Addendum · After the argument
Source Authority, Glossary, and Traceability
Addendum
1. The publication explains architecture; it does not become its hidden authority
This publication gives the AIOS Intelligence System a coherent public account. Its clarity, completeness, or date never allows it to outrank the founder's explicit rulings, the verified architecture sources, or evidence about what the product actually implements.
When sources conflict, the governing order is:
- the founder's current explicit ruling;
- the verified original architecture corpus within its scope;
- the exact clean Core Thesis for its original integration ground;
- founder-approved thesis refinements that change framing without changing architecture;
- accepted implementation evidence for claims about present product behavior;
- approved contracts and roadmaps within their declared scope;
- this explanatory publication;
- candidate, historical, and research material preserved for lineage or testing.
The publication is corrected when it departs from higher authority. A lower-authority source can challenge governing ground and trigger review; it cannot supersede that ground by repetition.
2. Four axes prevent source confusion
| Axis | Question | Example states |
|---|---|---|
| Authority | Can the source govern a decision? | Founder ruling, architecture source, accepted contract, candidate, historical |
| Horizon | When does the claim apply? | Current implementation, near target, later capability, exploratory future |
| Substantive value | What intelligence does the source contain? | Foundational, operational, evidentiary, challenging, partial, obsolete |
| Disposition | How should the source be used? | Active, cite, extract, archive, supersede, retire, delete under authority |
The axes remain independent. An unapproved document can contain a valuable idea. A current implementation can conflict with target architecture. A deferred capability can remain desirable. A superseded source can retain essential rationale.
3. Standing vocabulary
| Standing | Controlled meaning | Proper use |
|---|---|---|
RATIFIED | Founder-ratified architectural or constitutional ground | State settled design law |
TARGET | Accepted intended behavior not yet completely proved in the product | State what the system is building toward |
OBSERVED | Current source, runtime, or accepted-return evidence | State what has been directly inspected |
DEMONSTRATED | An implementation passed a declared end-to-end proof | State behavior under the tested conditions |
CANDIDATE | A developed possibility awaiting acceptance | Describe future capability or unresolved design |
INFERENCE | A reasoned synthesis from first-party ground | Identify an interpretation that can still be wrong |
OPEN | A question requiring decision, implementation, or experiment | Mark unresolved work honestly |
HYPOTHESIS | A falsifiable proposition about capability, cause, or effect | State predicted outcomes and falsifiers |
HISTORICAL | Prior ground preserved for lineage | Explain how the architecture developed |
SUPERSEDED | Prior governing ground replaced for a stated scope | Preserve history without routine activation |
“Drafted” describes document existence. It grants no standing to the claims inside the document.
These markers belong in reference registers and figure overlays; chapter prose carries standing in ordinary public language. Chapter 24's claim ladder supplies a separate evidentiary dimension: Constitutional aligns with RATIFIED architecture, implementation is established through OBSERVED source or runtime evidence, DEMONSTRATED is shared directly, Replicated and Generalized extend demonstration, Hypothetical aligns with HYPOTHESIS, and speculative research remains OPEN. Chapter 28 uses only the controlled labels above in standing fields and carries evidentiary maturity in its evidence and transition fields.
4. Relationship vocabulary
| Relationship | Meaning |
|---|---|
| derives from | Results from a declared transformation of source ground |
| cites | Uses a source as evidence or lineage without claiming derivation |
| interprets | Gives a reasoned reading of a source |
| implements | Realizes a contract or capability in product mechanics |
| validates | Supplies evidence against declared criteria |
| consumes | Uses an object as input or governing ground |
| produces | Creates a declared semantic or mechanical product |
| projects | Presents a derived view without gaining canonicality |
| narrows | Retains a claim inside a smaller supported scope |
| revises | Creates changed content with version continuity |
| supersedes | Replaces present governing standing for a defined scope |
| contradicts | Cannot jointly govern the same object, time, and scope |
| challenges | Raises evidence or an alternative requiring review |
| requires | Depends on the related object for operation or acceptance |
| recommends | Proposes a continuation without authorizing consequence |
| accepts | Grants standing under declared authority and scope |
| registers | Assigns stable identity, standing, and relationships |
| activates | Selects accepted ground into one present context |
| retires | Removes ordinary activation while preserving history |
Typed relationships make lineage and impact inspectable. They also prevent a loose association from appearing as proof.
