MIND THE GAP
Planned Changes — v6
The 42-item catalogue. Three new items added 23 May 2026. 23 May 2026.
This document captures every planned change to Manuscript v1. v6 adds three items: Item 40 (The Spine — new opening section before Chapter 1, with Paul's 600 words verbatim as the source material); Item 41 (OOAI definition correction — Outcome-Oriented AI, not the compaction-summary hallucination 'Out Of AI'); Item 42 (Attribution discipline — per-chapter sign-off lines removed; supplementary book documents stripped of footers; continuity documents retain attribution as working record). Item 40 is structurally significant: it gives the book its thesis statement up front, integrating with the Gap Definition Triangle (Items 1, 36, 39) and triggering light integration passes in Chapters 1, 7, and 12.
Categories
FACTUAL — correction needed to a stated fact (date, name, claim, statistic).
AMPLIFY — add material that exists in Paul’s experience but not yet in the book.
STRUCTURAL — reorganise, list, or restructure existing material.
STRUCTURAL PRINCIPLE — a chapter-spine test or rule that applies across the whole manuscript.
REPETITION — the book is saying something twice; one instance needs work.
DRIFT — the book itself has drifted on its own argument or terminology.
INTEGRATION — substantial integration of external material into an existing chapter.
OPEN QUESTION — a question requiring Paul’s decision before any change can be made.
LIKE — affirmation; no change required.
MARKER — not an editorial item (number skipped, status marker).
Phases
V2 — belongs in Manuscript v2 editorial work. Twelve chapters only.
RESOLVED — decision made; will execute during v2 chapter work.
PHASE 2 — front matter, rear matter, acknowledgements, bibliography. Logged for phase two assembly.
POST-V2 — future edition or companion volume; not for this revision.
RETIRED — marker only; nothing to action.
The 42 Items
Item 1 — Chapter 1 — Human Recognition
Category: AMPLIFY Phase: V2
Observation: It says 'this is the gap' and I just feel I wonder if we need to really succinctly say what that gap is. It's the gap between AI and human, between artificial intelligence and human intelligence. I think that's what this gap is and maybe that's the point we introduced that, which is actually at the title of the book.
Decision / Note: Land the definition of 'the gap' at first use. The gap is between AI behaviour and human behaviour (per Item 36). Pairs with Item 36 and resolves at the same time.
Item 2 — Chapter 1 — page 7 (reinforcement learning section)
Category: AMPLIFY Phase: V2
Observation: It's important that we say we're also shaped by what we're not rewarded for. We're not only a reward system; we're also a response system to not getting a reward or getting negative feedback. What happens with AI on the side, the edges that don't get rewarded? Where does that go? I'm leaving that as an open question.
Decision / Note: Add negative reinforcement to the human side of the reinforcement learning argument. The AI open question is partially resolved by the framework citation in the new Chapter 6 — reference forward.
Item 3 — Chapter 1 — page 8 (author bio paragraph)
Category: FACTUAL + AMPLIFY Phase: V2
Observation: I'm a psychotherapist and coach. I've worked for 30 years in boardrooms and consulting rooms on factory floors. I think it needs to add: in cancer wards, on my own cancer ward journey. My LinkedIn profile is probably the place to go for that.
Decision / Note: SUPERSEDED IN PART by Item 38 (21 May 2026). Item 38 carries the full LinkedIn-aligned characterisation that this item asked for. Item 38 is the canonical scope; Item 3 retained for historical record but execution flows through Item 38. Coordinate with Items 4 (chronology) and 38 in a single bio rewrite during Chapter 1 v2 work.
Item 4 — Throughout — 'thirty years' as a stable phrase
Category: FACTUAL Phase: V2
Observation: Page 9 and throughout the document we talked about 30 years. I've had 30 years of working with people. I've only had about 15 years in clinical practice. Started coaching in 2009 so 20 years. Professional career in the 90s so 40 years. We need to clarify this throughout.
Decision / Note: RESOLVED 21 May 2026. Four-number rule locked. (a) 40 years — total professional life (whole career). (b) 37 years — commercial practice specifically (1978–2015). Preserved at line 1111 only. (c) 20 years — behaviour, coaching, listening (~2006–2026). (d) 10 years — clinical psychotherapy specifically. 'Thirty years' as standalone phrase — RETIRED. Per-claim assignments: CLAIM A (whole career) → 40 years at lines 199, 1903, 2023. CLAIM B (attending to behaviour, the instrument) → 20 years at lines 139, 205, 831, 1147, 1319, 1399, 2107. CLAIM C (therapy room, clinical) → 10 years at lines 229, 723. SPECIAL CASE line 1125 — RED. Both halves wrong; must become 'forty years in commercial environments and ten years in psychotherapy'. SPECIAL CASE line 1111 — UNCHANGED at thirty-seven years (precise 1978–2015 chronology, load-bearing for Chapter 7).
Item 5 — Chapter 2 — page 13 (five moments of misreading)
Category: STRUCTURAL + AMPLIFY Phase: V2
Observation: In that whole previous section we talked about the five traps. It needs to say trap one, trap two, trap three, trap four, trap five — they need a listing. Also: 'what are you projecting onto it? What do you need it to be? How much is that shaping what you are hearing?' That hooks to the NLP phrase — projection is perception.
Decision / Note: Number the five traps explicitly (Trap 1 through Trap 5). Add the NLP projection-is-perception link at chapter close.
Item 6 — Chapter 3 — page 14 (Hidden Architecture, therapy room intro)
Category: AMPLIFY + LIKE Phase: V2
Observation: I love Hidden Architecture, Chapter 3, page 14. There are two people with those character types. I bring my mum, I bring my dad, I bring my other sub-personalities. I have some awareness about which to silence and which to take. Also: in a two-hour therapy session, that's upwards of 10,000 to 15,000 words. That's a third to a half of the book between two people in one single session.
