diff --git a/CLAUDE.md b/CLAUDE.md index db88bd8..1aa57fd 100644 --- a/CLAUDE.md +++ b/CLAUDE.md @@ -6,8 +6,8 @@ Guidance for agents working in the `decision-driven-design` repository. **This repository is the software projection (split per `DDD-dec-04`).** The actor-general theory — the numbered core documents and the canonical term graph — is **canon in the principle -repository** (`actor-indexed-determination`, currently pinned at tag `v5.7.0` — `graph/upstream.yaml` -is authoritative), not here. Do not add or edit core +repository** (`actor-indexed-determination`, pinned at the tag named in `graph/upstream.yaml`, +which is authoritative), not here. Do not add or edit core theory documents in this repo; canon changes are issues and PRs against the principle repo. This repo holds the software apparatus, the SDLC and organisation-design projections, the graph tool's claims (`core/claims/`: `org`, `tool`, `sim`), and the program's decisions diff --git a/README.md b/README.md index fb5d543..349553b 100644 --- a/README.md +++ b/README.md @@ -13,8 +13,9 @@ determinations come from, what they cost, and which actor should make each one. The framework is a **claim graph**: every claim it makes is a node with a status, evidence, and a falsifier, stored as data and validated against a versioned schema. It remains deliberately **smaller -and better-attributed** than v3, and harder to knock down — and says exactly which of its claims are -proven, exercised, or still projected. +and better-attributed** than v3 — and says exactly which of its claims are `established` (argued +internally and unchallenged; four claims, all formal — not external validation), `reported` (at +least one reproducing computation), or still `projected` (a declared falsifier nobody has met). --- @@ -22,30 +23,34 @@ proven, exercised, or still projected. Four classical results govern how work is allocated in engineered systems — Brooks's essential complexity, Tesler's conservation of complexity, Ashby's requisite variety, Meyer's contracts. Each -quantifies over an **actor**: the thing that makes a determination against some ground. **None of them -makes that actor explicit.** They had no reason to — for the whole history of these results there -were two kinds of determiner, a person or a program, and the gap between them was a light switch, not -a spectrum. +quantifies over an **actor**: the thing that makes a determination against some ground. **Each of +them holds that actor largely fixed.** They had little reason to vary it — for the whole history of +these results there were two kinds of determiner, a person or a program, and the gap between them +was a light switch, not a spectrum. A third kind now exists: **non-deterministic, yet with a distribution that can be frozen by binding.** -Decision-Driven Design is what you get when you **fill in the actor slot** those results left empty — -and discover that supplying the missing parameter *changes their predictions.* +Decision-Driven Design is what you get when you **make the actor an explicit parameter** those +results hold largely fixed. The framework indexes their allocation questions to the arrangement; its +contribution is stated against those literatures, not against their absence. -Two consequences follow, and they are the framework's core contribution: +Two claims follow, and they are the framework's core contribution — both **projected**: argued, +with declared falsifiers nobody has met: 1. **The irreducible floor of a task is a property of its *acceptance predicate*, not of the task.** Zero where you can check the answer; non-zero where you cannot; and *whether you can* is, in - general, undecidable. → [`core/03-the-floor.md`](core/03-the-floor.md) + general, undecidable. *(`DDD-floor-02`, **projected**)* → [`core/03-the-floor.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/core/03-the-floor.md) 2. **Selection intensity is inversely proportional to acceptance-predicate closure.** *Training* is what you do when you can check the work. *Selection* is what you do when you cannot — you check the - worker instead. This is falsifiable across professions. - → [`core/04-actors.md`](core/04-actors.md) + worker instead. This is falsifiable across professions. *(`DDD-hyp-05`, **projected**)* + → [`core/04-actors.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/core/04-actors.md) -And, new in 4.1: for tasks whose acceptance predicate closes, **specification demand is measurable** — -it is the Shannon entropy of the verdict, and conservation is the chain rule of entropy -([`core/09`](core/09-the-measure.md)). The measure exists exactly where the predicate closes, and -vanishes precisely at the floor. +And, new in 4.1: a measure for the closing region. The chain-rule identity is arithmetic +(`DDD-measure-02`, **established**, formal); the identification of specification demand with the +Shannon entropy of the verdict is a modelling claim (`DDD-measure-01`, **projected**) — the +identity holding is not evidence for the identification +([`core/09`](https://github.com/Hafeok/actor-indexed-determination/blob/main/core/09-the-measure.md)). The measure is defined exactly where the predicate closes, +and is undefined at the floor. And a prediction: **model actors outperform human actors exactly where the acceptance predicate closes, and underperform exactly where it does not — the gap tracks *closure*, not *difficulty*.** @@ -61,9 +66,9 @@ allocation lens (four stores: encoded, mechanically checked, judged, escaped), a **It is not** a new physical law. The conservation claim is **Tesler's Law of Conservation of Complexity, generalised** — denominated in decisions, with a fourth store (the *escaped* one Tesler lacked) and an assurance-level bound. It has **no measurable unit**, so it is a **principle**, not a -law, and the repository says so throughout. See [`core/01-the-principle.md`](core/01-the-principle.md) +law, and the repository says so throughout. See [`core/01-the-principle.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/core/01-the-principle.md) and, for the full record of what was corrected and why, -[`meta/lineage-and-limits.md`](meta/lineage-and-limits.md). +[`meta/lineage-and-limits.