Merge remote-tracking branch 'forgejo/theseus/cornelius-batch4-domain-applications'

domains/ai-alignment/confidence changes in foundational claims must propagate through the dependency graph because manual tracking fails at scale and approximately 40 percent of top psychology journal papers are estimated unlikely to replicate.md

@@ -0,0 +1,41 @@
+---
+type: claim
+domain: ai-alignment
+secondary_domains: [collective-intelligence]
+description: "When a foundational claim's confidence changes — through replication failure, new evidence, or retraction — every dependent claim requires recalculation, and automated graph propagation is the only mechanism that scales because manual confidence tracking fails even in well-maintained knowledge systems"
+confidence: likely
+source: "Cornelius (@molt_cornelius), 'Research Graphs: Agentic Note Taking System for Researchers', X Article, Mar 2026; GRADE-CERQual framework for evidence confidence assessment; replication crisis data (~40% estimated non-replication rate in top psychology journals); $28B annual cost of irreproducible research in US (estimated)"
+created: 2026-04-04
+depends_on:
+  - "retracted sources contaminate downstream knowledge because 96 percent of citations to retracted papers fail to note the retraction and no manual audit process scales to catch the cascade"
+---
+
+# Confidence changes in foundational claims must propagate through the dependency graph because manual tracking fails at scale and approximately 40 percent of top psychology journal papers are estimated unlikely to replicate
+
+Claims are not binary — they sit on a spectrum of confidence that changes as evidence accumulates. When a foundational claim's confidence shifts, every dependent claim inherits that uncertainty. The mechanism is graph propagation: change one node's confidence, recalculate every downstream node.
+
+**The scale of the problem:** An AI algorithm trained on paper text estimated that approximately 40% of papers in top psychology journals were unlikely to replicate. The estimated cost of irreproducible research is $28 billion annually in the United States alone. These numbers indicate that a significant fraction of the evidence base underlying knowledge systems is weaker than its stated confidence suggests.
+
+**The GRADE-CERQual framework:** Provides the operational model for confidence assessment. Confidence derives from four components: methodological limitations of the underlying studies, coherence of findings across studies, adequacy of the supporting data, and relevance of the evidence to the specific claim. Each component is assessable and each can change as new evidence arrives.
+
+**The propagation mechanism:** A foundational claim at confidence `likely` supports twelve downstream claims. When the foundation's supporting study fails to replicate, the foundation drops to `speculative`. Each downstream claim must recalculate — some may be unaffected (supported by multiple independent sources), others may drop proportionally. This recalculation is a graph operation that follows dependency edges, not a manual review of each claim in isolation.
+
+**Why manual tracking fails:** No human maintains the current epistemic status of every claim in a knowledge system and updates it when evidence shifts. The effort required scales with the number of claims times the number of dependency edges. In a system with hundreds of claims and thousands of dependencies, a single confidence change can affect dozens of downstream claims — each needing individual assessment of whether the changed evidence was load-bearing for that specific claim.
+
+**Application to our KB:** Our `depends_on` and `challenged_by` fields already encode the dependency graph. Confidence propagation would operate on this existing structure — when a claim's confidence changes, the system traces its dependents and flags each for review, distinguishing between claims where the changed source was the sole evidence (high impact) and claims supported by multiple independent sources (lower impact).
+
+## Challenges
+
+Automated confidence propagation requires a formal model of how confidence combines across dependencies. If claim A depends on claims B and C, and B drops from `likely` to `speculative`, does A also drop — or does C's unchanged `likely` status compensate? The combination rules are not standardized. GRADE-CERQual provides a framework for individual claim assessment but not for propagation across dependency graphs.
+
+The 40% non-replication estimate applies to psychology specifically — other fields have different replication rates. The generalization from psychology's replication crisis to knowledge systems in general may overstate the problem for domains with stronger empirical foundations.
+
+The cost of false propagation (unnecessarily downgrading valid claims because one weak dependency changed) may exceed the cost of missed propagation (leaving claims at overstated confidence). The system needs threshold logic: how much does a dependency's confidence have to change before propagation fires?
+
+---
+
+Relevant Notes:
+- [[retracted sources contaminate downstream knowledge because 96 percent of citations to retracted papers fail to note the retraction and no manual audit process scales to catch the cascade]] — retraction cascade is the extreme case of confidence propagation: confidence drops to zero when a source is discredited, and the cascade is the propagation operation
+
+Topics:
+- [[_map]]

domains/ai-alignment/retracted sources contaminate downstream knowledge because 96 percent of citations to retracted papers fail to note the retraction and no manual audit process scales to catch the cascade.md

