpose, engine, disciplines, moral axis, object, arenas, manner), marked as my reading; diagram redrawn as a hub with labelled arrows; being human rewritten as both at stake and sought; threat-to-value → being human link cited to 2023-11-26 and 2024-08-25. |
| C2 | Positive programme (thriving, value creation) almost absent | Confirmed | Fixed (see B3): C1, §2 purpose, value as the unifying concept, tone restated as “worry and wonder both sharpen”, lens A2. |
| C3 | Education has no territory, commitment or thread | Confirmed | Fixed (see B2). |
| C4 | Recent and mixed material in load-bearing parts; Core overused; lens 6 reproduces the “four filters” | Confirmed | Fixed in substance (see B1, B6); lens 6 restated in his 2018 terms as B3; provenance and era tags added to every lens. |
| C5 | “Arguments rarely move him” contradicted by the map’s own evidence | Confirmed (2023-05-25 “But there is a ‘but’ here”; 2024-06-30 “compelling arguments”; 2025-04-06) | Fixed. “How he updates” rewritten on the reviewer’s formula; consciousness shift added to the inferred changes table. |
| C6 | Tensions not separated into his admissions vs mapmaker inferences; four overreach | Confirmed for tensions 4, 9, 10, 14 and §2’s “public engagement” decline | Fixed. Every tension tagged [he says so] / [partly his] / [my reading], with an accurate count; tensions 4, 9, 10, 14 rewritten; “public engagement” removed from declining remedies. |
| C7 | Missing tensions and links (too human-centric; safety message first; RI nested in risk innovation; universities as ambivalence) | Confirmed (2023-05-31; 2026-05-10; 2020-07-30; 2016-01-31 and 2026-08-30) | Fixed. Tensions 15 and 16 added; 2020-07-30 nesting added to §3, C3 and the RI row; §7 change 7 rewritten as standing ambivalence. |
| C8 | Too many non-aligned partitions; ~15k words; §6 repeats §4 | Confirmed | Partly fixed. One architecture now orders §3–§4 and roughly §5 and §9; a crosswalk table links territories to commitments, concept tables, lenses and theme files; §6 “Core” paragraphs cut to one line and all “Connections” paragraphs removed; §2 tightened. The 9–10k-word target was not adopted: the breadth review required substantial additions (education, being human, perception, neurotech, systemic risk), and the map is now about 22k words. Further cutting would drop coverage the breadth review found missing. |
| C9 | Commitments not distinct; not all “held consistently”; no commitment on stories | Confirmed | Fixed. 19 → 17: “who decides” and “no one decides alone” merged (C5); symmetry and steering merged (C6); manipulation and language/formation merged with layered firmness (C14); humility moved into a method note, which also carries stories and imagination; thriving (C1) and education (C17) added; heading retitled. |
| C10 | Opening sentence misstates his position | Confirmed (FFTF p.22) | Fixed. |
| C11 | Lenses mirror commitments; unweighted; no defaults, provenance or missing lenses | Confirmed | Fixed. 24 lenses in six groups, each with “his usual landing”, era and provenance tags; added lenses for value creation, tool vs relationship, stories, care, capacity, perception. |
| C12 | Mapmaker labels presented as his terms | Confirmed (zero corpus hits for “symmetric risk”, “language turn”, “reversibility test”, “universal vulnerability”, “cognitive exposure”, “catastrophe as mass loss of value”, “analogy as mindset”) | Fixed. † convention added and applied throughout §3–§9. |
| C-minor | “Three constraints”; extinction statement “important”/”a lot of sympathy”; “futile” conditional; “too narrow (2024)” unsourced; “AI present from 2015” | Confirmed; no source found for “too narrow (2024)” | Fixed. “Three features”; sympathy added to C15; conditional kept with “he hopes not”; “too narrow” deleted; diagram now says “one converging strand to 2021; central from 2023”. |
Additional corrections found during verification#
- Tension 7: “He notes the irony himself” was attached to a footnote about footnotes (2026-05-10 fn9). Re-sourced to his remark on turning to “the very source of the problem” (2023-04-05) and retagged [partly his].
- Duplicate “fix frame” row removed from §5.2 (now in §5.9 with “normal” vs “human”).
- Full slugs restored for Appendix A entries that had been truncated.