# Source of Truth and Aging Rate

This doctrine governs how the method assigns authority to surfaces. A surface is authoritative for a class of claim because it is sized to that claim's aging rate. Mixing aging rates within one surface produces a surface that ages at the rate of its fastest-aging contents — usually badly. This doc names the split, the failure modes, and the routing rule.

[`AGENTS.md`](../AGENTS.md) carries the aging-rate principle as a workflow rule and the source-of-truth split as a brief enumeration. [`docs/method.md`](method.md) references both as elements within the broader method articulation. This doc carries them at full doctrine altitude: definition, the operative surface split, failure modes, boundaries with adjacent doctrine, and a practical decision rule.

## Definition

- **Source of truth** is the authoritative surface for a given class of claim. When two surfaces disagree, the authoritative one wins. Authority is assigned by claim type, not by recency or fluency.
- **Aging rate** is how quickly a claim type becomes stale. State ages fast. Rules age slowly. Source of intent ages slowly. Per-conversation tracking is ephemeral. Reflection moves at its own pace.
- **Each surface should be sized to one dominant aging rate.** A surface that mixes a slow-aging payload with fast-aging contents ages at the rate of the fastest contents. The slow payload becomes unreadable through the noise; the fast payload becomes stale-looking under the surface's slow-aging frame.

Source of truth and aging rate are the same problem viewed from two angles. Source of truth asks *which surface owns this claim?* Aging rate asks *how fast does this claim go stale?* Both questions resolve to: pick the surface whose aging cadence matches the claim's, and let that surface be authoritative.

## Core split

The method distinguishes the roles and authority of these surfaces:

- **Repo** — project state, public-safe truth, current artifacts, doctrine docs, code, schema. Authoritative for *repo-owned project truth and current repo state*; purpose, workflow, identity, operator state and delegated domain judgment remain with their own owners. Aging rate is project-specific but generally slow within categories (doctrine slower; state faster).
- **`AGENTS.md`** — repo-local workflow rules for executors working on the repo. Authoritative for *how repo work is done*. Slow-aging. Refreshes only when a workflow rule is added, removed, or materially revised — not when project state changes.
- **Grounding note** — slow-aging source-of-intent + external context (graduation origin, audience, durable role boundaries, foundational premises, durable loose threads). Authoritative for *why this project exists and what its enduring premises are*. Slow-aging by design. Lives operator-side, external to the repo.
- **Scratch** — event records, state chronology, working memory, in-flight per-conversation tracking, post-event recaps. Authoritative for *what happened, when, in what sequence*. Fast-aging. Lives operator-side.
- **[`method-ASK`](../README.md) docs** — cross-project method doctrine that has earned generalization. Authoritative for *the method itself*. Slow-aging. Refreshes only when the underlying method articulation has materially evolved.
- **[`control-surface`](https://github.com/apexSolarKiss/control-surface)** — execution protocol, workflow templates, prompts, project-instantiation workflow, the per-PR cadence. Authoritative for *how sessions operate against repos*. Slow-aging.
- **Articles** — reflection at its own pace. Authoritative for *the public reflective articulation of method elements*, not for current repo state or method doctrine. Own pace, distinct from doctrine cadence.
- **Current task context** — per-conversation tracking, in-flight task lists, thread history, and session state. Ephemeral working context, not a durable owner. Authorized source material may be read into it without transferring authority from its named owners. “Ephemeral” describes the task-state role, not a deletion rule for transcripts or governed records.
- **Private agent memory** — an executor's own persistent store (for example Claude Code's `MEMORY.md`). **Non-authoritative operator context.** It never owns project truth, workflow protocol, or current task state. Eligible residue is separately classified and requires exact-target, exact-payload operator approval before it is written. Persistence does not confer authority.

Each surface has a single dominant aging rate. The split is not separation-of-concerns alone; it is separation-by-aging-rate.