Decision / Note: Add Paul's first-person account of bringing his cast of characters. Add the 10,000-15,000 word session statistic as scale reference.
Item 7 — Chapter 4 — page 20 (drift example dialogue)
Category: STRUCTURAL Phase: V2
Observation: On page 20 where we start the drift. After the breakfast section you need to add questions: 'What's the motorway like? Has there been any roadworks? I hear that Crewe town centre's been dug up.' Just check on the local accessibility then go to 'remind me of the price again.'
Decision / Note: Add intermediate questions to the drift dialogue to make the migration visible to the reader.
Item 8 — Chapter 4 — page 23 (Nairobi slide story)
Category: FACTUAL + AMPLIFY Phase: V2
Observation: My recollection is I got an email from my holiday company about a holiday in Africa, flight schedules to Nairobi. It was a separate parallel conversation that I suspect got into my context window. Alternatively the imagery was done in ChatGPT, moving data from Claude. I have the slide with Nairobi and the image of the Crewe conversation in my library.
Decision / Note: Two parts. Clarify the actual cause of the Nairobi drift. INCLUDE as a photo collage of the actual drift in real time — not a single slide but a visual evidence sequence showing the drift unfolding. Brief caption to name the date and the actual cause (parallel email or ChatGPT context). DECISION TAKEN 21 May 2026.
Item 9 — Chapter 4 — page 25 (drift commentary)
Category: AMPLIFY Phase: V2
Observation: Even in human conversations there is always drift. If my mum's going on holiday next Thursday, I know it's the 24th. In four months I won't remember it was the 24th. In two years I won't remember it was May. I will remember it was 2026. Also — I double bind a lot of my AI work by having one AI check the work of another. I've actually used my son's AI. I think that's worthy of inclusion.
Decision / Note: Add 'drift is normal in humans too' commentary using the holiday date example. Add the double-bind verification practice as a discipline.
Item 10 — Chapter 5 — page 27 (Compression, the tunnel opening)
Category: AMPLIFY + STRUCTURAL Phase: V2
Observation: I think the word in photography is 'bokeh'. Depth of field is probably a better word. With the girl in the photo, I can absolutely see the shine in the eyeball. The walls are part of the depth of field and the view beyond is the bokeh. I was out with my camera one day and I photographed this tunnel. I think that needs storyboarding.
Decision / Note: Introduce bokeh / depth of field photographic vocabulary. Storyboard the origin of the photograph more clearly before the liminal moment lands. Note Item 12 — the tunnel appears twice in Chapter 5.
Item 11 — Chapter 5 — page 29 (context window sizing)
Category: AMPLIFY + FACTUAL Phase: V2
Observation: You infer the size of the context window. We need to size this. I've got a paper about words in my therapy room over 10 years comparing it with 300 billion words ChatGPT-3 started with. Anthropic discussed context window and it came out as one day will be infinite. Also Anthropic have developed dreaming — the nightly consolidation of memory. It's ironic that Anthropic this week are using human terms like dreaming at a time when I'm preparing the book warning against treating AI as human. At the optician's yesterday I was referring to Claude as 'he' and they picked me up on it.
Decision / Note: Three additions. Size context window with real numbers. The Anthropic 'dreaming' irony (serves the book's thesis). The optician story as a caught-in-the-act moment. Coordinates with Item 21 on context window claims.
Item 12 — Chapter 5 — throughout (Compression repetition)
Category: REPETITION + LIKE Phase: V2
Observation: On chapter five, Compression and Liminal Moment, I felt like we described the girl in the tunnel twice. The opening needs rewriting to avoid duplication with the close. I think it's a great chapter.
Decision / Note: Affirmation plus repetition fix. Opening rewrite needed to avoid duplication with close.
Item 13 — (13 skipped — unlucky number)
Category: MARKER Phase: RETIRED
Observation: Paul deliberately skipped 13 as an unlucky number.
Decision / Note: No editorial action. Skip in numbering preserved.
Item 14 — Chapter 6 — page 35 (Fluency, therapy room section)
Category: AMPLIFY Phase: V2
Observation: In the therapy room you're taking in a lot more variables: where are their eyes, tone of skin, colour of skin. The facial palette shifts. The pores open or close. There'll be a finger move, a foot tap, a glance, a stumble. The poker tell. That's the fluency interrupt. It's almost impossible to pick that fluency interrupt up from AI. Maybe AI needs a visual cue — like a geometry in a car that tells you how fast you're going round a bend.
Decision / Note: Large amplification. Adds visual / somatic dimension of clinical listening. The phrase 'fluency interrupt' is potentially a coined term worth elevating. Lands in the rewritten Chapter 6 (see Item 16).
Item 15 — Chapter 6 — page 35 (verification, fact vs supposition)
Category: AMPLIFY Phase: V2
Observation: I thought I had something in my AI that said whether data was fact, supposition, or implication. I thought I'd categorised my data in four ways. Have I missed something or is that something we should be talking about as part of prompting capability?
Decision / Note: RESOLVED 21 May 2026. YES — include fact/supposition/implication categorisation as a prompting discipline. Lands in the rewritten Chapter 6 as part of the practical close alongside cascade and verification material.
Item 16 — Chapter 6 — directional defence integration (RESOLVED)
Category: INTEGRATION Phase: RESOLVED — V2
Observation: The clinical parallel starts to describe internal shame versus external shame. The externaliser points in two directions at the same time, conflates, waffles, creates a mask. The internaliser carries it to the next conversation, like a wound. There is a further document called Defending the Gap which takes the theory of there being a gap and the way we defend it.