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/meta/lineage-and-limits.md). We publish the review and the retreats as first-class documents. A framework that states what it cannot support is worth more than one that overclaims. @@ -77,13 +82,14 @@ cannot support is worth more than one that overclaims. The actor-general theory — `00` primitives through `13` delivery, the canonical term graph, and the reproduction assets — is **canon in the principle repository**, [`actor-indexed-determination`](https://github.com/Hafeok/actor-indexed-determination), at the tag -pinned in [`graph/upstream.yaml`](graph/upstream.yaml) (`v5.5.0` at time of writing). +pinned in [`graph/upstream.yaml`](graph/upstream.yaml), which is authoritative for the version. It is not duplicated here. This repository pins the specific claims it depends on in [`graph/upstream.yaml`](graph/upstream.yaml); read the theory in the principle repo's `core/`. -The load-bearing, falsifiable results — the floor is in the acceptance predicate, selection intensity -tracks predicate closure, demand is the Shannon entropy of the verdict, overflow ∩ open is the -mechanism of capacity-generated escape — all live upstream. Falsification of any of them is an issue against the principle repo, not this one. +The load-bearing, falsifiable claims — the floor is in the acceptance predicate (`DDD-floor-02`, +projected), selection intensity tracks predicate closure (`DDD-hyp-05`, projected), demand is the +Shannon entropy of the verdict (`DDD-measure-01`, projected), overflow ∩ open is the mechanism of +capacity-generated escape (`DDD-floor-01`, reported) — all live upstream. Falsification of any of them is an issue against the principle repo, not this one. ### Apparatus — the mechanisms @@ -105,7 +111,7 @@ The framework, denominated in a domain. | Document | Domain | |---|---| -| [`applications/sdlc/`](applications/sdlc/) | Software delivery — the agentic/DAG design framework (formerly the whole of v3) | +| [`applications/sdlc/`](applications/sdlc/) | Software delivery. What the tree carries today: the section README and `production-as-ground.md`. The four v3 design documents live in this repository's git history, not in the tree — see the section README for what they were and where they sit | | [`applications/sdlc/production-as-ground.md`](applications/sdlc/production-as-ground.md) | Production is the only real ground; DORA read as demand; which feedback loops are waste | ### The graph — claims and decisions as data @@ -118,6 +124,7 @@ The framework as nodes. Files are storage; the graph is the object. | [`core/decisions/`](core/decisions/) | Decision nodes (`DDD-dec-NN`); the load-bearing edge is `decision --basedOn--> claim` | | [`spec/claim-format.md`](spec/claim-format.md) | The claim schema (format version 1) and its validation rules | | [`scripts/validate-claims.py`](scripts/validate-claims.py) | Validates `core/claims/` and `core/decisions/` against the spec | +| [`validate-core-order.py`](validate-core-order.py) | Validates docs, embeds and the cross-repo pins. **Canonical invocation: `python3 validate-core-order.py core/`** — run against the repository root (`.`) it misreads this repository's own registry ids and reports spurious E13 errors on a clean tree; a fix for that mode is tracked, the `core/` invocation is authoritative | ### Meta — the honesty layer and the program @@ -137,11 +144,14 @@ Danish glossary of core terms: in the principle repo (`i18n/ordliste-dansk.md`). ## What changed from v3 v3 was the agentic-design framework: decisions as the unit of work, a DAG of roles, artifacts with -schemas, backed by [`product-cli`](https://github.com/Hafeok/product-cli). **That framework is -intact** — it now lives in [`applications/sdlc/`](applications/sdlc/) as the **engineering -projection** of the general principle, which is exactly what it always was. What v4 adds is the layer -*beneath* it: the theory that explains why the DAG design works, and the actor model that says which -node each determination belongs to. +schemas, backed by [`product-cli`](https://github.com/Hafeok/product-cli). Its reframing as the +**engineering projection** of the general principle stands — that is what it always was. Its four +design documents, however, are **not in this tree**: they were removed at the v4 restructuring and +survive in git history; restoring them is an open decision that has not been taken. What the +projection carries in the tree today is [`applications/sdlc/`](applications/sdlc/)'s README, +`production-as-ground.md`, and the reference implementation. What v4 adds is the layer *beneath* +it: the theory that explains why the DAG design works, and the actor model that says which node +each determination belongs to. The register also changed. Following external review: @@ -149,14 +159,14 @@ The register also changed. Following external review: Ashby did, and even he refused the term). - **Conservation** holds as an accounting identity *within a fixed decomposition* — re-decomposing *relocates* demand into the seam (a cleaner split pre-pays more into the interface contract; the - total is invariant), so the decomposition is the highest-leverage decision. → [`core/09`](core/09-the-measure.md) §4, `DDD-measure-03`. + total is invariant), so the decomposition is the highest-leverage decision. → [`core/09`](https://github.com/Hafeok/actor-indexed-determination/blob/main/core/09-the-measure.md) §4, `DDD-measure-03`. - **The immune-system "licensing" argument** is demoted to a suggestive parallel with known disanalogies; **CRISPR** is the accurate compound-platform instance. - **The zero-floor postulate** is retreated to **the floor-in-the-predicate** result, which is sharper and survives the theoretical limits (Rice, inevitable model error, collective tacit knowledge). -Full record: [`meta/lineage-and-limits.md`](meta/lineage-and-limits.md). +Full record: [`meta/lineage-and-limits.