# Published vs. Draft Gap Analysis **Purpose:** This document identifies what is PRESENT in Nick Young's published papers but ABSENT or UNDERDEVELOPED in his drafts. It exists to inform the design of a quickfire-questions skill that helps bridge the outline-to-prose gap. **Published papers analysed (7):** Growing the Image (2025), Sounds as Properties (2021), Hearing Spaces (2017), Agents of Change (2022), Hearing Objects and Events (2018), Audition and Composite Sensory Individuals (2023), Can Movement Be Depicted? (2018) **Drafts analysed (5):** Generating Philosophy with AI, Generative Aesthetics of AI, Virtual Reality and the Structure of Perception, Auditory Images (Audible Presence), Life in the Chat --- ## Level 2: Section-Level Organisation ### What published papers do Each section in a published paper has a single clear function: - **Survey sections** map the landscape of existing views (Agents of Change §2, Hearing Objects §1, Sounds as Properties §2) - **Critical sections** consider and reject one specific alternative (Hearing Spaces §§2–4, each rejecting one account) - **Model-building sections** reconstruct someone else's view as a template for the author's own (Agents of Change §4 on Hoerl, §5 on Horgan) - **Positive-proposal sections** introduce the author's own view (Hearing Objects §5, Hearing Spaces §5, Sounds as Properties §4) - **Application sections** show the positive view working on cases (Growing the Image §IV, Audition and Composite §3) Crucially, sections never try to do two things at once. A survey section surveys; a critical section criticises; a proposal section proposes. The division of labour is clean. ### What drafts do and don't do **Generating Philosophy** follows this pattern well in its developed sections. Sections 1–3 each have clear functions (present the problem, develop the knowledge-transforming framework, apply to AI writing). But Sections 4 and 5 are burden statements rather than executed sections: they say what the section needs to do without doing it. Section 4 reads: > Section 4 can make the positive argument... The norms of philosophical practice are publicly codifiable and textually manifest, and the philosophical corpus is itself the record of an evaluative feedback loop that a model can learn from. This is a planning sentence, not philosophical prose. **VR Paper** has sections that blur their functions. Section 4 (Virtual Reality and the Structure of Perception) tries simultaneously to: (a) establish the proximity principle, (b) enumerate VR's manipulation capabilities, (c) draw parallels with earlier technologies, and (d) argue for VR's uniqueness as an artistic medium. In a published paper, these would be separated. The result is repetition — the proximity point is stated three times across subsections. **Auditory Images** has clean sections through Section 3, then the organisation breaks down. Section 4 contains two overlapping versions of the headphone-listening material (one in prose, one in bullets). Section 5 exists as two separate attempts at the same section, neither complete. ### The gaps **GAP 2.1 — Sections that state their burden instead of discharging it.** Published sections DO the argumentative work. Draft sections sometimes describe what they need to do. The skill should ask: "What specific argumentative work does this section need to accomplish? Can you state the conclusion this section reaches?" **GAP 2.2 — Sections with blurred functions.** Published sections have one job each. Draft sections sometimes try to survey AND criticise AND propose in one section. The skill should ask: "What is this section's single job? Is it surveying, criticising, or proposing?" **GAP 2.3 — Missing intermediate sections.** Published papers often have a model-building section between the critical work and the positive proposal — a section that reconstructs someone else's argument as a template. (Agents of Change §§4–5 reconstructs Hoerl and Horgan before building the original argument in §§6–7.) Drafts tend to jump from criticism to positive proposal without this intermediary. **GAP 2.4 — Duplicate/overlapping sections.** Published papers never have two versions of the same section. Auditory Images has two Section 5 attempts. This indicates the author doesn't yet know which version of the argument to commit to. --- ## Level 3: Move-Level Analysis ### The move vocabulary of published papers Across 7 published papers, the following argumentative moves recur: | Move | Description | Frequency | |------|-------------|-----------| | **SURVEY** | Map existing views on a question | Every paper | | **DISTINGUISH** | Draw a conceptual distinction | Every paper | | **REJECT** | Show why an alternative fails | Every paper (2–4 per paper) | | **ILLUSTRATE** | Provide a vivid concrete example | Every paper | | **ANALOGIZE** | Build structural parallels between domains | 6 of 7 papers | | **PROPOSE** | State the positive view | Every paper | | **DERIVE** | Show consequences of the positive view | 6 of 7 papers | | **ANTICIPATE OBJECTION** | State and respond to a likely criticism | 6 of 7 papers | | **CONCEDE** | Grant a point to the