## Failure modes

- **Mixing fast state into slow grounding notes.** Event chronology, review outcomes, or current-stage status absorbed into the grounding note. The grounding note then ages at the rate of the events it carries, losing its slow-source-of-intent property.
- **Duplicating repo truth in grounding notes.** Repeating current project state in the grounding note creates two sources for the same claim. The grounding note drifts as the repo updates; the repo is authoritative but the grounding note is read first.
- **Treating scratch as doctrine.** A well-formed scratch artifact reads as authoritative even though it was sized to fast-aging contents. Promoting it into doctrine without absorption discipline (see boundary below) bypasses the surface-sizing check.
- **Treating articles as repo specs.** Articles reflect at their own pace; they are not authoritative for current repo state. A reader who treats the article as the operating rule risks acting against a frame the repo has already moved past.
- **Treating generated summaries as authoritative without artifact verification.** A chat-side summary of file contents, branch state, or PR diff can sound coherent while being stale. Authority belongs to the artifact, not to the summary.
- **Letting chat-state outrank repo-state.** Conversational fluency can produce confident claims about current file content that the actual file has already superseded. The repo decides; the conversation may frame.
- **Letting a surface age at the rate of its fastest-aging contents.** The meta-failure. Any surface that admits material from a faster-aging class will gradually take that faster cadence — regardless of intent. The fix is upstream (refuse the material) rather than downstream (try to age the surface differently).

These failure modes share a structural cause: authority assignment by content urgency rather than by surface fit. The remedy is surface-sizing applied at the moment of writing, before the material lands.

## Boundary with source of intent

[`docs/source-of-intent.md`](source-of-intent.md) carries the doctrine governing source-of-intent origination, recovery, validation, and the grounding-note refresh preflight. This doc references it as the upstream concept for the *source of intent* layer in the surface split above.

Specifically:

- Source of intent is operator-owned. The method can elaborate purpose endlessly and propose new purposes or governing standards, but it cannot authorize those proposals merely by generating or executing them. The grounding note is the surface that carries operator-validated source of intent.
- Recovery and validation precede repo encoding. The repo is not the first artifact; recovered intent is. The repo encodes the structure that validated intent earned.
- The source-of-intent guard prevents unresolved means from becoming unresolved purpose. Most apparent purpose-level gaps are actually architectural-means questions that should be answered method-side, not routed back to the operator.
- The grounding-note refresh preflight classifies candidate content before it enters the grounding note. State / event records route to scratch; slow source-of-intent routes to the grounding note; repo-local public-safe truth routes to the repo; future roadmap is held; advisor scratch waits for validation; premature implementation architecture is held.

The aging-rate principle determines surface authority. The source-of-intent doctrine determines what counts as authoritative source-of-intent at the slow-aging layer. The two docs compose: source-of-intent doctrine names the upstream concept; this doc names the surface that carries it and the cadence at which it ages.

## Boundary with absorption discipline

[`docs/absorption-discipline.md`](absorption-discipline.md) carries the doctrine governing how material moves between surfaces. This doc references it as the operative mechanism for changing a claim's surface assignment.

Specifically:

- Absorption is the move from one surface to another when material has earned a more durable or more general home. The surface split above defines the *destinations*; absorption discipline governs the *movement*.
- Material earns a new surface only when it has earned that aging rate and that level of generality. Aging-rate fit alone is insufficient if the claim has not earned slow-surface durability. Generality alone is insufficient if the claim has not been pressure-tested.
- n=1 stays observation-level unless the structural argument is strong enough to justify generalization before repetition. The default is restraint.
- Absorption can move material upward (toward more durable or more general surfaces), downward (out of a surface where it no longer fits, into operator-side or scratch), or nowhere (held material that has not earned movement either way).

The surface split is a topology; absorption discipline is the routing rule that respects it. A surface assignment without an absorption discipline produces drift; an absorption discipline without a surface split has nowhere to route material to.

## Worked pressure surfaces

Method altitude mentions only. Domain detail lives in the project repos and in the project examples (`examples/`).