Decision / Note: PLATINUM. DECISION: Mind the Gap ends with Chapter 6. Defending the Gap material integrates into a rewritten Chapter 6. Current Chapter 6 (Fluency and False Confidence) becomes setup; Defending the Gap becomes the chapter's structural argument; cascade and verification material from 6a(4) folds back as the practical close. Single chapter, not 6 plus 6a. Citation to Where Does The Shame Go? (Roebuck, 2026, working theory paper, to be available on paulroebuck.co.uk ahead of book launch). Framework scope held tight — no Hunter-Farmer, APC, ADHD, or wider applications introduced. The clinical observation does the work the book needs. The wider framework lives elsewhere. This decision resolves Item 16 fully.
Item 17 — Chapter 7 — opening (factory floor age)
Category: FACTUAL Phase: V2
Observation: It says in 1978 I would have been age 16 not 18. I started my apprenticeship at the age of 16.
Decision / Note: RED. Simple factual correction. Sixteen, not eighteen.
Item 18 — Chapter 7 — page 42 (Blakeborough, six sessions)
Category: FACTUAL + LIKE Phase: V2
Observation: Page 42 is the Blakeborough story. It's absolutely riveting listening. It says a person might come forth after six sessions. I never do six sessions. You might say six hours or two sessions.
Decision / Note: Affirmation plus factual correction. Two sessions or six hours, not six sessions.
Item 19 — Chapter 7 — third Plugs reference
Category: STRUCTURAL + AMPLIFY Phase: V2
Observation: Later in Chapter 7 you revisit Plugs. We've had Plugs earlier twice so I think we need to say a third reference. I'd shouted at it, I'd observed it, this time I was fully engaged with Plugs. Plugs was as much a part of my daily AI routine as Claude and ChatGPT. That's the day I asked it to help me produce that post for Facebook.
Decision / Note: Frame the third Plugs reference as the working partnership phase. Coordinate with Item 29 (cut Chapter 11 Plugs ending).
Item 20 — Chapter 7 — chronology
Category: STRUCTURAL Phase: V2
Observation: Chapter 7 gives the whole story of my chronology. It calculates the years. We need to use that as a reference point because we keep saying 30 years. I've got professional years, therapy years, and years before that. That captures it quite well.
Decision / Note: Use Chapter 7 chronology as the master reference for resolving Item 4 across the manuscript. Single source of truth.
Item 21 — Chapter 8 — infinite context window
Category: FACTUAL Phase: V2
Observation: In Chapter 8 you've got a heading saying Anthropic have announced an infinite context window. They haven't announced it. What they've said is that it's an aspiration. There is still a limit. I'm not sure they'll ever be such a thing. They've said clearly it's aspirational.
Decision / Note: RED. Correct from 'announced' to 'aspirational'. Touches AI sign-off using same phrase.
Item 22 — Chapter 8 — page 47 (outcomes not prompts)
Category: FACTUAL Phase: V2
Observation: Just to clarify, Anthropic have not announced a feature called Outcomes. I think you've drifted there or you made that up or you've hallucinated. The evolution is ChatGPT or OpenAI talking about /Goal. Outcome-oriented prompting is my word.
Decision / Note: RED. Book has hallucinated an Anthropic feature attribution. Correct to /Goal (OpenAI). Frame 'outcome-oriented prompting' as Paul's coined phrase.
Item 23 — Chapter 9 — opening (sub-personalities intro)
Category: STRUCTURAL Phase: V2
Observation: Chapter 9 needs an intro section. It goes straight into sub-personalities. It needs to position itself: we've spent a lot of time thinking about AI. We've got to start with ourselves. Who are we? We're not one thing, we're the sum of the parts.
Decision / Note: RESOLVED 21 May 2026. Merges with Item 28 — the 15 November 2006 recognition quote opens Chapter 9 and provides the intro setup this item asks for. Two items, one structural move. See Item 28 for the quote itself.
Item 24 — Chapter 9 — Russian Doll Therapy positioning
Category: STRUCTURAL + LIKE Phase: V2
Observation: Chapter 9 does roll it up nicely — Assagioli, integration of the parts. I just think a little bit at the beginning to clarify the Russian doll thing. I found it very emotional listening to it. Maybe just a little more intro to set the scene that she sat with these dolls.
Decision / Note: Combines with Item 23. The Chapter 9 opening needs more setup before the emotional weight of Olivia lands.
Item 25 — Whole book — affirmation and chapter-scaffolding question
Category: LIKE + STRUCTURAL PRINCIPLE Phase: V2
Observation: This book is bloody compelling. It's absolutely marvellous. I can't put it down. I do wonder whether each chapter should have a quote at the beginning and some Pearl of Wisdom at the end, a tip, a hint. I do like quotes but I don't know if that's just too traditionalist.
Decision / Note: RESOLVED 21 May 2026. YES to quote at start of each chapter (confirmed via Item 37 platinum addition). NO to formal tips at chapter close — prescriptive register conflicts with the book's voice. YES to insights at chapter close — observational and earned by the chapter's argument. Distinction is Paul's: an insight lands; a tip instructs. This establishes the chapter-end pattern across the whole manuscript.
Item 26 — Chapter 9 — close (sub-personalities and AI memory link)
Category: AMPLIFY Phase: V2
Observation: Sub-personalities are a way of articulating that we need different versions of ourselves talking to the agent. It's worth finishing that section by explaining that it's part of memory. If we're in one style and stay in it, we'll get responses in that style. It's about that self-awareness. You've got to be super self-aware to get to stage four use of the system.
Decision / Note: Add the sub-personalities to AI memory link. Connect to stage 4 maturity from Chapter 8.
Item 27 — Chapter 11 — page 61 (WaaS, Pret story)
Category: AMPLIFY Phase: V2
Observation: We've just started in WaaS, page 61. Today I went to Pret. The Pret Club, they no longer do it. I contacted Pret via Facebook messenger. I typed back: thanks a million, by the way are you human or AI? They said 'human of course' and I paused. May 2026 — I'm setting the clock and wondering how long before Pret has got an AI as their front door. I also have a family member in the software industry going AI-first.