opponent | 5 of 7 papers | | **DEFINE** | Provide precise conditions for a concept | 5 of 7 papers | | **DIAGNOSE** | Identify a hidden assumption | 4 of 7 papers | | **APPEAL TO PHENOMENOLOGY** | Invoke what experience is like | 4 of 7 papers | | **DECOUPLE** | Separate two questions standardly treated together | 3 of 7 papers | | **MODEL-BUILD** | Reconstruct someone's argument as a template | 2 of 7 papers | The signature moves — the ones most distinctive to this author — are ILLUSTRATE and ANALOGIZE. Every published paper builds its positive view through analogy: audition-to-vision, temporal-to-spatial, gardening-to-AI, Hoerl-to-the-author's-own-argument. And every paper grounds abstract claims in vivid, often playful examples: bowling balls rolling, Be My Baby's reverb, ice cream melting, wine pouring. ### What drafts do and don't do **Moves present in drafts:** - SURVEY: Present in Generating Philosophy (§1 on Floridi and Zahavy), Generative Aesthetics (§§3–4), Auditory Images (§§1–2) - DISTINGUISH: Present in all developed drafts - ILLUSTRATE: Present but less vivid — the drafts tend toward abstract examples rather than the concrete, playful ones in published papers - PROPOSE: Present where the positive section exists (Generative Aesthetics §§5–6, VR §§3–5) **Moves absent or underdeveloped in drafts:** **GAP 3.1 — Missing REJECT moves.** Published papers typically have 2–4 well-developed rejections. The VR paper has zero. Generating Philosophy develops rejections of Floridi and Zahavy but in compressed form — the counterarguments are sketched rather than pressed to their conclusion. Auditory Images has good rejections in §§1–2 but the later material lacks them. **GAP 3.2 — Missing ANTICIPATE OBJECTION moves.** Published papers routinely ask "but what about...?" and then answer. This is almost entirely absent from the drafts. Generating Philosophy has "%%this sentence is twatty%%" and "%%don't like this paragraph at all%%" — the author's internal critic is active but hasn't been converted into explicit objection-response pairs. The Generative Aesthetics draft is an exception — it handles the "metaphor objection" to semiotic forces talk. **GAP 3.3 — Missing DERIVE moves.** Published papers state their positive view and then derive consequences from it: "If this is right, then..." The VR paper states that VR can alter perceptual structure but doesn't derive what follows from this for aesthetics, for the philosophy of perception, or for understanding VR art specifically. What are the consequences of the thesis? **GAP 3.4 — Missing CONCEDE moves.** Published papers regularly grant points to opponents: "Even if we accept Leddington's reasons..." (Sounds as Properties), "Even if belief in a changing present is less common than we might have thought..." (Agents of Change). Drafts tend to assert rather than negotiate. The skill should ask: "What's the strongest version of the opposing view? What can you grant to it?" **GAP 3.5 — Underdeveloped ANALOGIZE moves.** The published papers' most distinctive move — building positive views through analogy — is present but less developed in drafts. The VR paper draws parallels between VR and prism glasses / cochlear implants / hemianopia devices, but the parallels are listed rather than argued for. In published papers, analogies are defended: the author explains WHY the analogy holds, what it illuminates, and where it breaks down. In drafts, analogies are asserted. **GAP 3.6 — Missing DIAGNOSE moves.** Some of the published papers' best moments are when the author identifies a hidden assumption shared by all existing views (Audition and Composite: all views "postulate sensory individuals that are non-composite"). Drafts rarely make this move — they proceed directly to engagement rather than stepping back to diagnose what the entire debate is assuming. --- ## Level 4: Paragraph-Level Patterns ### What published paragraphs do Published paragraphs follow a small set of templates: **Template A — Claim-Development-Consequence:** 1. State a claim or observation (1–2 sentences) 2. Develop with evidence, examples, or quotation (3–6 sentences) 3. Draw a consequence for the paper's argument (1–2 sentences) Example from Hearing Spaces: The paragraph about Be My Baby (1) observes that reverberation is conspicuous in certain recordings, (2) describes the specific reverberant qualities heard, (3) concludes that this is evidence of space being auditorily represented. **Template B — Present-and-Respond:** 1. Present another philosopher's view, often via block quote (2–4 sentences) 2. Gloss or restate the view in the author's own terms (1–2 sentences) 3. Evaluate: endorse, qualify, or reject (2–4 sentences) Example from Agents of Change: Horgan's "what-it's-like of self as source" is quoted, glossed, and then endorsed as providing the right phenomenological resource. **Template C — Consider-Reject:** 1. Present a possible account or objection (2–3 sentences) 2. Show why it fails (3–5 sentences) 3. Indicate what this means for the available options (1–2 sentences) Example from Hearing Objects: The paragraph considering whether "hearing a rolling without