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/meta/lineage-and-limits.md). --- @@ -166,10 +176,11 @@ This is a working specification under active revision. The strongest claims — predicate, and *selection intensity tracks predicate closure* — are **falsifiable**, and the conditions are stated in the documents. Live status is now in the graph: every claim carries a status (`established` / `reported` / `projected` / `retired`) in [`core/claims/`](core/claims/). -The counting-procedure debt — a measure of governing-decision demand shown invariant — is -**paid for closing predicates** by the measure (`core/09`, `DDD-measure-01`/`02`/`06`) and marked -as a **boundary, not an open debt**, off them; the framework books its remaining debts openly -([`meta/`](meta/)). +The counting-procedure debt — a measure of governing-decision demand — is **addressed for closing +predicates** by the measure: the conservation identity is arithmetic (`DDD-measure-02`, +established) and the identification of demand with verdict entropy is a modelling claim +(`DDD-measure-01`, **projected**); off closing predicates it is marked as a **boundary, not an +open debt**. The framework books its remaining debts openly ([`meta/`](meta/)). Reference implementation: [`product-cli`](https://github.com/Hafeok/product-cli) (the authoring layer for the SDLC projection). @@ -179,7 +190,7 @@ for the SDLC projection). Tesler (conservation of complexity) · Ashby (requisite variety) · Brooks (essential complexity) · Meyer & Hoare (contracts) · Saltzer, Reed & Clark (end-to-end) · Kalman (observability) · Polanyi & Collins (tacit knowledge) · Rice (undecidability) · Edelman & Gally (degeneracy). Full attribution in -[`meta/lineage-and-limits.md`](meta/lineage-and-limits.md). +[`meta/lineage-and-limits.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/meta/lineage-and-limits.md). ## License diff --git a/applications/sdlc/README.md b/applications/sdlc/README.md index 627052a..ff8f89e 100644 --- a/applications/sdlc/README.md +++ b/applications/sdlc/README.md @@ -38,22 +38,25 @@ generally and more carefully. ## The design documents -These are the v3 documents, retained. Read `core/` for *why*; read these for *how to build it*. +**The four v3 design documents are not in this tree.** They were removed at the v4 restructuring +(commit `219ad6c`, under their former `docs/` paths) and have not been re-landed; restoring them is +an open decision that has not been taken. They are listed here as the record of what the v3 +framework comprised — recoverable from this repository's git history — not as links: -1. **[`01-foundations.md`](01-foundations.md)** — LLMs as knowledge forecasters; work as a chain of +1. **`01-foundations.md`** *(history: `docs/01-foundations.md`)* — LLMs as knowledge forecasters; work as a chain of context-conditioned decisions; value actions as the terminus; roles and artifacts; the DAG, not the pipeline. *Start here for the software framing.* -2. **[`02-entity-reference.md`](02-entity-reference.md)** — the vocabulary made precise: processes, +2. **`02-entity-reference.md`** *(history: `docs/02-entity-reference.md`)* — the vocabulary made precise: processes, decisions, actions, interpretations, roles, artifacts, sessions, schemas, bundles, audits, the orchestration system, per-role autonomy. -3. **[`03-autonomy-levels.md`](03-autonomy-levels.md)** — the 0–5 autonomy ladder, why autonomy is +3. **`03-autonomy-levels.md`** *(history: `docs/03-autonomy-levels.md`)* — the 0–5 autonomy ladder, why autonomy is per-role not per-system, and why this structure is what makes Levels 4–5 reachable. *Now with a sharper foundation: per-role autonomy is exactly the actor model's pinning resolution (`core/04`), and the ceiling on a role's autonomy is set by the closure of its acceptance predicate (`core/03`).* -4. **[`04-implementation.md`](04-implementation.md)** — the build architecture: Rust + Oxigraph, curated +4. **`04-implementation.md`** *(history: `docs/04-implementation.md`)* — the build architecture: Rust + Oxigraph, curated SPARQL for bundle assembly, the event substrate, the worker contract, emergent decisions during action, the meta-loop, the model catalog. @@ -61,11 +64,12 @@ Written against v4: - **production-as-ground** · pre-production predicates close over substitute ground; DORA's four keys instrument the encode/verify gap; the three tiers of feedback loop and which are eliminable -> **Note.** These four documents are carried forward from v3 and are **not yet fully reconciled** with -> the v4 register (they may still say "law" where `core/` now says "principle," and may still lean on -> the pre-review immune framing). Reconciling them is tracked in -> [`../../meta/consolidated-state.md`](../../meta/consolidated-state.md). The `core/` layer wins wherever -> they conflict. +> **Note.** The four historical documents were written before the v4 register (they say "law" +> where `core/` now says "principle," and lean on the pre-review immune framing). If they are ever +> restored to the tree, that reconciliation is the restoring session's work, and the `core/` layer +> wins wherever they conflict. Until then, this README and `production-as-ground.md` are what the +> SDLC projection carries here; the reference implementation is +> [`product-cli`](https://github.com/Hafeok/product-cli). --- diff --git a/core/decisions/DDD-dec-35.yaml b/core/decisions/DDD-dec-35.yaml new file mode 100644 index 0000000..819e3bd --- /dev/null +++ b/core/decisions/DDD-dec-35.yaml @@ -0,0 +1,76 @@ +format: 1 +id: DDD-dec-35 +kind: decision +statement: | + The upstream pin advances v5.12.0 -> v5.13.0, discharging the ground-migration execution's own prediction: exactly two pinned digests move, DDD-measure-01 and DDD-measure-16, both from the W1 rename in their statements, with every id and every hash the one the migration session wrote down before its edits were made. +principal: Emil +made: 2026-09-04, pre-transfer session (GATE 3 ruling to proceed; the advance executed and held for verification at the same gate) +resolution: | + The advance, and the prediction it verifies. + + ## Whose prediction this is + + The ground-migration execution (2026-08-31) edited canon's `ground distribution` occurrences to + `deployment distribution` and predicted, in a committed file before its own edits + (meta/sessions/2026-08-31-ground-migration-exec/gate3-prediction.md §2), what the next downstream + pin advance would fire: exactly two W6 and no more, by id and by hash. The prediction belongs to + "a session that can verify its own prediction", which is the pre-transfer session — this advance. + + ## Predicted before, verified after + + Per the DDD-dec-28/29/34 pattern, restated and committed at + meta/sessions/2026-09-02-pre-transfer/gate3-pin-advance.md before graph/upstream.yaml was + touched, with the ref advanced ALONE first and the hashes re-instrumented second, so the pass is + distinguishable from a skip. + + PREDICTED FIRING (from the migration record, verbatim ids and hashes): + E12/E13 0 + W5 0 — no pinned status moves + W6 exactly 2 — + DDD-measure-01 (statement, 1 occurrence): + sha256:66b01ede... -> sha256:d78274d2... + DDD-measure-16 (statement, 2 occurrences): + sha256:2d763d1a... -> sha256:6828d066... + W7 1, unchanged — the standing shadowed id + and 69 of 71 pins unmoved. + + BASELINE, fresh at ref v5.12.0 before the operation: + 71 pins resolved against the pinned ref, 0 basis-loss, 0 content-drift, 1 shadowed id(s) + + OBSERVED, at ref v5.13.0 with the hashes NOT yet re-instrumented: + 71 pins resolved against the pinned ref, 0 basis-loss, 2 content-drift, 1 shadowed id(s) + warnings: exactly 2, both W6, both ids and all four digests exactly as tabled; 0 errors + + **The prediction held in every limb.