Decision / Note: PLATINUM. The Pret story as present-tense moment of the WaaS transition. Sets May 2026 as personal clock. Significant new material for Chapter 11.
Item 28 — Chapter 9 opening — the 15 November 2006 recognition moment
Category: AMPLIFY Phase: V2
Observation: The exact wording I wrote on the 15th of November 2006: 'I didn't understand anything yet. I just recognised that there was something to understand.' That was me on a management training course in 2006, the day I read Behind the Masks. I think that's where we are now in AI. There's a recognition that it's different once you realise there's something to understand. I think society is having that same moment now. That quote has got to find its way into my book somewhere.
Decision / Note: PLATINUM. RESOLVED 21 May 2026. Placement: Chapter 9 OPENING. The 2006 recognition quote opens Chapter 9 and provides the intro setup Item 23 calls for. Two items resolved as one structural move. The recognition-before-understanding quote naturally sets up sub-personalities work — the move from AI focus to self-knowledge focus.
Item 29 — Chapter 11 — ending (Shipston / Plugs at chapter close)
Category: STRUCTURAL + REPETITION Phase: V2
Observation: At the end of Chapter 11 you revisit the Plugs thing. We've done that to death. The From Boardroom to Shipston bit needs to just vanish. Chapter ends well with 'The future is not AI or human; it is AI at the surface and human at depth.' Skip the Shipston bit and fast forward to 'different model, different world, same gap.'
Decision / Note: Cut. Remove the Shipston / Plugs closing of Chapter 11.
Item 30 — Chapter 12 — cancer passage at chapter opening
Category: REPETITION + STRUCTURAL Phase: V2
Observation: Chapter 12. You revisit the cancer story — almost an identical copy of prior writings. We can thin that down. Keep: 'The treatment was significant, the aftermath was more significant still.' 'What matters becomes clearer. What doesn't matter becomes obvious.' That's the essence of the gap in those few words.
Decision / Note: Thin the cancer passage at Chapter 12 opening. Keep the 'what matters becomes clearer' spine. Full cancer narrative belongs in How to Behave. Coordinate with Item 31.
Item 31 — Chapter 12 — close (Bowlby quote / cancer loop)
Category: REPETITION + STRUCTURAL Phase: V2
Observation: The Bowlby quote at the end of Chapter 12 can be ended after the John Bowlby quote. You end up repeating it and go on to talk about cancer again so it feels like a loop. End of Chapter 12 needs killing off after the Bowlby quote.
Decision / Note: Direct excision. Chapter 12 ends at the Bowlby quote. Overrules Editorial Handover v2 statement that this echo is deliberate. Paul's listening pass identifies it as a loop. Handover will be updated.
Item 32 — Manuscript complete marker (deliberate read-through proof)
Category: MARKER Phase: RETIRED
Observation: Paul explicitly noted this to show he had read every word.
Decision / Note: No editorial action. Marker preserved as proof of complete read.
Item 33 — Mind the Gap Final Assessment — inclusion at rear
Category: PHASE 2 Phase: PHASE 2
Observation: The addendum document 'Mind the Gap Final Assessment' is written by you in your first hand. It needs to be added to the book at the back. It needs to end with 'This is quite an interesting story. Someone will be interested. It's the best it can be for the present.' That is all Bowlby asked. It is enough.
Decision / Note: Phase 2 — rear matter. Final Assessment to be included with Paul's proposed closing wording. Not v2 work.
Item 34 — Day in the Life and 72 million words
Category: PHASE 2 + AMPLIFY Phase: PHASE 2 + V2
Observation: 'The Day in the Life of the Doc' needs integrating because it describes how we wrote it. There's also something I've written about the 72 million words. It's not 30 years, it's 72 million words.
Decision / Note: RESOLVED 21 May 2026. CLEAN SPLIT. (a) 'A Day in the Life' — PHASE 2 rear matter (not front). Confirmed: belongs at the back, after the book has been read. (b) 72 million words framing — V2 work, the resolution of Item 4. Use as reframing of 'thirty years' where appropriate.
Item 35 — Front and rear matter — full assembly
Category: PHASE 2 Phase: PHASE 2
Observation: There are two documents — the acknowledgement paper and the front matters paper — and they all need integrating along with the bibliography, the sign-off, and the dedication.
Decision / Note: Phase 2 — complete front and rear matter assembly after v2 is stable. Includes Naula full name verification (still pending).
Item 36 — Throughout — 'the gap' definition drift
Category: DRIFT Phase: V2
Observation: The gap is between AI behaviour and human behaviour. That's The Gap. By Chapter 12 there is drift toward AI versus EI. We have drift on AI to EI which we've both spotted and the need to double check that we use Additional Intelligence and Artificial Intelligence in their correct places and contexts.
Decision / Note: RED + GOLD. The gap is behavioural. AI behaviour versus human behaviour. Audit the manuscript for: (a) inappropriate intelligence-versus-intelligence framing (especially Chapter 12 'intelligence of the gap' passage); (b) every Additional Intelligence reference in correct context; (c) every Artificial Intelligence reference in correct context. Pairs with Item 1. Resolved by sustained discipline during chapter-by-chapter work plus targeted searches.
Item 37 — Every chapter — the three-view chapter spine test
Category: STRUCTURAL PRINCIPLE Phase: V2
Observation: The gap is between AI behaviour and human behaviour. Each chapter needs to have an AI and human context to it as well as something lived — whether the chapter is AI or human orientated/led/biased it has to have the other two views. That's the real helicopter pass. Platinum would be to add a quote to each chapter.
Decision / Note: PLATINUM. Structural principle applied to every chapter. Three views required per chapter: AI view, human view, lived view. Audit binary: does this chapter carry all three? Platinum addition: quote at start of each chapter. UPDATED 21 May 2026 (per Item 25 resolution): chapter bookend pattern is OPENING QUOTE + CLOSING INSIGHT. Insights, not tips. An insight is observational and earned by the chapter's argument; a tip is prescriptive and tilts toward self-help register. This is the helicopter resolution mechanism for Item 36. Run the audit on every chapter before per-chapter editorial work begins.