hearing that which rolls" is coherent — presents the idea, identifies the difficulty (raises more questions than it answers), and notes the consequence (the objectless view is unattractive). **Paragraph lengths** in published papers: typically 6–15 sentences, with considerable variety. Survey/expository paragraphs tend longer; critical paragraphs tend shorter. No published paragraph is shorter than 4 sentences except in conclusions. ### What draft paragraphs do **Where drafts have polished prose** (Generating Philosophy §§1–3, Generative Aesthetics throughout, Auditory Images §§1–3, Life in the Chat), paragraphs follow the published templates well. The quality of individual paragraphs is often high — the prose is capable and the patterns are recognisable. **Where drafts break down**, the following problems appear: **GAP 4.1 — Paragraphs that state without developing.** Draft paragraphs sometimes make a claim and then move on without the development step. The VR paper's Section 4 does this repeatedly: a claim about VR's capability is stated, then the next claim begins. In published papers, every claim gets at least 3 sentences of development. Example from VR Paper §4: Claims about field-of-view modification, depth perception alteration, and spatial geometry distortion are each stated in 1–2 sentences. In a published paper, each would get its own fully developed paragraph. **GAP 4.2 — Bullet-point lists where paragraphs should be.** Published papers never use bullet points. Drafts frequently do — especially at the transition from critical to constructive work. The VR paper's §4 uses bullet lists; Auditory Images' §§4–5 collapse into bullets; Generating Philosophy's §§4–5 are outlined in bullets. Bullets indicate the author knows WHAT needs to be said but hasn't yet worked out HOW to say it as connected prose. **GAP 4.3 — Missing consequence sentences.** The published paragraph templates almost always end with a "so what" sentence — a sentence that connects the paragraph's content back to the paper's argument. In drafts, paragraphs often end on the evidence/example rather than drawing out the consequence. The last sentence of a published paragraph typically begins with "This shows that..." or "If this is right..." or "The upshot is..." Draft paragraphs tend to stop before this step. **GAP 4.4 — Missing gloss sentences after quotations.** Published papers always gloss block quotes — restating the quoted passage in the author's own terms before evaluating it. Some draft passages quote and then proceed directly to evaluation, or quote and then move on. The Auditory Images draft has several quotations from Nanay, Matthen, and Vishwanath that appear without gloss. --- ## Level 5: Sentence-Level Craft ### What published sentences do **Hedging and qualification:** Published papers use a precise hedging vocabulary. The author never says "X is the case" when the argument only warrants "X seems to be the case" or "arguably, X." But hedging is targeted — it appears on controversial claims, not on established facts. Common hedging devices: "arguably," "at least," "it seems," "I suggest," "plausibly," "at best." Example: "Even if we accept Leddington's reasons for preferring Property over Identity and Parthood, as I think we should, questions remain about a further detail of his view." (Sounds as Properties) — one hedge ("I think we should"), one assertive frame ("questions remain"). **Colons and semicolons:** The author uses colons to introduce elaborations and semicolons to balance parallel claims. This is a distinctive habit across all papers. Example: "Two dimensions matter. A representation can be more or less accurate, depending on whether the dependence relations it encodes actually obtain; and it can be more or less comprehensive, depending on how much of the relevant network it captures rather than omitting." (Generating Philosophy draft — this is one of the polished sections) **First person:** "I argue," "I suggest," "I want to be careful here" in solo papers. Never "the author argues." In co-authored papers, "we argue," "we suggest." **Signposting sentences:** Published papers use explicit sentences to tell the reader where the argument is going: "The rest of the paper develops that case," "In what follows I will argue that," "Three considerations count against this view." These are structural sentences — they carry no philosophical content but manage the reader's expectations. ### What draft sentences do **GAP 5.1 — Notes-to-self instead of hedges.** Published papers hedge for the reader; drafts hedge for the author. The Generating Philosophy draft uses `%%comments%%` (Obsidian hidden comments) as a private hedging mechanism: "%%this sentence is twatty%%", "%%not a clear description of my position%%", "%%don't like this paragraph at all%%". These identify problems but don't solve them. In the finished version, these comments would need to become either revisions or reader-facing qualifications. **GAP 5.2 — Missing signposting sentences.