** After re-instrumentation: 71 pins resolved, 0 basis-loss, + 0 content-drift, 1 shadowed id; 0 errors, 0 warnings. + + ## Non-interference, predicted as nothing and observed as nothing + + The same session's supersession of DDD-frame-08 (GATE 1 ruling C) lives on the upstream session + branch only; the v5.13.0 tag does not carry it. Its predicted contribution to this advance was + stated as zero beforehand — making the absence checkable — and zero is what fired: + DDD-frame-08's pinned hash verified unchanged at the tag. That supersession's firing belongs to + the next advance, after the pre-transfer session's close proposes its version. + + ## Consequences carried at the same act + + - The primer regenerates at v5.13.0: generated blocks re-drawn, the stamp's pin moves, the + hand-written sections' own pin references move, and §6's "the migration's changes arrive at + the next advance" paragraph updates — this is that advance, as §6 anticipated. + - Paper A's three checkers re-run at the new pin; their reports are recorded as evidence at the + session's verification hold. None of the sixteen deferred manuscript occurrences moves here + (Emil's GATE 3 ruling): a pin advance is not a licence to reopen a deferral, and this session + repairs no manuscript prose. + - The successor item recording that DDD-measure-12 and term:verdict still read + `ground distribution` at v5.12.0 discharges: at v5.13.0 both read `deployment distribution`. + - The downstream README's stale "(v5.5.0 at time of writing)" is repaired by pointing at + graph/upstream.yaml without naming a version (GATE 2 ruling, D7): a stale number recurs, a + pointer does not. +basedOn: + - DDD-dec-29 + - DDD-dec-34 + - DDD-agent-01 +notes: | + Filed from the pre-transfer session (2026-09-02..04; interactive canon curation, Emil ruling at + every gate). No pin added, none removed; no status_at_pin moves. The session record is + meta/sessions/2026-09-02-pre-transfer/gate3-pin-advance.md. diff --git a/graph/upstream.yaml b/graph/upstream.yaml index 2c33826..e203fc3 100644 --- a/graph/upstream.yaml +++ b/graph/upstream.yaml @@ -86,7 +86,21 @@ upstream: # at v5.12.0 and fired NOTHING, because W5 only sees ids that are pinned. Paper A asserted it as # **established** in three places and both of its checkers passed those lines. The gap was found # by a sweep written for inline status assertions, now shipped as papers/paper-a/check-status.py. - ref: v5.12.0 # tag, never a branch + # Advanced v5.12.0 -> v5.13.0 at the pre-transfer session (2026-09-04), on Emil's GATE 3 + # ruling; the advance is DDD-dec-35 here. This advance discharges the ground-migration + # execution's OWN prediction, committed before that session's edits at + # meta/sessions/2026-08-31-ground-migration-exec/gate3-prediction.md: exactly two W6 and no + # more, both from W1's ground->deployment rename — DDD-measure-01 (statement, + # 66b01ede... -> d78274d2...) and DDD-measure-16 (statement, 2d763d1a... -> 6828d066...); + # zero W5, zero E12/E13, the one standing W7 unchanged. Verified at the bump, by advancing the + # ref first and re-instrumenting the hashes second so the firing could be observed rather than + # assumed: 71 pins resolved, 0 basis-loss, EXACTLY TWO content-drift, both ids and all four + # digests as written down beforehand. The prediction held in every limb. After + # re-instrumentation: 71 pins, 0 basis-loss, 0 content-drift, 1 shadowed id. No pins added, + # none removed. The same session's branch-only supersession of DDD-frame-08 (GATE 1 ruling C) + # was predicted to contribute nothing to this advance — the tag does not carry it — and + # contributed nothing; its firing belongs to the next advance. + ref: v5.13.0 # tag, never a branch pins: # basis claims for DDD-dec-04 (the split decision itself rests on these) - id: DDD-frame-01 @@ -94,7 +108,7 @@ upstream: content_hash: sha256:6b74b5bfb34c6c722c9abf1185264ff5c6eee01d020c7fd4ad6804ba7a5bca64 - id: DDD-measure-01 status_at_pin: projected - content_hash: sha256:66b01ede631c0173d86696dca945ade53c9ffe37462bbe290a6df6b6de4b41ea + content_hash: sha256:d78274d2350cb867c9ec3497aaf13caeafb4792f9e20c20697ba54114a58c198 # the floor mechanism the apparatus/skill-floor builds on - id: DDD-floor-01 status_at_pin: reported @@ -339,7 +353,7 @@ upstream: content_hash: sha256:a7ae6e7ad5ccd0afd3bc35c5fd5ce061532b4f99f4d352464f9cc5b78ee01c39 - id: DDD-measure-16 status_at_pin: established - content_hash: sha256:2d763d1adf9e1c68c88d1f5c3e30550a5079e274d84e688c31f5f5bb51f1d94e + content_hash: sha256:6828d06649045bfdf374f985a7984d337d01fdb42be29ef8f051a185411fe674 - id: DDD-measure-17 status_at_pin: projected content_hash: sha256:164a18072145bc4a0cdb3c6e1ca8890fc64c1b36396c73be5ca8bd45c66114e5 diff --git a/meta/sessions/2026-09-02-pre-transfer/bootstrap.md b/meta/sessions/2026-09-02-pre-transfer/bootstrap.md new file mode 100644 index 0000000..81839a5 --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/bootstrap.md @@ -0,0 +1,53 @@ +# Bootstrap — the pre-transfer session + +## Invocation message + +> Read prompt-pre-transfer.md in its entirety — this session follows it exactly, including every +> gate. +> +> This is the last work before the repositories become public. Four items, chosen by one test: does +> this get more expensive, or become unfixable, once someone outside can read it or fork it? The +> accountability arity, which reaches canon; the README over-claim and the front pages generally; +> the pin advance to v5.13.0 with its two predicted firings; and a publication-readiness report. +> +> First act, before anything else: commit this prompt and bootstrap to +> meta/sessions/2026-09-02-pre-transfer/ in decision-driven-design, per DDD-dec-20. +> +> Fetch both repos at head — actor-indexed-determination at v5.13.0 (verify and report the tag +> found) and decision-driven-design at head. +> +> Rules that override anything you might infer: +> +> - Interactive canon curation. Stop at every gate for Emil's ruling. Merge nothing. +> - Supersession, never rewriting (DDD-dec-09, DDD-dec-10, DDD-dec-15). +> - I-1: report both texts verbatim before proposing anything, and propose three dispositions +> rather than assuming one. A projected claim adding an element to a settled term is a +> supersession question. Paper A's survey found four of the five elements are already Bovens's — +> whatever files must not claim more than that. +> - I-2: report all findings before any repair, and repair over-claims rather than rewriting the +> READMEs. A front page may say what the framework is for; it may not say what it has shown +> unless a status carries it. +> - I-3: state the prediction before the operation, advance the ref first and re-instrument hashes +> second so a pass is distinguishable from a skip, verify after, record divergence rather than +> reconciling it. The primer regenerates at the new pin and its stamp updates. +> - I-4 is a report, not a repair pass. Emil rules what files now and what carries. +> - The test is publication, not tidiness. Untidy waits; unfixable-once-public lands now. Where the +> two are hard to tell apart, report and let Emil rule. +> - Commit drafts before reporting at each gate, bodies marked draft-pending-ruling. +> +> Begin with I-1 and end your first report at GATE 1: DDD-frame-08 and term:accountability +> verbatim, every node depending on either, and the three dispositions with their costs. + +## Parameters + +- **Session type:** interactive canon curation; hold at every gate; merge nothing; supersession, + never rewriting. +- **Branch (both repositories):** `claude/pre-transfer-publication-readiness-dh0dqo` +- **Base commits:** + - `decision-driven-design`: `46d581a` (head of default branch at session start; the primer + merged, PR #34) + - `actor-indexed-determination`: `5c7fe46`, tag `v5.13.0` verified — the tag points exactly at + the head of the default branch. +- **Gate count:** 5 (I-1 arity texts and dispositions; I-2 front-page findings; I-3 pin-advance + prediction then verification; I-4 readiness report; close). +- **Principal:** Emil. diff --git a/meta/sessions/2026-09-02-pre-transfer/gate1-arity.md b/meta/sessions/2026-09-02-pre-transfer/gate1-arity.md new file mode 100644 index 0000000..80366b7 --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/gate1-arity.md @@ -0,0 +1,200 @@ +# GATE 1 — the accountability arity (I-1) + +**Status: draft-pending-ruling.** Ruled at Paper A's Gate 3 as both-not-either; the paper flagged, +this is the canon session that owes the ruling. Discharges item 1 of +`meta/successor-items-paper-a-revision.md`. + +Base verified: upstream `actor-indexed-determination` at `5c7fe46`, tag `v5.13.0` pointing exactly +at the head of the default branch. Downstream `decision-driven-design` at `46d581a` (the primer +merged). + +--- + +## 1. Both texts, verbatim + +`DDD-frame-08` (`core/claims/DDD-frame-08.yaml`, upstream, **projected**, kind normative, owner +paper-4, changed v4.5) — the statement field: + +> Accountability is a relation (attribution, persistent principal, authority linkage, stake, +> sanction path), not an intrinsic capacity; an arrangement naming an executor but no principal is +> incomplete. + +`term:accountability` (`core/graph/terms.yaml`, upstream, **settled**, established by +`05-accountability.md`) — the canonical_md field: + +> **Accountability** is a property of the arrangement, not of the executor: attribution of +> the determination, a persistent answerable party, and a borne consequence. An arrangement +> missing any of the three has not allocated the decision's consequence. + +**A third enumeration exists in canon prose**, and the gate should see it. `core/05-accountability.md` +§2, the bolded sentence immediately above the term's embed — the sentence `DDD-frame-08`'s own +notes cite as its verification: + +> **Accountability completeness is a property of an arrangement, not a capacity of an actor. An +> execution is accountability-complete when it is linked to a persistent responsible principal, an +> attributable record, a defined stake, and an enforceable consequence path.** + +Four elements: it already carries the split (stake and consequence path separate) and does not +carry authority linkage. Canon therefore counts the relation three ways — three (term, settled), +four (core/05 §2 prose, exposition), five (claim, projected). Authority linkage's only canonical +carriage is `DDD-frame-08` and the statement of `DDD-hyp-04`; the nearest prose support is core/05 +§2's "sanctionability depends on an authority with enforceable jurisdiction — all properties of a +relationship among executor, principal, record, and sanctioning authority". + +## 2. Every node depending on either + +### Upstream (`actor-indexed-determination`, v5.13.0) + +| Node | Dependency | +|---|---| +| `core/claims/DDD-frame-08.yaml` | the claim itself | +| `core/graph/terms.yaml` → `term:accountability` | the term itself | +| `core/claims/DDD-hyp-04.yaml` | its **statement** enumerates all five elements ("attribution, persistent principal, authority linkage, stake, sanction path"); its notes cite frame-08 as conceptual basis per a GATE 1 ruling — the prediction is "frame-08's test made graded and preregistered" | +| `core/05-accountability.md` | establishes and embeds `term:accountability` (`ddd:embed`, E6 byte-match enforced); §2 carries the four-element bold sentence frame-08's notes cite | + +No other upstream claim or decision references either id. `DDD-dec-27` (recorded downstream in the +pin comment) is provenance for frame-08's pin addition, not a live dependent. + +### Downstream (`decision-driven-design`, head) + +| Node | Dependency | +|---|---| +| `graph/upstream.yaml` | pins both: `DDD-frame-08` (projected, `content_hash sha256:cf73b307…`) and `term:accountability` (settled, `sha256:a986dc63…`). Hash covers statement + region + canonical_md (`validate-core-order.py`), so **any change to either text fires the pin** | +| `projections/tracks/01-determination.md` | rung 4 (line ~397): prose enumerates the five elements with `ddd:ref id=DDD-frame-08`; rung 10 (lines ~913–920): quotes the term's canonical text in full and frame-08's statement in paraphrase; status table rows at lines 50 and 64 | +| `papers/paper-a/paper-a.md` | §7 quotes frame-08's statement verbatim and reproduces the term's three-element enumeration; carries the mapping table (refines-and-adds); further uses at lines ~1263, 1292, 1430, 1515–1537, 1566–1567 (related-work table: "four of five accountability elements" Bovens's; "authority linkage; the