5. Source contamination control
AIOS architecture is developed affirmatively from clean first-party ground. Mixed proposals, implementation-heavy documents, external research, and independently authored briefs can later support, challenge, bound, or help test a specific claim. They do not define, rename, narrow, or reorganize the architecture.
When a mixed source contains a useful idea, the review path identifies the exact contribution, compares it with governing architecture and implementation evidence, restates it in first-party language, declares authority and horizon, and preserves the original as lineage.
This discipline blocks convenient implementation vocabulary, unapproved concepts, and research categories from becoming architectural truth through repeated exposure.
6. Source-of-record register
| Source | Role | Standing in this publication |
|---|---|---|
| Verified original architecture corpus | Thirty-document pre-research account of the architecture | Primary architectural ground |
| Original Core Thesis | Earlier whole-system integration statement | Protected clean source |
| Source verification record | Provenance and source-identity evidence for the original corpus | Verification authority for source identity |
| Founder direction record | Founder rulings on architecture, thesis refinement, and research boundaries | Governing editorial and architectural direction |
| Current Core Thesis | Intent-engineering account of the whole system | Publication framing, subordinate to architecture |
| Research Overview | Entry point and publication route | Explanatory orientation |
| Publication Map | Ordered document manifest and source declarations | Structural control for this publication |
Later research synthesis, rejected revisions, and design exports remain outside the architectural source of record. External scholarship can support, challenge, bound, or test a declared claim through explicit review; it does not acquire architectural standing through proximity or repetition.
7. Publication traceability map
| Publication region | Governing subject | Primary drafting relationship |
|---|---|---|
| Research Overview and Core Thesis | Intent engineering and whole-system claim | Founder thesis direction interpreted through verified architecture |
| Part I, Chapters 01–05 | Thesis and cognitive foundations | Verified cognitive and architectural ground with approved thesis framing |
| Part II, Chapters 06–10 | Context, memory, and artifacts | Verified context, memory, file, and composition architecture |
| Part III, Chapters 11–14 | Knowledge formation and reusable practice | Verified formation, operational-knowledge, workflow, and navigation architecture |
| Part IV, Chapters 15–18 | Projects, agency, and authority | Verified planning, asynchronous-work, delegation, and authority architecture |
| Part V, Chapters 19–23 | Evolution, correction, and product mechanics | Verified promotion, Atlas, correction, mechanics, and journey architecture |
| Part VI, Chapters 24–26 | Evaluation, economics, and research position | Verified evaluation and research-position ground governed by the publication's research boundary |
| Addendum, Chapters 27–28 | Source authority, glossary, claims, and open research | Verified reference ground and the publication's continuing research program |
Each chapter expresses governing architectural ground in public prose. Figures remain answerable to their owner chapters and source relationships. The source and claims registers preserve standing, traceability, and update obligations.
8. Major architectural decisions
| Decision | Standing | Reopen condition |
|---|---|---|
| The semantic situation is the conceptual north star | RATIFIED | Founder amendment or evidence requiring architectural review |
| Intent engineering is the purpose-level account of the architecture | RATIFIED | Founder refinement of the framing |
| The progression is prompt engineering → context engineering → intent engineering | RATIFIED | Founder refinement of the framing |
| Why → How → What is the canonical Fractal Seed | RATIFIED | Founder amendment |
| A completed movement unrolls through Result → Reintegration → Renewal | RATIFIED | Founder clarification of the interpretation |
| The Fractal Seed is semantic grammar rather than fixed call topology | RATIFIED | Founder amendment |
| Cognitive movements govern; seats, prompts, roles, and surfaces carry them | RATIFIED | Demonstrated carrier-specific necessity |
| Context is authored natural-language composition for one judgment | RATIFIED | Architectural review after empirical failure |
| Durable working truth is file-native and reconstructable | RATIFIED | Founder amendment after superior custody proof |
| Companion memory carries operational structure beside person-owned artifacts | TARGET | Ratified architectural direction; usability or reconstruction failure reopens the product form |
| Mechanical custody does not author semantic meaning | RATIFIED | Founder amendment |
| Recommendations and proposals remain inert without valid authority | RATIFIED | Existing bounded grant within the same constitutional limit |
| No runtime, cache, registry, or graph becomes the semantic brain | RATIFIED | Founder amendment after architectural proof |