** Published papers are heavily signposted. The VR paper lacks most inter-section signposting — sections begin with their content rather than announcing what they will do. Auditory Images has good signposting in §§1–3 and loses it in §§4–5. The Generating Philosophy draft has a clear roadmap paragraph but individual sections often lack opening signpost sentences. **GAP 5.3 — Incomplete sentences and placeholders.** Published papers have no incomplete sentences. Several drafts do: - VR Paper §2: "In Section 4, we will argue that" — sentence stops mid-thought - Auditory Images §4: "it seems xxx" — placeholder for an unwritten conclusion - Auditory Images §5: "**The reason why is**" — heading for an argument that doesn't exist - Generating Philosophy §0: "[needs to be linked better]" — note-to-self in brackets **GAP 5.4 — (REF) placeholders.** Published papers have complete references. Drafts frequently use "(REF)" as a placeholder: Generating Philosophy §0 has several, Auditory Images §§4–5 has many ("Nanay (REF)", "Vishwanath (REF)", "Dokic (REF)"), VR Paper §2 has one. These indicate claims that are being made without grounding — the author asserts that someone has argued X but hasn't verified the citation. **GAP 5.5 — Rhetorical questions as placeholders rather than moves.** Published papers use rhetorical questions at structurally important moments to focus the reader's attention: "Can we hear a rolling without hearing that which rolls?" (Hearing Objects). In drafts, questions sometimes appear not as rhetorical devices but as genuine uncertainties the author hasn't resolved: "Irrespective of this possibility, we might still wonder whether..." (Auditory Images §5, trailing off). **GAP 5.6 — Sentence flatness.** In the VR paper especially, sentences tend toward declarative exposition: "VR technology enables users to experience embodiment within virtual bodies that diverge from their own." Published papers vary sentence structure more — conditionals, concessive constructions ("While X, Y"), rhetorical questions, parenthetical observations. The draft sentences convey information but lack the dialectical energy of the published prose. --- ## Summary: The Gaps That Questions Should Target Ordered by structural level, from macro to micro: ### Section Level 5. Sections that state their burden rather than discharging it 6. Sections with blurred functions (trying to do too many things) 7. Missing intermediate/model-building sections 8. Duplicate/overlapping sections (indicating unresolved commitments) ### Move Level 9. Missing REJECT moves (alternatives not considered) 10. Missing ANTICIPATE OBJECTION moves (strongest criticism not addressed) 11. Missing DERIVE moves (consequences of the view not drawn out) 12. Missing CONCEDE moves (no negotiation with opponents) 13. Underdeveloped ANALOGIZE moves (analogies asserted rather than argued) 14. Missing DIAGNOSE moves (hidden assumptions not surfaced) ### Paragraph Level 15. Paragraphs that state without developing 16. Bullet-point lists where paragraphs should be 17. Missing consequence sentences (no "so what" at paragraph end) 18. Missing gloss sentences after quotations ### Sentence Level 19. Notes-to-self instead of reader-facing hedges 20. Missing signposting sentences 21. Incomplete sentences and placeholders 22. (REF) placeholders for unverified citations 23. Rhetorical questions as genuine uncertainties rather than moves 24. Sentence flatness (lack of dialectical variety) --- ## Implications for Skill Design The gaps cluster into three types, each requiring different kinds of questions: ### Type A: Content gaps (the author doesn't yet know what to say) Gaps 2, 7, 8, 9, 10, 14 — These require questions that help the author THINK: "What's the strongest objection to your view?" "What hidden assumption do all the views you're criticising share?" "Which version of §5 are you committing to and why?" ### Type B: Execution gaps (the author knows what to say but hasn't converted it to prose) Gaps 1, 5, 15, 16, 17, 18, 20 — These require questions that help the author WRITE: "Can you state the conclusion of this section in one sentence?" "What does the reader need to know before reading this section?" "What follows from the point you just made?" ### Type C: Polish gaps (the prose exists but lacks finishing) Gaps 4, 12, 13, 19, 21, 22, 23, 24 — These require questions that help the author REFINE: "Where are you hedging for yourself vs. for the reader?" "Can you fill in what (REF) should point to?" "What makes the analogy hold — and where does it break down?" The most productive area for quickfire questions is likely **Type B** — the execution gaps. This is where the author has done the thinking but stalls at converting it to connected prose. The skill's primary job is to ask the micro-decision questions that turn an outline or a burden statement into a paragraph: "What is the first sentence of this section?" "How does this paragraph connect to the previous one?" "What example would make this concrete?" **Type A** questions are also valuable but higher-risk — they may require thinking the author isn't ready to do. The skill should offer these but not force them. **Type C** questions are lowest priority — they apply only to prose that already exists and is close to finished.