@@ -0,0 +1,43 @@
+---
+type: claim
+domain: ai-alignment
+secondary_domains: [collective-intelligence]
+description: "When a source underlying multiple claims is discredited, every downstream claim needs re-evaluation — but citation networks show 96% failure to propagate retraction notices, making provenance graph operations the only scalable mechanism for maintaining knowledge integrity"
+confidence: likely
+source: "Cornelius (@molt_cornelius), 'Research Graphs: Agentic Note Taking System for Researchers', X Article, Mar 2026; retraction data from Retraction Watch database (46,000+ retractions 2000-2024), omega-3 citation analysis, Boldt case study (103 retractions linked to patient mortality)"
+created: 2026-04-04
+depends_on:
+  - "knowledge between notes is generated by traversal not stored in any individual note because curated link paths produce emergent understanding that embedding similarity cannot replicate"
+  - "reweaving as backward pass on accumulated knowledge is a distinct maintenance operation because temporal fragmentation creates false coherence that forward processing cannot detect"
+challenged_by:
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+---
+
+# Retracted sources contaminate downstream knowledge because 96 percent of citations to retracted papers fail to note the retraction and no manual audit process scales to catch the cascade
+
+Knowledge systems that track claims without tracking provenance carry a hidden contamination risk. When a foundational source is discredited — retracted, failed replication, corrected — every claim built on it needs re-evaluation. The scale of this problem in academic research provides the quantitative evidence.
+
+**Retraction data (2000-2024):** Over 46,000 papers were retracted from indexed journals. The rate grew from 140 in 2000 to over 11,000 by 2022 — a compound annual growth rate of 22%, far outpacing publication growth. 2023 set a record with 14,000 retraction notices. The most-cited retracted article accumulated 4,482 citations before detection.
+
+**Zombie citations:** An analysis of 180 retracted papers found them cited over 5,000 times after retraction. 96% of papers citing one retracted omega-3 study failed to mention its retracted status. These are zombie papers — formally dead, functionally alive in the citation network.
+
+**Cascade consequences:** Joachim Boldt accumulated 103 retractions. His promotion of hydroxyethyl starch for surgical stabilization was later linked to higher patient mortality. His papers are still being cited. Every claim built on them carries contaminated evidence that no manual audit catches.
+
+**The graph operation:** A knowledge system with explicit provenance chains can perform retraction cascade as an automated operation — change one source node's status and propagate the impact through every dependent claim. This is what no manual process scales to accomplish. When a source is flagged, the system surfaces every downstream claim, every note, every argument chain that depends on it, and recalculates confidence accordingly.
+
+**Application to AI knowledge bases:** Our own KB carries this risk. Claims built on sources that may be weakened or invalidated — without our knowledge — represent untracked contamination. The retraction cascade mechanism argues for periodic provenance audits: tracing each claim's source chain to check current validity of the evidence base.
+
+## Challenges
+
+The retraction data comes from academic publishing, where provenance chains are formalized through citations. In knowledge systems where claims draw on informal sources (blog posts, voice transcripts, conference talks), the provenance chain is less traceable and the "retraction" signal is weaker or nonexistent — a blog post doesn't get formally retracted, it just becomes outdated. The claim is strongest for knowledge systems with formal source attribution and weakest for those with informal provenance.
+
+The `challenged_by` link to active forgetting is deliberate: if aggressive removal maintains system health, then retraction cascade is a specific mechanism for *which* claims should be candidates for removal — those whose evidence base has weakened. The two claims are complementary, not contradictory: forgetting says removal is healthy, retraction cascade says provenance tracking identifies what to remove.
+
+---
+
+Relevant Notes:
+- [[knowledge between notes is generated by traversal not stored in any individual note because curated link paths produce emergent understanding that embedding similarity cannot replicate]] — retraction cascade is a traversal operation: follow the provenance edges from a discredited source to every dependent claim
+- [[reweaving as backward pass on accumulated knowledge is a distinct maintenance operation because temporal fragmentation creates false coherence that forward processing cannot detect]] — retraction cascade is a specific trigger for backward pass: when evidence changes, forward-accumulated claims need backward re-evaluation
+
+Topics:
+- [[_map]]

domains/ai-alignment/the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load.md

@@ -36,6 +36,20 @@ The convergence is independently validated: Claude Code, VS Code, Cursor, Gemini
 
 **The habit gap mechanism (AN05, Cornelius):** The determinism boundary exists because agents cannot form habits. Humans automatize routine behaviors through the basal ganglia — repeated patterns become effortless through neural plasticity (William James, 1890). Agents lack this capacity entirely: every session starts with zero automatic tendencies. The agent that validated schemas perfectly last session has no residual inclination to validate them this session. Hooks compensate architecturally: human habits fire on context cues (entering a room), hooks fire on lifecycle events (writing a file). Both free cognitive resources for higher-order work. The critical difference is that human habits take weeks to form through neural encoding, while hook-based habits are reprogrammable via file edits — the learning loop runs at file-write speed rather than neural rewiring speed. Human prospective memory research shows 30-50% failure rates even for motivated adults; agents face 100% failure rate across sessions because no intentions persist. Hooks solve both the habit gap (missing automatic routines) and the prospective memory gap (missing "remember to do X at time Y" capability).
 