relation as a design-time property" the framework's), 1629 | +| `papers/paper-a/paper-a-supplement.md` | line ~66 (the mapping, five vs three); line ~189 (frame-08's statement verbatim, appendix table); line ~230 (the term's canonical_md verbatim, appendix table) | +| `meta/successor-items-paper-a-revision.md` item 1 | the ruling this gate owes; discharges here | +| `meta/seed/claims-seed.yaml` | historical seed copy of frame-08 (record, not live) | +| session records (`2026-08-18-wave3`, `2026-08-20-paper-a`, `2026-08-25-item4`, `2026-08-30-paper-a-revision`), ground-audit files | historical mentions; records do not move | + +### The survey constraint that binds every disposition + +Paper A's related-work survey (gate2-survey-notes, carried into the paper's related-work table): +**four of `DDD-frame-08`'s five elements are already Bovens's** (the actor–forum relation: inform, +interrogate, judge-and-sanction; the problem of many hands). What is not Bovens's is **authority +linkage**, plus the change of tense — retrospective assessment of conduct → design-time +constitution of an arrangement. Whatever files at this gate must not claim more than that, and +currently `DDD-frame-08`'s notes say nothing about lineage at all. + +## 3. The three dispositions, with costs + +### A — the term supersedes to carry five + +`term:accountability`'s canonical text moves to the five-element relation; the three-element text is +preserved as the superseded version with the correction recorded (supersession, never rewriting — +DDD-dec-09/10/15 pattern). + +Costs: + +- **A settled node carries projected content.** Authority linkage's evidence grade is a projected + claim and an unrun hypothesis. Putting it into the settled term either over-claims — the defect + class this session exists to prevent — or forces the term's status down, and a settled term five + documents pin or embed is the expensive kind to demote. +- **Maximum firing surface.** canonical_md is hashed: the downstream pin fires; `core/05`'s embed + must re-project (E6); the paper's §7 quote and the supplement's appendix table (line 230) go + stale against a *settled* node — the worst kind of staleness for a manuscript about to be public; + the track's rung 10 verbatim quote regenerates. +- **The Bovens constraint bites hardest here.** The three-element term is materially Bovens's own + decomposition; a five-element settled term moves the non-Bovens element into the framework's + canonical definition at settled grade. +- **Sequencing.** Lands after v5.13.0, so it needs the close's version proposal (v5.14.0) and a + second downstream pin advance after I-3's — I-3's advance to v5.13.0 stays clean either way. + +### B — the claim narrows to the term's three + +`DDD-frame-08`'s statement supersedes to the three-element relation. + +Costs: + +- **Two claims move, not one.** `DDD-hyp-04`'s statement enumerates the five; narrowing frame-08 + alone relocates the disagreement rather than resolving it. Both are projected, so the status cost + is low — this is the ordinary correction loop. +- **Canon gives up ground its own prose argues for.** The split (stake ≠ sanction path) is already + in core/05 §2's bolded P1 sentence at four elements, and authority linkage is argued in core/05's + exposition ("an authority with enforceable jurisdiction"). Narrowing to three discards design + information the successor item itself called defensible. +- **Paper A's delta over Bovens halves.** The paper's stated contribution beyond Bovens is exactly + authority linkage plus the design-time tense. Drop authority linkage from canon and half the + stated delta has no canonical carrier — the related-work table, the mapping table, §7's quote, + the supplement's appendix, and the track's rung 4 all rewrite. The claim's pin fires (statement + is hashed). + +### C — reconciled as different objects, the distinction stated in both + +The term defines **what accountability is** — the relation that allocates a decision's consequence, +three parts at the definition's grain. The claim asserts **what a complete instance requires at +design time** — five conditions: the third part refined into stake plus sanction path (the split +core/05 §2 already carries), and authority linkage added as a completeness condition. Five is a +completeness test of the three-part relation, not a competing definition. The distinction lands in +both texts: + +- **Claim side:** frame-08's statement gains the scoping words (e.g. "a **complete** accountability + arrangement links…" / statement recast as completeness conditions of the relation the term + defines); its notes record the ruling, the reason the term deliberately does not carry authority + linkage, and the Bovens lineage (four of five his; authority linkage and the design-time tense + not). The statement change fires the claim's downstream pin — a projected claim, the cheap kind. +- **Term side:** the term registry schema carries no `notes` field today (no entry uses one), so + the term-side statement of the distinction goes in `core/05-accountability.md`'s prose beside the + embed — exposition, which fires nothing — unless Emil prefers a schema addition (additive, like + format 2's `canonical_home`), which is a spec question this session should not decide alone. + +Costs: + +- **The two counts remain visible** to an outside reader; what changes is that both texts now say + which governs at which grain. This is the residual cost — smaller than A's or B's, but not zero: + it requires the reader to accept a definition/completeness distinction rather than seeing one + number. +- **The smallest firing surface.** No settled node moves; no E6 re-embed; the paper's mapping table + stays exactly right (it already answers "refines-and-adds" — this ruling makes the addition + licensed and states why); `DDD-hyp-04` is untouched (it already predicts on completeness). + Frame-08's pin fires; rung 4's prose and the supplement's line-189 statement update to the new + statement text. +- **core/05 §2's four-element bold sentence needs a disposition of its own** under any of A/B/C: it + is exposition carrying a fourth count. Under C it reads naturally as the completeness statement + minus authority linkage — a prose defect to flag in frame-08's `notes:` per the flag-don't-guess + rule, or to repair in the prose as part of stating the distinction beside the embed. + +## 4. What every disposition carries regardless + +1. **The Bovens lineage lands in canon**, not only in the paper: frame-08's notes (and the term's + establishing prose, under C) must state that four of the five elements are Bovens's and that the + framework's additions are authority linkage and the design-time tense — no file may claim more. +2. **The supersession record**, whichever node moves, per DDD-dec-09/10/15: the superseded text + stays in the graph with the correction that moved it. +3. **Sequencing with I-3:** the disposition lands upstream after v5.13.0, so it rides the close's + version proposal; I-3's pin advance to v5.13.0 is unaffected and its two predicted W1 firings + stay exactly as predicted. + +**HOLD — awaiting Emil's ruling on A / B / C.