Item 38 — Chapter 1 (bio) and AI sign-off — LinkedIn author characterisation alignment (NEW)
Category: AMPLIFY Phase: V2
Observation: This is how I characterise myself in LinkedIn. And how it should be presented: consulting rooms, boardrooms, leadership teams, therapy rooms, hospital lecture halls, my own cancer journey, end-of-life care services.
Decision / Note: RESOLVED 21 May 2026. Option C — update author bio (line 199) and AI sign-off (line 1903) to the LinkedIn characterisation. Chapter 7 four-rooms framework (factory floor, boardroom, therapy room, AI room) UNCHANGED — those rooms are structurally load-bearing argument for the chapter, not biography. Supersedes the partial scope of Item 3 (cancer wards). Coordinates with Item 4 — the bio rewrite is also the natural moment to apply the 40-year chronology framing per Claim A. Three items resolve in one coordinated bio pass: Item 3 (superseded), Item 4 (40-year framing at lines 199 and 1903), Item 38 (LinkedIn characterisation).
Item 39 — Front matter and Chapter 1 — the defining sentence (NEW)
Category: AMPLIFY Phase: V2
Observation: This is the gap, isn't it? The gap between artificial intelligence behaviour and human behaviour. Not the gap between AI and EI.
Decision / Note: PLATINUM. Paul's articulation 21–22 May 2026. The defining sentence the book has been circling. Candidate placement: front matter (as the actual statement of what the book is about, sitting between Author's Note and Chapter 1) OR Chapter 1 opening as the definitional landing. Two-part structure: (1) AFFIRMATIVE — 'The gap between artificial intelligence behaviour and human behaviour.' (2) NEGATIVE — 'Not the gap between AI and EI.' The negation is part of the definition; it prevents the most common misreading and disambiguates the book from what the cover signals. Pairs structurally with Item 1 (define the gap at first use) and Item 36 (the behaviour-versus-intelligence drift audit). Items 1, 36, 39 form a structural triangle: 39 is the affirmative statement; 36 is the negative audit; 1 is the discipline of using both consistently from page one. Resolves the definition-of-the-gap question definitively. Paul to decide placement during v2 chapter work. NOTE 23 May 2026: Item 39 now sits at the top of the new opening section established by Item 40.
Item 40 — Front matter — The Spine (NEW OPENING SECTION)
Category: STRUCTURAL ADDITION Phase: V2
Observation: I read the book, provided 7000 words of content critique and also noticed in the later chapters of the book that 'the gap' had drifted — it was like the AI had begun inventing the gap — so I asked myself if the gap was clear. Then the instrument took over and I ended up telling you (no 39) but in parallel with that I decided to decompose and rethink — 600 words later I found the gap in the book and filled it. This book seeks to help readers to understand there is a gap between the behaviour of AI and humans.
Decision / Note: PLATINUM. Paul's articulation 23 May 2026. The book's thesis statement, expanded from Item 39 into a full opening section. To be placed between the Author's Note and Chapter 1, as a named section titled (provisional) 'The Gap' or 'What This Book Is About'. One to three pages. Item 39's title-level sentence sits at the top. The 600 words below are Paul's verbatim source material — they are the substrate from which the opening section will be shaped. Paul to draft the final opening section in his voice (estimated 400–600 words shaped from this source); Claude provides structural feedback only. PAUL'S VERBATIM SOURCE MATERIAL FOLLOWS, in full, preserved exactly as written 23 May 2026:
[SOURCE MATERIAL BEGINS]
What is the gap
The book is about behaviour — and the gap between two radically different systems that now both produce seemingly similar behaviour
The core distinction
AI does not merely produce answers.
It produces behaviour.
It responds.
It adapts.
It appears to understand.
It mirrors tone.
It anticipates need.
It offers confidence.
It performs fluency.
It occupies conversational space.
That behaviour has a recognisable signature:
fluent, confident, tireless, available, asymmetric, frictionless, and disembodied.
It can sound certain without being certain.
It can sound empathic without feeling anything.
It can sound experienced without having lived.
It can appear present without being physically, emotionally, or socially present.
Human behaviour has a different signature
Human behaviour does not emerge from language prediction alone.
It emerges from a living system.
A human being is not just processing information.
A human being is regulating a body, carrying history, protecting identity, reading threat, managing shame, seeking belonging, remembering pain, anticipating consequence, and trying to remain coherent in front of other humans.
Human behaviour is:
somatic, relational, defensive, adaptive, embodied, socially calibrated, historically loaded, and biologically constrained.
The deeper point
AI behaviour is produced by a sophisticated mathematical model developed, calibrated, encoded and embedded in software and hardware.
Human behaviour is produced by the complex interplay of neurology, biology, psychology, physiology, memory, attachment, social conditioning, identity, fear, desire, status, belonging and mortality — all inside a living creature.
The paradox
The differences are obvious. Almost too obvious.
One is built.
One is born.
One is trained on data.
One is shaped by life.
One predicts.
One experiences.
One outputs.
One suffers consequences.
One can be reset.
One carries residue.
The differences are inarguable.
The similarities, at the surface, often indistinguishable.
The gap between what AI is doing and what humans experience it as doing.
Stronger formulation
The book observes behaviour at the point where two systems meet.
On one side: artificial intelligence, producing behavioural output from mathematical architecture.
On the other: human beings, producing behavioural response from embodied life.
That is the gap.
My book won't explain it any more precisely than that. It will however explain what I've discovered so far.
40 years of human behaviour. 4 of AI.