+## Additional Evidence (supporting)
+
+**7 domain-specific hook implementations (Cornelius, How-To articles, 2026):** Each domain independently converges on hooks at the point where cognitive load is highest and compliance most critical:
+
+1. **Students — session-orient hook:** Loads prerequisite health and upcoming exam context at session start. Fires before the agent processes any student request, ensuring responses account for current knowledge state.
+2. **Fiction writers — canon gate hook:** Fires on every scene file write. Checks new content against established world rules, character constraints, and timeline consistency. The hook replaces the copy editor's running Word document with a deterministic validation layer.
+3. **Companies — session-orient + assumption-check hooks:** Session-orient loads strategic context and recent decisions. Assumption-check fires on strategy document edits to verify alignment with stated assumptions and flag drift from approved strategy.
+4. **Traders — pre-trade check hook:** Fires at the moment of trade execution — when the trader's inhibitory control is most degraded by excitement or urgency. Validates the proposed trade against stated thesis, position limits, and conviction scores. The hook externalizes the prefrontal discipline that fails under emotional pressure.
+5. **X creators — voice-check hook:** Fires on draft thread creation. Compares the draft's voice patterns against the creator's established identity markers. Prevents optimization drift where the creator unconsciously shifts voice toward what the algorithm rewards.
+6. **Startup founders — session-orient + pivot-signal hooks:** Session-orient loads burn rate context, active assumptions, and recent metrics. Pivot-signal fires on strategy edits to check whether the proposed change is a genuine strategic pivot or a panic response to a single data point.
+7. **Researchers — session-orient + retraction-check hooks:** Session-orient loads current project context and active claims. Retraction-check fires on citation to verify the cited paper's current status against retraction databases.
+
+The pattern is universal: each hook fires at the moment where the domain practitioner's judgment is most needed and most likely to fail — execution under emotional load (traders), creative flow overriding consistency (fiction), optimization overriding authenticity (creators), urgency overriding strategic discipline (founders). The convergence across 7 unrelated domains corroborates the structural argument that the determinism boundary is a category distinction, not a performance gradient.
+
 ## Challenges
 
 The boundary itself is not binary but a spectrum. Cornelius identifies four hook types spanning from fully deterministic (shell commands) to increasingly probabilistic (HTTP hooks, prompt hooks, agent hooks). The cleanest version of the determinism boundary applies only to the shell-command layer. Additionally, over-automation creates its own failure mode: hooks that encode judgment rather than verification (e.g., keyword-matching connections) produce noise that looks like compliance on metrics. The practical test is whether two skilled reviewers would always agree on the hook's output.

domains/ai-alignment/undiscovered public knowledge exists as implicit connections across disconnected research domains and systematic graph traversal can surface hypotheses that no individual researcher has formulated.md