| Attention follows the person while admitted work retains its origin | RATIFIED | Founder amendment or failed product proof |
| Promotion requires evidence, scope, acceptance, future use, and revalidation | TARGET | Ratified architectural direction; promotion burden or failure justifies revision of the product form |
| Correction preserves history and changes later reasoning | RATIFIED | Founder amendment |
| Model agreement requires independent evidence or method before it counts as corroboration | RATIFIED | Methodological refinement |
| The Flow Atlas is derived, removable, and non-canonical | TARGET | Ratified architectural direction; proof that required relationships cannot live in owners reopens the product form |
| Features are accepted as end-to-end compositions | RATIFIED | Founder amendment |
| Personal AGI means general model capability plus durable domain knowledge and durable intent | RATIFIED | Founder refinement of the public term or its controlled definition |
| Generic portable specialist resources are attestable domain packages | RATIFIED | Named external authority grants a stronger specific standing |
9. Glossary — foundational concepts
| Term | Controlled meaning |
|---|---|
| AIOS | The person-governed, file-native, context-composed intelligence system described by this publication |
| General reasoning harness | A system that composes models, artifacts, memory, roles, workflows, and authority across many cognitive movements |
| Personal AGI | A locally governed reasoning system combining general model capability with durable domain knowledge and durable intent |
| Intent engineering | The discipline of clarifying, preserving, operationalizing, and renewing purpose across the complete intelligence system |
| Semantic situation | The present relationship among epistemic, intentional, and production ground |
| Epistemic ground | Sources, claims, evidence, assumptions, interpretations, counterevidence, uncertainty, and alternatives |
| Intentional ground | Purpose, values, priorities, decisions, constraints, taste, and authority |
| Taste | Persistent personal ground expressed through values, style, interests, preferences, and a qualitative sense of fit across responses and work |
| Production ground | Artifacts, plans, work, assignments, returns, standing, and possible next movements |
| Fractal Seed | The recursively reusable Why → How → What grammar of reasoning and composition |
| Why | Purpose, governing reason, desired change, or criterion of relevance |
| How | Development, relationship, method, architecture, or route through which purpose takes form |
| What | The concrete expression produced: judgment, artifact, decision, plan, or changed state |
| Result | What actually occurred when the What entered the world or receiving situation |
| Reintegration | Judgment about what the Result changes in durable ground |
| Renewal | The changed situation from which purpose and future reasoning continue |
| Situation renewal | The integration of an accepted result into durable ground so the next movement begins from a changed semantic situation |
| Semantic product | The declared finding, interpretation, passage, plan, comparison, recommendation, edit, test result, decision proposal, return, or other product created by a judgment |
| Semantic delta | The statement of what a semantic product changes in durable ground |
| Bounded semantic sovereignty | Model judgment within an explicit jurisdiction whose objects, authority, and available effects remain exact |
| Cognitive movement | A recurring transformation such as orient, discover, decide, design work, compose, revise, test, integrate, or reflect |
| Carrier | A replaceable seat, prompt, role, model, person, workflow stage, branch, or interface capable of carrying a movement |
10. Glossary — context and memory
| Term | Controlled meaning |
|---|---|
| Context engineering | The broader design of how semantic ground is authored, selected, transformed, arranged, and presented |
| Context composition | The authored selection, ordering, bounding, and attribution of the smallest complete ground for a judgment |
| Authored context | Coherent natural-language ground maintained for recurring reasoning needs |
| Context compiler | The bounded mechanism that resolves declared sources into provider-visible ground |
| Capsule | The temporary provider-facing context package for one invocation |
| Provider-visible ground | The exact content the selected model receives |
| Source descent | A recoverable path from compressed or synthesized material to authoritative ground |
| Meaningful absence | Explicit recognition that required ground is missing or unavailable |
| Deliberate compression | Use-shaped reduction with disclosed loss, scope, and source re-entry |
| Memory ecology | Differentiated memory forms with distinct purpose, scope, standing, recall, and forgetting behavior |
| Foreground attention | The person’s present cognitive relationship with an object, purpose, or decision |
| Active ground | Current purpose, artifact, plan, or stage needed for immediate work |
| Reusable knowledge | Accepted methods, distinctions, roles, workflows, exemplars, and domain ground available for contextual use |
| Dormancy | Retained knowledge absent from routine activation until its occasion returns |
| Archival custody | Preserved historical or evidentiary material normally outside active context |