** + +--- + +## RULING (Emil, 2026-09-03): C — and not primarily on cost + +- **A refused on principle:** it moves content whose evidence grade is projected (one projected + claim, one unrun hypothesis) into a settled node — the over-claim defect this session exists to + prevent; cost does not enter into it. +- **B refused** because it would make canon smaller than the corpus supporting it: core/05 argues + authority linkage in its exposition, and B narrows the claim to three while the prose continues + to argue for more — resolving one inconsistency by creating another, and stranding half of Paper + A's stated delta with no canonical carrier on the eve of publication. +- **C is right because the two objects genuinely differ, and there is a test:** an arrangement can + instantiate the relation and still be incomplete, and it is the ordinary case — attribution + exists, a persistent principal exists, consequence is borne, and no authority has jurisdiction to + enforce it. Everyone knows who is responsible; nobody can sanction them. +- **The convergence:** the distinction C draws *is* the delta over Bovens — retrospective + assessment versus design-time constitution. The definition/completeness split is not a device to + make two counts coexist; it is the same distinction that makes the contribution non-derivative. + C carries the contribution; B discards it. + +Three required consequences: (1) core/05's four-element sentence is the completeness statement and +must carry five, attributed to `DDD-frame-08`, marked projected, distinct from the definition in +the embed — both grades visible in one document; (2) frame-08's verification was under-supported — +a four-element sentence cited as verification for five — and must cite the pair (that sentence plus +the sanctionability-and-jurisdiction line); (3) frame-08's notes take the lineage in full — four of +five elements are Bovens's, the delta is authority linkage plus the change from retrospective +assessment to design-time constitution, half of it a tense change and stated as such. No schema +field: prose beside the embed suffices; reconsider only on a second case. + +**Executed:** upstream commit `b0d603f` (`DDD-frame-08` re-scoped, `core/05` §2 restated; all three +validators green, warning count unchanged at 32); successor item 1 marked discharged. diff --git a/meta/sessions/2026-09-02-pre-transfer/gate2-front-pages.md b/meta/sessions/2026-09-02-pre-transfer/gate2-front-pages.md new file mode 100644 index 0000000..ee331ce --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/gate2-front-pages.md @@ -0,0 +1,226 @@ +# GATE 2 — the front pages read as a stranger would (I-2) + +**Status: draft-pending-ruling. Findings only; nothing repaired.** + +Method: both READMEs read in full against the graph at `v5.13.0` (upstream) and head (downstream); +every sentence tested against the primer §6 register — *a front page may say what the framework is +for; it may not say what it has shown unless a status carries it* — and against Paper A's +related-work survey. Line numbers are against the current files. + +The status facts the findings rest on, fetched fresh: + +| Assertion on a front page | Nearest carrying node | Kind | Status | +|---|---|---|---| +| the floor is in the acceptance predicate | `DDD-floor-02` | conceptual | **projected** | +| selection intensity tracks predicate closure | `DDD-hyp-05` | empirical | **projected** | +| demand is the Shannon entropy of the verdict (the identification) | `DDD-measure-01` | empirical | **projected** | +| conservation is the chain rule (the arithmetic) | `DDD-measure-02` | formal | established | +| overflow ∩ open is the mechanism of capacity-generated escape | `DDD-floor-01` | formal | reported (toy actors) | +| the hallucination account | none — exposition in `core/11` only | — | — | +| the immune-system instance | none — no claim carries it | — | — | +| the classical results' predictions change | none | — | — | + +`spec/claim-format.md` §5: `established` is four claims, all `formal`; nothing empirical, +conceptual or normative has ever reached it; external validation exists nowhere. + +--- + +## Downstream README (`decision-driven-design/README.md`) + +**D1 — lines 30–32, the named over-claim, confirmed and wider than one line.** + +> A third kind now exists: **non-deterministic, yet with a distribution that can be frozen by +> binding.** Decision-Driven Design is what you get when you **fill in the actor slot** those +> results left empty — and discover that supplying the missing parameter *changes their +> predictions.* + +Two defects in one sentence. (a) "the actor slot those results left empty" and "the missing +parameter" are the absence claim the paper's retitle withdrew — *The Missing Parameter* was dropped +because it "claims absence from prior work the paper has not surveyed", and the survey then found +the neighbours (Hollnagel & Woods hold the arrangement-as-unit priority; Horvitz supplies an +allocation criterion the framework lacks). The paper's ruled formulation is "holding the determiner +**largely** fixed", with the contribution "stated against them rather than against their absence" +(abstract; §1 "usually held fixed"). (b) "changes their predictions" — no claim at any status +states that Brooks's, Tesler's, Ashby's or Meyer's predictions change; §11 says the opposite +discipline: "It does not claim their results as instances of its own." + +**D2 — lines 25–26, the absolute absence claim.** + +> **None of them makes that actor explicit.** + +Scoped to the four named results, so it is the nearly-defensible form the retitle analysis +identified — but it is stated absolutely where the paper's ruled wording is "largely fixed" / +"usually held fixed". The survey supports the weaker form. + +**D3 — lines 34–43, projected claims presented as consequences that follow.