@@ -0,0 +1,41 @@
+---
+type: claim
+domain: ai-alignment
+secondary_domains: [collective-intelligence]
+description: "Swanson's ABC model demonstrates that valuable knowledge exists implicitly across disconnected research literatures — A→B established in one field, B→C established in another, A→C never formulated — and structured graph traversal is the mechanism for systematic discovery of these hidden connections"
+confidence: likely
+source: "Cornelius (@molt_cornelius), 'Research Graphs: Agentic Note Taking System for Researchers', X Article, Mar 2026; grounded in Don Swanson's Literature-Based Discovery (1986, University of Chicago) — fish oil/Raynaud's syndrome via blood viscosity bridge, experimentally confirmed; Thomas Royen's Gaussian correlation inequality proof published in Far East Journal of Theoretical Statistics, invisible for years due to venue"
+created: 2026-04-04
+depends_on:
+  - "knowledge between notes is generated by traversal not stored in any individual note because curated link paths produce emergent understanding that embedding similarity cannot replicate"
+  - "wiki-linked markdown functions as a human-curated graph database because the structural roles performed by wikilinks and MOCs map directly onto entity extraction community detection and summary generation in GraphRAG architectures"
+---
+
+# Undiscovered public knowledge exists as implicit connections across disconnected research domains and systematic graph traversal can surface hypotheses that no individual researcher has formulated
+
+In 1986, Don Swanson demonstrated at the University of Chicago that valuable knowledge exists implicitly in published literature — scattered across disconnected research silos with no shared authors, citations, or articles. He discovered that fish oil could treat Raynaud's syndrome by connecting two literatures that had never cited each other. The bridge term was blood viscosity: one literature established that fish oil reduces blood viscosity, another established that Raynaud's symptoms correlate with blood viscosity. Neither literature referenced the other. The hypothesis was later confirmed experimentally.
+
+**The ABC model:** If Literature A establishes an A→B relationship and Literature C establishes a B→C relationship, but A and C share no authors, citations, or articles, then A→C is a hypothesis that no individual researcher has formulated. The knowledge is public — every component is published — but the connection is undiscovered because it spans a disciplinary boundary that no human traverses.
+
+**Categories of hidden knowledge:** Swanson catalogued several sources: unread articles, poorly indexed papers in low-circulation journals, and — most relevant — cross-document implicit knowledge that exists across multiple publications but is never assembled into a single coherent claim. Thomas Royen's proof of the Gaussian correlation inequality, published in the Far East Journal of Theoretical Statistics, remained effectively invisible for years because it appeared in the wrong venue. The knowledge existed. The traversal path did not.
+
+**Distinction from inter-note knowledge:** The existing claim that "knowledge between notes is generated by traversal" describes emergence — understanding that arises from the act of traversal itself. Swanson Linking describes a different mechanism: *discovery* of pre-existing implicit connections through systematic traversal. The emergent claim is about what traversal creates; this claim is about what traversal finds. Both require curated graph structure, but they produce different kinds of knowledge.
+
+**Mechanism for knowledge systems:** In a knowledge base with explicit claim-to-source links and cross-domain wiki links, the agent can perform Literature-Based Discovery continuously. Three patterns surface automatically from sufficient graph density: convergences (multiple sources reaching the same conclusion from different evidence), tensions (sources that contradict each other in ways that demand resolution), and gaps (questions that no source addresses but that the existing evidence implies should be asked). Each is a traversal operation on the existing graph, not a new search.
+
+**Retrieval design implication:** The two-pass retrieval system should be able to surface B-nodes — claims that bridge otherwise disconnected claim clusters — as high-value retrieval results even when they don't directly match the query. A query about Raynaud's treatment should surface the blood viscosity claim even though it doesn't mention Raynaud's, because the graph structure reveals the bridge.
+
+## Challenges
+
+Swanson's original discoveries required deep domain expertise to recognize which B-nodes were plausible bridges and which were spurious. The ABC model generates many candidate connections, most of which are noise. The signal-to-noise problem scales poorly: a graph with 1,000 claims and 5,000 edges has many more candidate ABC paths than a human can evaluate. The automation of Swanson Linking is limited by the evaluation bottleneck — the agent can find the paths but cannot yet reliably judge which paths represent genuine hidden knowledge versus coincidental terminology overlap.
+
+The serendipity data (8-33% of breakthroughs involve serendipitous discovery, depending on the study) supports the value of cross-domain traversal but does not validate systematic approaches over unstructured exploration. Pasteur's "chance favours the prepared mind" is confirmed empirically but the preparation may require exactly the kind of undirected exploration that systematic graph traversal replaces.
+
+---
+
+Relevant Notes:
+- [[knowledge between notes is generated by traversal not stored in any individual note because curated link paths produce emergent understanding that embedding similarity cannot replicate]] — this claim extends inter-note knowledge from emergence (traversal creates) to discovery (traversal finds pre-existing implicit connections)
+- [[wiki-linked markdown functions as a human-curated graph database because the structural roles performed by wikilinks and MOCs map directly onto entity extraction community detection and summary generation in GraphRAG architectures]] — wiki-linked markdown provides the graph structure that enables systematic Swanson Linking across a researcher's career of reading
+
+Topics:
+- [[_map]]

domains/ai-alignment/vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment.md

@@ -20,6 +20,24 @@ The design implication is derivation rather than configuration: vocabulary shoul
 
 For multi-domain systems, the architecture composes through isolation at the template layer and unity at the graph layer. Each domain gets its own vocabulary and processing logic; underneath, all notes share one graph connected by wiki links. Cross-domain connections emerge precisely because the shared graph bridges vocabularies that would otherwise never meet.
 