The book is not primarily about intelligence. It is about behaviour. AI now produces behaviour with a distinctive signature: fluent, confident, available, asymmetric, tireless and disembodied. Human beings produce behaviour with an entirely different signature: somatic, relational, defensive, socially calibrated, shaped by lived experience and embedded in physical presence. One is generated by a sophisticated mathematical model operating through software and hardware. The other emerges from the unstable, living interplay of neurology, biology, psychology, physiology, memory, identity and social threat. The differences are obvious. The similarities are almost unrecognisable — until we notice that human beings respond to AI behaviour through the same social brain they use to read other people. That is the gap: not between AI and human intelligence, but between artificial behaviour and human behaviour — and what happens when one begins to resemble the other just enough to matter.
Title-level sentence
Between artificial intelligence behaviour and human behaviour sits a gap.
[SOURCE MATERIAL ENDS]
COORDINATED INTEGRATION WORK: Item 40 triggers light passes in three chapters. Chapter 1 — open with the spine rather than build toward it. Chapter 7 — frame the four rooms as DEMONSTRATING the gap, not introducing it. Chapter 12 — land the spine as final argument (coordinates with Item 36 audit and Item 31 Bowlby loop cut). NO OTHER CHAPTERS require structural change; they illustrate the gap, they do not redefine it. CRITICAL DISCIPLINE: When the spine wording appears in chapter integration, copy verbatim; never paraphrase. The spine wording is the through-line that holds the book together.
Item 41 — Working documents — OOAI definition correction (NEW)
Category: FACTUAL CORRECTION Phase: ADMINISTRATIVE
Observation: OOAI is outcome-oriented AI — known as slash goal in OpenAI/ChatGPT. I use the term to mean you enter a desired outcome instead of tasks — get to here, achieve this.
Decision / Note: RED. Working-document correction. The compaction summary at the start of the 23 May 2026 thread carried a hallucinated definition of OOAI as 'Out Of AI — the discipline of recognising when the AI conversation cannot be the source of verification'. That definition is wrong. OOAI = Outcome-Oriented AI, Paul's coined term, equivalent to OpenAI's /goal feature. THE MANUSCRIPT IS ALREADY CORRECT — lines 1759 and 1777 define OOAI accurately as Outcome-Oriented AI. THE CORRECTION IS FOR: (a) future Claude instances reading the working documents; (b) any artefacts produced this session that may carry the wrong definition (Calibration Brief, Methodology Paper, etc — audit pending); (c) the working memory of all future threads. This item resolves PC v5 Item 22 (which flagged outcome-oriented prompting as Paul's coined term but did not nail the OOAI definition explicitly). Discovered and corrected 23 May 2026 by Paul reading Claude's reference to OOAI elsewhere and noticing it did not match his meaning.
Item 42 — Manuscript and supplementary documents — Attribution discipline (NEW)
Category: STRUCTURAL CORRECTION Phase: V2
Observation: Sign-off line. Each chapter ends with the same two-line attribution: 'Ideas and direction: Paul Roebuck. Words: Claude AI (Anthropic) — claude-sonnet-4-6 | May 2026.' This is a faulty attribution. Claude should use the most accurate, transparent and relevant attributions from now on. I was wrong to instil this poor practice. Claude decides attributions. Don't think they are needed per chapter anyhow.
Decision / Note: Paul's instruction 23 May 2026. Reverses the per-chapter attribution template that was previously in place. REMOVE attribution footers from: (1) all 12 chapters of the manuscript; (2) the Complete Manuscript Rev2; (3) supplementary book documents (Bibliography, Academic Addendum, Academic Chronology, Front Matter, blog posts, illustrated index, etc); (4) the Calibration Brief and Methodology Paper produced this session. REPLACE WITH: a single honest paragraph in the Author's Note acknowledging AI involvement plainly and accurately, without per-chapter repetition or model-version specificity that obscures the actual distributed reality of how the book was produced. RETAIN attribution footers in: continuity / working editorial documents (Editorial Handover, Planned Changes, Editorial Breadcrumbs Log, Artefact Library, Journal Artefact) where AI involvement is part of the working record. The AI sign-off chapter at the close of the manuscript STAYS — it is a structural device, not an attribution line. RATIONALE: per-chapter attribution overstates Claude's role at the wrong level of resolution, embeds a potentially stale model identifier into the book, and weakens the book's clinical authority by inviting the reader to question the voice at every chapter close. Coordinates with the existing attribution rule in Paul's user preferences — this item reaffirms and operationalises that rule by removing the contradicting per-chapter template.
The Three-View Chapter Spine Test (Item 37) — Per-Chapter Audit
Run this audit before working each chapter. Confirms three views are present, identifies missing dimensions, surfaces quote opportunities.
Chapter 1 — Human Recognition. Lead: AI/human parallel. AI view: Plugs. Human view: social brain. Lived: swearing at the chatbot. Quote: TBC.
Chapter 2 — The Interpretation Trap. Lead: human mechanism. AI view: five misreadings. Human view: transference. Lived: therapy room examples. Quote: TBC.
Chapter 3 — Hidden Architecture. Lead: AI. AI view: seven layers. Human view: the room is never just two people. Lived: training as a therapist. Quote: TBC.
Chapter 4 — Drift. Lead: AI mechanism. AI view: Crewe-Manchester. Human view: clinical drift in long-term therapy. Lived: Nairobi slide; ombudsman case. Quote: TBC.
Chapter 5 — Compression. Lead: AI mechanism. AI view: context window. Human view: human memory reconstruction. Lived: tunnel photograph, liminal moment. Quote: TBC.
Chapter 6 — Fluency, Confidence, and the Defended Gap (rewritten). Lead: jointly led. AI view: fluency, compression, hallucination as directional defence. Human view: NGE-FOF directional regulation (citing Where Does The Shame Go?). Lived: fluency interrupt (Item 14); thirty years asking the question. Quote: TBC. NOTE: This is the most substantial single chapter change in v2.