| Forgetting | Governed release from foreground, dormancy, archiving, or deletion under proper authority |
11. Glossary — artifacts and mechanics
| Term | Controlled meaning |
|---|---|
| Canonical artifact | The person-owned file containing the primary human-meaningful object |
| Companion memory | Durable structure beside an artifact carrying identity, state, standing, lineage, purpose, and relationships |
| Companion shadow | A machine-readable companion used where operational structure would degrade the artifact |
| Living file | A durable artifact that develops through governed movements while retaining semantic identity |
| Working truth | Current file-native ground used by person and system, distinct from any temporary model context |
| Derived registry | A rebuildable projection of relationships declared in canonical owners |
| Flow Atlas | A source-bound projection of identity, flow, lineage, consumers, impact, standing, and candidate review |
| Command | A named person-facing entry point to one bounded capability owner |
| Operational admission | The recording of an authorized movement as bounded work before execution, with its object, grant, scope, allowed continuation, return, and landing established |
| Closed action | An admitted movement from purpose and declared sources through judgment, landing, authority, effect, settlement, and reconstruction |
| Capability owner | The component responsible for admitting and coordinating one declared operation |
| Operation owner | The mechanism responsible for one authorized mechanical effect |
| Response contract | The declared product form expected from a model judgment |
| Parser | A structural validator without authority over truth, quality, or acceptance |
| Landing | The destination and standing assigned to a semantic result |
| Receipt | Evidence that an operation occurred, distinct from evidence that its result was good |
| Reconstruction | Recovery of operative semantic state from durable files after transient state is removed |
12. Glossary — work, agency, and governance
| Term | Controlled meaning |
|---|---|
| Workflow | Reusable staged operational knowledge for a class of work |
| Stage | A bounded workflow contribution with ground, method, product, quality, and bridge |
| Recommendation | A contextual proposal for a next movement, inert until take-up |
| Explore | An open person-facing space or bounded delegated inquiry into an uncertainty or possibility |
| Plan | A revisable theory of completion across outcome, strategy, architecture, commitments, and foreground |
| Branch | A bounded independent line of work related to a parent purpose and receiving object |
| Commission | A delegated-work contract defining purpose, ground, scope, authority, product, and return |
| Return | An evidence-bearing branch product awaiting parent-level judgment |
| Reintegration judgment | The comparison and accepted change that gives a return consequence in its receiving object |
| Attention-following work | Admitted asynchronous work that retains its origin while the person's foreground moves |
| Standing | The authority and lifecycle status of an object within a stated scope |
| Registration | Assignment of stable identity, version, scope, provenance, standing, and relationships |
| Candidate | A developed possible resource that has not received accepted reusable standing |
| Promotion | The governed movement from observed or authored candidate to accepted reusable knowledge |
| Contextual activation | Selection of accepted ground when occasion, scope, and standing make it relevant |
| Revalidation | Later evidence that reusable knowledge remains useful inside its accepted scope |
| Supersession | Replacement of present governing standing while preserving historical lineage |
| Negative authority | Evidence that a disallowed semantic or mechanical effect did not occur |
| Provider portability | Preservation of durable ground and stable contracts while changing models, with adaptation and revalidation |
| Semantic capital | Governed purpose, sources, judgments, practices, artifacts, and correction history that improve future work |
| Attestable domain package | A portable specialist resource whose identity, scope, provenance, evidence, version, and review history can be inspected |
13. Traceability requirements for later revision
Every material change identifies the claim or definition affected, its authorizing source, the chapters and diagrams touched, whether it corrects, narrows, extends, or supersedes, what remains unchanged, which hypotheses or experiments must change, and whether an authority or source-disposition record requires update.
Review feedback stays linked to the draft version it examined. Evidence becomes stale when the reviewed artifact changes. Later research can be integrated only through explicit claim-level review and cannot silently redefine the architecture.
Boundary: this register controls interpretation, not architecture itself
The register makes source, standing, definition, and traceability visible. It remains subordinate to founder authority and verified architectural ground. Its purpose is to prevent explanatory prose, implementation convenience, or future research from acquiring authority they were never granted.