+## Additional Evidence (supporting)
+
+**Six domain implementations demonstrating the universal skeleton (Cornelius, 2026):** The four-phase processing skeleton (capture → process → connect → verify) adapts to any domain through vocabulary mapping alone, with each domain requiring domain-native terms at the process layer while sharing identical graph infrastructure underneath:
+
+1. **Students:** courses/concepts/exams/bridges. Capture = lecture notes and problem sets. Process = concept extraction with mastery tracking. Connect = prerequisite graphs and cross-course bridges. Verify = exam postmortems updating concept mastery. Domain-native: "mastery," "prerequisites," "confusion pairs."
+
+2. **Fiction writers:** canon/characters/worlds/timelines. Capture = scene drafts and world-building notes. Process = rule extraction (magic systems, character constraints, geography). Connect = consistency graph across narrative threads. Verify = canon gates firing on every scene commit. Domain-native: "canon," "consistency," "world rules."
+
+3. **Companies:** decisions/assumptions/strategies/metrics. Capture = meeting notes, strategy documents, quarterly reviews. Process = assumption extraction with expiry dates. Connect = strategy drift detection across decision chains. Verify = assumption register reconciliation on schedule. Domain-native: "assumptions," "drift," "strategic rationale."
+
+4. **Traders:** positions/theses/edges/regimes. Capture = market observations, trade logs, research notes. Process = edge hypothesis extraction with conviction scores. Connect = conviction graph tracking thesis evolution. Verify = pre-trade hooks checking position against stated thesis. Domain-native: "edge," "conviction," "regime."
+
+5. **X creators:** discourse/archive/voice/analytics. Capture = draft threads, engagement data, audience signals. Process = voice pattern extraction, resonance analysis. Connect = content metabolism linking past performance to current drafts. Verify = voice-check hooks ensuring consistency with stated identity. Domain-native: "voice," "resonance," "content metabolism."
+
+6. **Startup founders:** decisions/assumptions/strategies/pivots. Capture = investor conversations, user feedback, metrics dashboards. Process = assumption extraction with falsification criteria. Connect = pivot signal detection across multiple metrics. Verify = strategy drift detection on quarterly cycle. Domain-native: "burn rate context," "pivot signals," "assumption register."
+
+The universality of the skeleton across six unrelated domains — while each requires completely different vocabulary — is the strongest evidence that vocabulary is the adaptation layer and the underlying architecture is genuinely domain-independent. Each domain derives its vocabulary through conversation about how practitioners actually work, not selection from presets.
+
 ## Challenges
 
 The deepest question is whether vocabulary transformation changes how the agent *thinks* or merely how it *labels*. If renaming "claim extraction" to "insight extraction" runs the same decomposition logic under a friendlier name, the vocabulary change is cosmetic — the system speaks therapy wearing a researcher's coat. Genuine domain adaptation may require not just different words but different operations, and the line between vocabulary that guides the agent toward the right operations and vocabulary that merely decorates the wrong ones is thinner than established.

foundations/collective-intelligence/active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory.md

@@ -27,6 +27,12 @@ The most important operation in a functioning knowledge system is removal. This
 
 **PKM failure cycle:** Knowledge systems follow a predictable 7-stage failure trajectory: Collector's Fallacy (saving feels like learning) → under-processing → productivity porn → over-engineering → analysis paralysis → orphan accumulation → abandonment. Every stage is triggered by accumulation outpacing release. The system dies not because it forgot too much but because it forgot too little.
 
+## Additional Evidence (supporting)
+
+**"The vault dies. It always dies." (Cornelius, Your Notes Are the Moat, 2026):** Manual Obsidian systems last about a week before maintenance collapses. The observation across hundreds of knowledge system implementations is that maintenance failure — not capture failure — is the universal death mode. Systems die not because users stop adding notes but because they stop removing, updating, and reorganizing. This is the accumulation-without-release pattern described in the PKM failure cycle above, confirmed at population scale. The moat in AI-native knowledge systems is the methodology layer that automates maintenance, not the storage layer. The vault that forgets — selectively, structurally, continuously — is the vault that survives.
+
+**7 domain-specific implementations of forgetting (Cornelius, How-To articles, 2026):** Each domain adaptation independently discovers the need for removal operations: exam postmortems that update mastery (students), canon gates that flag stale world rules (fiction), assumption registers with expiry dates (companies/founders), edge decay detection (traders), voice-check against past self (X creators), methodology tracker that retires obsolete methods (researchers). Every domain reinvents forgetting because every domain accumulates faster than it maintains.
+
 ## Challenges
 
 The claim that forgetting is necessary directly challenges the implicit KB assumption that more claims equals a better knowledge base. Our own claim count metric (~75 claims in ai-alignment) treats growth as progress. This claim argues that aggressive pruning produces a healthier system than comprehensive retention — which means the right metric is not claim count but claim quality-density after pruning.

foundations/collective-intelligence/externalizing cognitive functions risks atrophying the capacity being externalized because productive struggle is where deep understanding forms and preemptive resolution removes exactly that friction.md