- **[`asset-pipeline-ASK`](https://github.com/apexSolarKiss/asset-pipeline-ASK)** surfaced the aging-rate principle and the source-of-truth split under mature project pressure. The friction of mixing rules, state, and context in a single doc was the original surfacing event. The doctrine that resulted — repo carries project state, `AGENTS.md` carries workflow rules, grounding note carries source-of-intent, current task context carries per-conversation tracking — generalized cleanly upstream into [`control-surface`](https://github.com/apexSolarKiss/control-surface) workflow templates and into [`docs/method.md`](method.md#disciplines-during-execution)'s §Disciplines During Execution treatment.
- **[`urban-observatory`](https://github.com/apexSolarKiss/urban-observatory)** surfaced source-of-intent recovery, the post-bootstrap grounding-note trim pass, and state-vs-context routing. The grounding-note layer-classification gate (now in [`docs/source-of-intent.md`](source-of-intent.md#grounding-note-refresh-preflight) §Grounding-note refresh preflight) was the structural absorption of this pressure. State / event records route to scratch even when the event was important.
- **[`design-system-ASK`](https://github.com/apexSolarKiss/design-system-ASK)** surfaced the presence-vs-structure verification observation and the freshly-built-reference-implementation-vs-proven-inheritance distinction. The presence-vs-structure observation is currently carried at observation-level; n=1 across child projects. Promotion to rule-level treatment in this doc, in [`docs/absorption-discipline.md`](absorption-discipline.md), or in `control-surface` templates waits for a second child project surfacing the same failure mode. The reference-implementation-vs-proof distinction is captured in [`examples/design-system-ASK.md`](../examples/design-system-ASK.md) as the worked-example framing.

## Practical rule

Before writing or updating any artifact, ask in order:

- What kind of claim is this? (project state / workflow rule / source of intent / event record / cross-project doctrine / execution protocol / reflection / task-context ephemera)
- Which surface owns this claim? (repo / `AGENTS.md` / grounding note / scratch / `method-ASK` doc / `control-surface` / article / current task context)
- How fast will it age? (slow / own pace / fast / ephemeral)
- What would become stale if placed here? (the test for wrong-surface fit)
- What would be lost if it stays where it is? (the test for under-absorption)
- What later evidence would move it? (the explicit promotion / demotion condition)

The first three questions establish surface fit. The fourth surfaces the staleness risk. The fifth surfaces the under-absorption risk. The sixth records the trigger that would warrant a future move, so the assignment is auditable rather than presumed permanent.

The rule is a tripwire, not a ceremony. Most material answers the first three questions trivially and routes without friction. The rule earns its keep on the cases where the answers disagree.

## Self-superseding clause

This doctrine should be superseded by:

- [`docs/relative-externality.md`](relative-externality.md) if the surface-topology pieces (the core split, the boundary with adjacent surfaces, the recursive externality relationships) earn a more focused home there and this doc compresses toward the aging-rate and authority-assignment substrate
- [`docs/voice-surface-boundary.md`](voice-surface-boundary.md) if the doctrine's failure-mode treatment for voice-bearing surfaces (article-as-spec confusion, summary-as-authority drift) earns specific treatment in that doc
- compression into [`docs/absorption-discipline.md`](absorption-discipline.md) if the surface-split topology folds into absorption-discipline's §Layer destinations as the routing surface, and the aging-rate principle compresses into a shorter cross-reference
- a future `docs/layer-classification.md` if multiple downstream projects produce enough surface-routing decisions to justify a more formal central routing doc that absorbs both this doc's surface split and the source-of-intent grounding-note refresh preflight

The doctrine is not finished. The nine-surface split is the current best articulation of the method's lived surface topology. As surfaces shift — articles becoming codified differently, new repos earning their own classes, executor tooling changing what a session holds — the split will refine. The aging-rate principle itself is more stable than the specific surface list; the surfaces are the substrate, the aging-rate is the doctrine.

Anchor reading lives in [`docs/articles.md`](articles.md). The articles most directly substrating this doctrine are *Lessons from the First Prototype Phase* (the structural retrospective; the original surfacing of aging-rate and source-of-truth as load-bearing principles), *The Rule the Repo Could Not Contain* (externality as a relation, not a property; surfaces carry the structural category but not always the payload), and *Three Agents Got Into an Argument // The Repo Won* (repo-as-arbiter for current-state questions; the operating rule that splits chat-side advisors from the artifact-owning surface).