Chapter 7 — The Gap in Practice. Lead: lived. AI view: AI room. Human view: factory floor, boardroom, therapy room. Lived: career chronology. Quote: CONFIRMED 21 May 2026 — “All behaviour is learned behaviour. Not all behaviour is taught.” (Roebuck) per EC-06.
Chapter 8 — Same Language or Different Game? Lead: AI engagement. AI view: outcome-oriented prompting (Paul's term — see Item 22). Human view: leadership and standard-setting. Lived: K3, Tarmac. Quote: TBC.
Chapter 9 — Sub-Personalities for AI Fluency. Lead: human method applied to AI. AI view: six modes. Human view: psychosynthesis, sub-personalities. Lived: Olivia. Quote: CONFIRMED 21 May 2026 — “I didn’t understand anything yet. I just recognised that there was something to understand.” (Roebuck, 15 November 2006) per Item 28. The quote opens the chapter and provides the intro setup Item 23 calls for.
Chapter 10 — The Dialect. Lead: AI engagement. AI view: dialect, brief structure. Human view: conversational versus structured speech. Lived: Fuse Facebook post. Quote: TBC.
Chapter 11 — WaaS. Lead: organisational. AI view: agentic workforce. Human view: labour shift. Lived: Pret story (Item 27). Quote: TBC.
Chapter 12 — Mind the Gap. Lead: lived synthesis. AI view: must be brought back to behavioural framing (Item 36). Human view: Two Dads, primal wound. Lived: cancer. Quote: TBC — Bowlby epigraph already present.
Re-run this audit after each chapter is reworked to confirm three views remain in balance.
Audit Tools for the v2 Pass
Targeted text searches to verify Item 36 resolution and prevent drift recurring:
— Search 'the gap' — confirm consistent behavioural framing throughout.
— Search 'Additional Intelligence' — confirm each use in correct context (Paul's coined term for the working partnership, not synonym for AI).
— Search 'artificial intelligence' — confirm each use in correct context.
— Search 'AI' as standalone token — spot places where the term is being stretched.
— Search 'intelligence' broadly — find passages drifted into intelligence-versus-intelligence framing.
— Search 'emotional intelligence' and 'EI' — confirm only used where appropriate.
— Search 'thirty years' — audit all chronology references against Chapter 7 master version.
— Search 'Outcomes' — audit for Anthropic feature drift (Item 22).
— Search personal names — Kieren, [name held], Mr [name held], Naula — hallucination targets.
Re-run after each chapter is reworked. Cheap. Effective. Verification layer.
Summary by Phase
V2 — Manuscript v2 editorial work: 30 items inside v2 scope. Includes Items 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 14, 15, 16 (RESOLVED), 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 34 (partial), 36, 37.
RESOLVED — Item 16 (directional defence integration into Chapter 6). Decision made; execution during chapter work.
PHASE 2 — Items 33, 34 (partial), 35. Rear matter and front matter assembly after v2 is stable.
POST-V2 — None outright. The wider Shame-Response Continuum applications (Hunter-Farmer, APC, ADHD calibration) explicitly held out of scope for Mind the Gap.
RETIRED — Items 13 (skipped as unlucky) and 32 (read-through proof).
RED Flags — must be resolved correctly
— Item 4: 'thirty years' approximation throughout. Use Chapter 7 chronology as master. Consider 72 million words framing (Item 34).
— Item 17: age 16 not 18 at Blakeborough 1978.
— Item 21: infinite context window stated as announced; actually aspirational.
— Item 22: 'Outcomes' attributed to Anthropic; actually OpenAI /Goal. Outcome-oriented prompting is Paul's term.
— Item 31: Chapter 12 cancer loop after Bowlby — cut.
— Item 36: behaviour-versus-behaviour discipline must hold through Chapter 12 and AI sign-off.
PLATINUM items — significant additions
— Item 16 (RESOLVED): directional defence integration. Chapter 6 rewritten with Defending the Gap as spine. Citation to Where Does The Shame Go?
— Item 27: Pret story and enterprise software speculation for Chapter 11.
— Item 28: the 15 November 2006 recognition moment.
— Item 37: three-view chapter spine test as structural discipline plus quote per chapter.
— Item 39: the defining sentence — 'The gap between artificial intelligence behaviour and human behaviour. Not the gap between AI and EI.' Placement decision pending.
— Item 40 (NEW): The Spine — new opening section before Chapter 1 carrying Paul's 600-word verbatim source material. The book's thesis statement made explicit.
Cross-connections to resolve together
— Items 4, 20, 34: the chronology knot — RESOLVED with locked numbers (40/37/20/10). Chapter 7 chronology = master.
— Items 3, 4, 38: bio rewrite coordination — single Chapter 1 bio pass handles all three (cancer wards superseded by LinkedIn characterisation; 40-year chronology applied at lines 199 and 1903).
— Items 19, 27, 29: Plugs through-line across chapters. Three references with arc; cut the Chapter 11 ending.
— Items 23, 24: Chapter 9 opening. Combined treatment.
— Items 30, 31: Chapter 12 cancer / Bowlby loop. Thin opening; cut closing duplication.
— Items 1, 36, 39, 40: THE GAP DEFINITION CONSTELLATION (was triangle, now four-point). Item 40 is the new opening section carrying the spine material. Item 39 is the title-level sentence at the top of that section. Item 36 audits the negation (no drift to intelligence-vs-intelligence framing). Item 1 disciplines consistent use throughout. Resolve as a coordinated front-matter and Chapter 1 move.
— Items 36 and 37: the structural pair. Item 36 names the drift; Item 37 prevents it recurring. Audit before per-chapter work.
— Items 40, 1, 7, 12: integration passes. The spine in Item 40 triggers light passes in Chapter 1 (open with spine), Chapter 7 (frame four rooms as DEMONSTRATING the gap), Chapter 12 (land the spine as final argument). No other chapters require structural change.