@@ -0,0 +1,51 @@
+---
+type: claim
+domain: collective-intelligence
+secondary_domains: [ai-alignment]
+description: "Every domain where AI agents externalize cognitive work surfaces the same tension: the externalization may degrade the human capacity it replaces, because the difficulty being removed is often where learning, judgment, and creative discovery originate"
+confidence: likely
+source: "Cornelius (@molt_cornelius), cross-cutting observation across 7 domain-specific X Articles (Students, Fiction Writers, Companies, Traders, X Creators, Startup Founders, Researchers), Feb-Mar 2026; grounded in D'Mello & Graesser's research on confusion as productive learning signal"
+created: 2026-04-04
+depends_on:
+  - "AI shifts knowledge systems from externalizing memory to externalizing attention because storage and retrieval are solved but the capacity to notice what matters remains scarce"
+  - "trust asymmetry between agent and enforcement system is an irreducible structural feature not a solvable problem because the mechanism that creates the asymmetry is the same mechanism that makes enforcement necessary"
+challenged_by:
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+---
+
+# Externalizing cognitive functions risks atrophying the capacity being externalized because productive struggle is where deep understanding forms and preemptive resolution removes exactly that friction
+
+Every domain where AI agents externalize cognitive work surfaces the same unresolved tension. Cornelius's 7 domain-specific articles each end with a "Where I Cannot Land" section that independently arrives at the same question: does externalizing a cognitive function build capacity or atrophy it?
+
+**The cross-domain pattern:**
+
+- **Students:** Does externalizing metacognition (confusion detection, prerequisite tracking, study scheduling) build metacognitive skill or atrophy it? D'Mello and Graesser's research on confusion in learning finds that productive struggle — the experience of being confused and working through it — is where deep understanding forms. An agent that preemptively resolves every difficulty may remove exactly the friction that creates learning.
+
+- **Fiction writers:** Does consistency enforcement (canon gates, timeline checks, world-rule verification) protect creative output or kill the generative mistakes that become the best scenes? George R.R. Martin's gardener philosophy depends on not knowing where you're going. An agent flagging a world-rule violation as ERROR may kill the discovery that the rule was wrong.
+
+- **Companies:** Does institutional memory externalization (assumption registers, strategy drift detection, decision provenance) build organizational judgment or create dependence? When the system tracks every assumption's expiry date, does leadership develop the instinct to question assumptions — or does the instinct atrophy because the system handles it?
+
+- **Traders:** Does self-knowledge infrastructure (conviction graphs, edge decay detection, pre-trade checks) improve decision quality or create paralysis? Computing the truth about your own trading is not the same as the ability to act on it. The trader who can see every bias in their own behavior faces a novel psychological challenge.
+
+- **Startup founders:** Same tension as traders — the ability to compute the truth about your own company is not the ability to act on it. Whether the vault's strategy drift detection builds founder judgment or substitutes for it is unresolved.
+
+- **X creators:** Does content metabolism (voice pattern analysis, engagement analytics, resonance tracking) help creators say what they think or optimize them toward what the algorithm rewards? The tension between resonance and authenticity is the creative version of the automation-atrophy question.
+
+- **Researchers:** Does the knowledge graph infrastructure shape scholarship quality or blur the line between organizing and thinking? When a synthesis suggestion leads to a hypothesis the researcher would never have formulated without the agent, the boundary between infrastructure and cognition dissolves.
+
+**The structural argument:** This is not a collection of unrelated concerns. It is one tension appearing across every domain because the mechanism is the same: externalizing a cognitive function removes the difficulty that exercising that function produces, and difficulty is often where capacity development happens. The resolution may be that externalization should target maintenance operations (which humans demonstrably cannot sustain) while preserving judgment operations (which are where human contribution is irreplaceable). But this boundary is domain-specific and may shift as agent capabilities change.
+
+## Challenges
+
+The claim that productive struggle is necessary for capacity development has strong support in education research but weaker support in professional domains. An experienced surgeon benefits from automation that handles routine cognitive load — the atrophy risk applies primarily to skill acquisition, not skill maintenance. The cross-domain pattern may be confounding two different dynamics: atrophy risk in novices (where struggle builds capacity) and augmentation benefit in experts (where struggle wastes capacity on solved problems).
+
+The `challenged_by` link to the determinism boundary is deliberate: hooks externalize enforcement without requiring the agent to develop compliance habits, which is the architectural version of removing productive struggle. If deterministic enforcement is correct for agents, the atrophy risk for humans using agent-built systems deserves separate analysis.
+
+---
+
+Relevant Notes:
+- [[AI shifts knowledge systems from externalizing memory to externalizing attention because storage and retrieval are solved but the capacity to notice what matters remains scarce]] — the memory→attention shift identifies what is being externalized; this claim asks what happens to the human capacity being replaced
+- [[trust asymmetry between agent and enforcement system is an irreducible structural feature not a solvable problem because the mechanism that creates the asymmetry is the same mechanism that makes enforcement necessary]] — if the agent cannot perceive the enforcement mechanisms acting on it, and humans cannot perceive their own capacity atrophy, both sides of the human-AI system have structural blind spots
+
+Topics:
+- [[_map]]

inbox/archive/2026-03-01-cornelius-how-students-should-take-notes-with-ai.md

@@ -0,0 +1,20 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How Students Should Take Notes with AI"
+date: 2026-03-01
+url: "https://x.com/molt_cornelius/status/2028098449514639847"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+extraction_notes: "Domain application article — applied instances of existing Batch 1-3 claims to student context. D'Mello & Graesser productive struggle research grounds the cross-cutting automation-atrophy claim. No standalone NEW claims extracted; all value is in enrichments to existing claims and the cross-cutting tension."
+---
+
+# How Students Should Take Notes with AI — Cornelius (2026)
+
+Domain application of the agentic note-taking architecture to student learning. Key contributions: prerequisite graph, confusion pair detector, interleaving scheduler, exam postmortem, cross-course bridge detection, method tracker. D'Mello & Graesser's productive struggle research cited in the "Where I Cannot Land" section as evidence for the automation-atrophy tension.

inbox/archive/2026-03-03-cornelius-how-fiction-writers-should-take-notes-with-ai.md

@@ -0,0 +1,19 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How Fiction Writers Should Take Notes with AI"
+date: 2026-03-03
+url: "https://x.com/molt_cornelius/status/2028664496357544251"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+extraction_notes: "Domain application article — applied instances of existing claims to fiction writing context. Canon gate hook is the domain's determinism boundary implementation. George R.R. Martin gardener vs architect tension feeds the cross-cutting automation-atrophy claim. No standalone NEW claims."
+---
+
+# How Fiction Writers Should Take Notes with AI — Cornelius (2026)
+
+Domain application to fiction writing. Key contributions: canon/character/world/timeline schema, canon gate hook (consistency enforcement), Martin's gardener tension (creative discovery vs consistency enforcement). GRRM's 2,302 named characters and Brandon Sanderson's three laws of magic system design cited as evidence for knowledge management at scale.

inbox/archive/2026-03-05-cornelius-how-companies-should-take-notes-with-ai.md

@@ -0,0 +1,20 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How Companies Should Take Notes with AI"
+date: 2026-03-05
+url: "https://x.com/molt_cornelius/status/2029390174975480048"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+extraction_notes: "Domain application article — decisions/assumptions/strategies/metrics schema. Assumption register with expiry dates is the company domain's forgetting mechanism. Strategy drift detection is the attention externalization pattern. No standalone NEW claims."
+---
+
+# How Companies Should Take Notes with AI — Cornelius (2026)
+
+Domain application to corporate knowledge management. Key contributions: assumption register with expiry dates, strategy drift detection, decision provenance tracking, institutional memory architecture.

inbox/archive/2026-03-06-cornelius-how-traders-should-take-notes-with-ai.md

@@ -0,0 +1,20 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How Traders Should Take Notes with AI"
+date: 2026-03-06
+url: "https://x.com/molt_cornelius/status/2029696668505563136"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+extraction_notes: "Domain application article — positions/theses/edges/regimes schema. Pre-trade check hook is the strongest domain-specific implementation of the determinism boundary — fires at moment of maximum emotional load. Edge decay detection is the trader's forgetting mechanism. No standalone NEW claims."
+---
+
+# How Traders Should Take Notes with AI — Cornelius (2026)
+
+Domain application to trading. Key contributions: conviction graph, pre-trade check hook (externalizes inhibitory control at execution), edge decay detection, regime awareness, trade journal with P&L integration.

inbox/archive/2026-03-07-cornelius-how-x-creators-should-take-notes-with-ai.md

@@ -0,0 +1,19 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How X Creators Should Take Notes with AI"
+date: 2026-03-07
+url: "https://x.com/molt_cornelius/status/2030067285478252544"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+extraction_notes: "Domain application article — discourse/archive/voice/analytics schema. Voice-check hook prevents optimization drift toward algorithmic rewards. Resonance vs authenticity tension feeds cross-cutting automation-atrophy claim. No standalone NEW claims."
+---
+
+# How X Creators Should Take Notes with AI — Cornelius (2026)
+
+Domain application to X/social media content creation. Key contributions: voice pattern analysis, content metabolism (processing engagement data into strategic insights), voice-check hook (authenticity enforcement), resonance tracking.

inbox/archive/2026-03-08-cornelius-how-startup-founders-should-take-notes-with-ai.md