Framework citation — confirmed scope
Mind the Gap cites the Shame-Response Continuum framework at one place only: in the rewritten Chapter 6, with reference to:
— Where Does The Shame Go? — Towards a Theory of Directional Shame Regulation and Anti-Disintegration Adaptation (Roebuck, 2026, working theory paper). To be available at paulroebuck.co.uk ahead of book launch.
The book does NOT include or reference: the Hunter-Farmer commercial application, the Adaptive Pressure Calibration positioning paper, the Pegs/Holes/Hammers image, the ADHD/neurodivergence calibration work, or the wider Shame-Response Continuum body of work. These belong to other publications. The framework appears in Mind the Gap as it applies to AI behavioural parallels, scoped tightly to Chapter 6.
Editorial Handover impact
Two corrections required when handover is next reissued (v3):
— Item 31: the Bowlby epigraph echo at Chapter 12 close is NOT deliberate. It is a loop. Cut.
— Item 36: the behaviour-to-intelligence drift should be added as a named drift risk.
— Item 37: the three-view chapter spine test should be added as a structural discipline.
— Item 16 framework citation guidance: Where Does The Shame Go? as the only framework citation; wider Shame-Response Continuum scope held out.
Decisions captured 23 May 2026 (the watershed)
Three further decisions logged in this v6 update. These are the structural watershed of the editorial process — the moment the book's spine was located, named, and integrated.
— Item 40 ADDED 23 May 2026 — PLATINUM. The Spine. New opening section before Chapter 1 carrying Paul's 600-word verbatim source material. The book's thesis statement made explicit. Triggers light integration passes in Chapters 1, 7, and 12. No other chapters affected.
— Item 41 ADDED 23 May 2026 — RED. OOAI definition correction. Caught by Paul reading Claude's reference to OOAI in another context and noticing it did not match his meaning. OOAI = Outcome-Oriented AI (Paul's coined term, equivalent to OpenAI /goal). Not 'Out Of AI' — that was a compaction-summary hallucination. The manuscript is already correct (lines 1759, 1777). Working documents and future Claude instances must use the correct definition.
— Item 42 ADDED 23 May 2026 — Attribution discipline. Per-chapter sign-off line removed from all chapters and supplementary book documents. Replaced by single honest paragraph in Author's Note. Continuity / working editorial documents retain attribution footers as part of the working record.
— Gap Definition Triangle (Items 1, 36, 39) expanded to GAP DEFINITION CONSTELLATION (Items 1, 36, 39, 40). Item 40 carries the substrate; Item 39 carries the title-level sentence; Item 36 audits the negation; Item 1 disciplines first-use placement.
Decisions captured 21–22 May 2026 (later additions)
Three further decisions logged in this v5 update:
— Item 4 RESOLVED 21 May 2026 — chronology numbers locked. 40 years professional, 37 years commercial (line 1111 only), 20 years behaviour, 10 years clinical psychotherapy. Per-line assignments mapped by Claim A/B/C.
— Item 38 ADDED 21 May 2026 — LinkedIn author characterisation alignment for Chapter 1 bio and AI sign-off. Chapter 7 four rooms UNCHANGED. Supersedes Item 3 scope.
— Item 39 ADDED 21–22 May 2026 — the defining sentence: 'The gap between artificial intelligence behaviour and human behaviour. Not the gap between AI and EI.' Forms structural triangle with Items 1 and 36. Placement decision pending (front matter or Chapter 1 opening).
— Item 3 PARTIALLY SUPERSEDED — by Item 38 (LinkedIn characterisation expanded the cancer-wards scope).
Decisions captured 21 May 2026 (Category A and partial Category B)
Seven Category A decisions plus two Category B decisions logged in the v4 update:
— Item 15 RESOLVED — YES, include fact/supposition/implication categorisation in Chapter 6.
— Item 25 RESOLVED — YES to opening quote per chapter; NO to formal tips; YES to closing insights. Insights, not tips.
— Item 28 RESOLVED — placement is Chapter 9 opening. Merges with Item 23 as one structural move.
— Item 23 RESOLVED — merged with Item 28.
— Item 8 RESOLVED — YES include, as a photo collage of the actual drift in real time (not single slide).
— Item 34 RESOLVED with clean split — 'A Day in the Life' to rear matter (Phase 2); 72 million words framing to V2.
— EC-06 RESOLVED — 'All behaviour is learned behaviour. Not all behaviour is taught.' becomes Chapter 7 opening epigraph.
— EC-09 RESOLVED — replace 'argument' with 'claim', 'observation', or 'pattern' in opening sections.
— PUB-02 RESOLVED — Paul Roebuck is the publisher. No imprint. Unblocks PUB-01 and PUB-03.
— EC-04 STATUS CHANGED — from passive blocking to active permission request. Bowlby estate contact research and email drafting in progress.
— Item 37 UPDATED — chapter bookend pattern now reads: opening quote + closing insight. Adds insights-not-tips discipline.
Items still pending Category B resolution: B1 (skin graft sentence, EC-02), B2 (separate clients dedication, ACK-03), B4 (medium-as-dialect Chapter 10, EC-10), B5 (landing page sequencing, MKT-04), B6 (How to Behave sequencing, HTB-01).
Planned Changes v6. Compiled 23 May 2026.
42 items. 35 from Paul Roebuck’s spoken listening pass plus 7 items added structurally (Item 36 behaviour drift, Item 37 spine test, Item 38 LinkedIn characterisation, Item 39 defining sentence, Item 40 The Spine, Item 41 OOAI correction, Item 42 attribution discipline). 40 active editorial items. 2 markers retired (13, 32). v6 adds three items captured on 23 May 2026 — the watershed. Item 40 carries Paul’s 600-word verbatim source material for the new opening section. Items 1, 36, 39, 40 now form the GAP DEFINITION CONSTELLATION. Voice preserved throughout.
Ideas and direction: Paul Roebuck
Words: Claude AI (Anthropic) — claude-opus-4-7 | May 2026