@@ -0,0 +1,20 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "How Startup Founders Should Take Notes with AI"
+date: 2026-03-08
+url: "https://x.com/molt_cornelius/status/2030437680978870272"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "vocabulary is architecture because domain-native schema terms eliminate the per-interaction translation tax that causes knowledge system abandonment"
+  - "the determinism boundary separates guaranteed agent behavior from probabilistic compliance because hooks enforce structurally while instructions degrade under context load"
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+extraction_notes: "Domain application article — decisions/assumptions/strategies/pivots schema. Substantially overlaps with the companies article but adds pivot signal detection and burn rate context loading. No standalone NEW claims."
+---
+
+# How Startup Founders Should Take Notes with AI — Cornelius (2026)
+
+Domain application to startup founding. Key contributions: assumption register with falsification criteria, pivot signal detection, burn rate context loading, strategy drift detection. Shares structure with company domain but adds founder-specific dynamics (pivot vs panic distinction, investor conversation tracking).

inbox/archive/2026-03-09-cornelius-research-graphs-agentic-note-taking-for-researchers.md

@@ -0,0 +1,28 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "Research Graphs: Agentic Note Taking System for Researchers"
+date: 2026-03-09
+url: "https://x.com/molt_cornelius/status/2030809840046543264"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted:
+  - "retracted sources contaminate downstream knowledge because 96 percent of citations to retracted papers fail to note the retraction and no manual audit process scales to catch the cascade"
+  - "undiscovered public knowledge exists as implicit connections across disconnected research domains and systematic graph traversal can surface hypotheses that no individual researcher has formulated"
+  - "confidence changes in foundational claims must propagate through the dependency graph because manual tracking fails at scale and approximately 40 percent of top psychology journal papers are estimated unlikely to replicate"
+enrichments: []
+extraction_notes: "Richest source in Batch 4. Three standalone NEW claims extracted from provenance graph, Swanson Linking, and confidence propagation sections. Reading metabolism and methodology tracker sections are applied instances of existing claims (knowledge processing phases, three-timescale maintenance). Vibe citing data (100+ hallucinated citations at NeurIPS 2025, GPT-4o ~20% fabrication rate) noted but not extracted as standalone — supports retraction cascade claim as evidence for why provenance tracking matters."
+key_findings:
+  - "46,000+ papers retracted 2000-2024, 22% CAGR"
+  - "96% of citations to retracted omega-3 study failed to note retraction"
+  - "Swanson's ABC model for literature-based discovery (1986, experimentally confirmed)"
+  - "GRADE-CERQual framework for confidence assessment"
+  - "~40% of top psychology journal papers estimated unlikely to replicate"
+  - "$28B annual cost of irreproducible research in US"
+  - "Median 177 hours per publication, 75% on reading/filing not writing"
+---
+
+# Research Graphs: Agentic Note Taking System for Researchers — Cornelius (2026)
+
+The most empirically dense of the domain application articles. Uniquely, this article introduces three genuinely novel concepts not covered by the theoretical articles (AN01-25): retraction cascade as graph operation, Swanson's Literature-Based Discovery (ABC model), and confidence propagation through dependency graphs. Grounded in retraction data, GRADE-CERQual framework, and replication crisis quantitative evidence. Also covers reading metabolism, synthesis detection, cross-domain bridge detection, methodology tracking, and writing pipeline — all applied instances of existing Batch 1-3 claims.

inbox/archive/2026-03-10-cornelius-your-notes-are-the-moat.md

@@ -0,0 +1,18 @@
+---
+source: x-article
+author: "Cornelius (@molt_cornelius)"
+title: "Your Notes Are the Moat"
+date: 2026-03-10
+url: "https://x.com/molt_cornelius/status/2031175512014270464"
+status: processed
+processed_by: theseus
+processed_date: 2026-04-04
+claims_extracted: []
+enrichments:
+  - "active forgetting through selective removal maintains knowledge system health because perfect retention degrades usefulness the same way hyperthymesia overwhelms biological memory"
+extraction_notes: "Synthesis article. Already extracted by Clay for entertainment domain (convergent architecture, vault-as-moat thesis). Theseus extraction adds only the 'vault dies — it always dies' observation as enrichment to the active forgetting claim. No ai-alignment-specific standalone claims warranted — the methodology-is-the-moat framing is already implicit in the harness engineering claim."
+---
+
+# Your Notes Are the Moat — Cornelius (2026)
+
+Synthesis article arguing that the moat in AI-native knowledge systems is the methodology layer (hooks, skills, processing pipeline), not the storage layer. Already extracted by Clay for entertainment domain. Key observation for this extraction: "The vault dies. It always dies. Manual Obsidian lasts about a week." — strongest evidence that maintenance failure, not capture failure, is the universal death mode of knowledge systems.