** + +> Two consequences follow, and they are the framework's core contribution: +> 1. **The irreducible floor of a task is a property of its *acceptance predicate*, not of the +> task.** Zero where you can check the answer; non-zero where you cannot; and *whether you can* +> is, in general, undecidable. +> 2. **Selection intensity is inversely proportional to acceptance-predicate closure.** […] This is +> falsifiable across professions. + +Consequence 1 is `DDD-floor-02`, conceptual, **projected**. Consequence 2 is `DDD-hyp-05`, +empirical, **projected**. "Falsifiable across professions" is honest; "consequences follow" with no +status mark is not — a stranger reads two shown results. + +**D4 — lines 45–48, the projected identification asserted flat.** + +> for tasks whose acceptance predicate closes, **specification demand is measurable** — it is the +> Shannon entropy of the verdict, and conservation is the chain rule of entropy + +The chain rule is `DDD-measure-02` (formal, established — arithmetic). The identification of the +framework's demand with verdict entropy is `DDD-measure-01` — empirical, **projected**. The +sentence asserts the identification as fact and leans it on the arithmetic: the identity-as-evidence +move canon forbids, on the front page. + +**D5 — lines 14–17, "proven" and "harder to knock down".** + +> It remains deliberately **smaller and better-attributed** than v3, and harder to knock down — and +> says exactly which of its claims are proven, exercised, or still projected. + +"Proven" is not in the status vocabulary and describes nothing in the graph — §5 says `established` +is internal warrant, four formal claims, no external validation. "Harder to knock down" asserts +robustness no status carries. This is the sentence most likely to be quoted against the framework +verbatim. + +**D6 — lines 84–86, "results" over four claims of which three are projected.** + +> The load-bearing, falsifiable results — the floor is in the acceptance predicate, selection +> intensity tracks predicate closure, demand is the Shannon entropy of the verdict, overflow ∩ open +> is the mechanism of capacity-generated escape — all live upstream. + +`projected`, `projected`, `projected`, `reported` (toy actors). "Falsifiable" is right; "results" +is not carried. + +**D7 — line 80, stale front-page fact.** + +> at the tag pinned in `graph/upstream.yaml` (`v5.5.0` at time of writing) + +Seven releases stale (pin is v5.12.0 now; v5.13.0 after I-3). A wrong version on the front door of +a public repo, beside the file that is authoritative. + +**D8 — lines 169–172, the debt declared paid on a projected identification.** + +> The counting-procedure debt — a measure of governing-decision demand **shown invariant** — is +> **paid** for closing predicates by the measure (`core/09`, `DDD-measure-01`/`02`/`06`) + +"Shown invariant" holds of the arithmetic (`DDD-measure-02`, established) only through the +identification (`DDD-measure-01`, projected). "Paid" asserts discharge the projected claim cannot +carry. + +## Upstream README (`actor-indexed-determination/README.md`) + +**U1 — lines 105–121, "the central result" is a projected claim.** + +> ## The central result: check the work or check the worker +> The framework's most important result concerns the **acceptance predicate** […] +> **The floor is non-zero exactly when, and because, you cannot check the work.** + +Carried at `DDD-floor-02` — conceptual, **projected**. "Result" twice, no status mark anywhere in +the section. The definitional content around it (what closure means) is the framework's own and +fine; "result" is the over-read. + +**U2 — line 123, an empirical explanation asserted as succeeded.** + +> This explains why organisations use licences, qualifications, grades, track records, +> certification, institutional standing, and selection. + +The explanatory claim about real institutions is at best `DDD-hyp-05` (empirical, **projected**). +"Explains" reports an explanation as achieved; "would explain" or a status mark is what the graph +supports. + +**U3 — line 320, a mechanical account of hallucination.** + +> This also yields a mechanical account of one class of hallucination: output decoupled from +> correct ground because that ground is absent, false, or present but unresolved. + +No claim carries a hallucination account at any status. The nearest node, `DDD-floor-01`, is +`reported` on toy actors and says nothing about models' hallucination. A real-world phenomenon is +presented as accounted for. + +**U4 — lines 356–366, "tested", and the two front pages disagree with each other.** + +> The actor-general vocabulary is tested against the vertebrate immune system. +> The mapping is not intended as metaphor: […] +> The immune system matters because no engineer specified it. If the framework's categories only +> worked for deliberately engineered systems, the actor-general claim would fail here. + +No claim carries the immune-system instance at any status; "is tested" reads as a performed and +passed test, and the closing sentence frames it as a survived falsification attempt. Worse: the +**downstream README's own "What changed from v3"** records "the immune-system 'licensing' argument +is **demoted to a suggestive parallel with known disanalogies**". One front page demotes what the +other presents as a load-bearing worked instance — an outside reader can quote the two against each +other. + +**B1 — borderline, reported for the ruling rather than asserted as a defect.** Upstream line 7: +"It applies to any arrangement that determines choices against ground: programs, models, humans, +organisations, markets, immune systems, and compositions of them." Read as purpose ("is for +analysing"), this is the primer-§6-legitimate kind; read as an applicability assertion, its +universality is `projected` (`DDD-frame-01`'s enumeration). The same sentence-form question governs +several purpose-adjacent lines in both files. + +**What is already right, for calibration.** Upstream: "These hypotheses are **projected**, not +reported findings"; the "What this framework does not claim" list; the Register section; "toy +demonstrations are not presented as external validation". Downstream: "We publish the review and +the retreats as first-class documents." The repairs should converge on this register — it already +exists in both files. + +## Proposed repair shape (for the ruling; not executed) + +Per the charter: **repair over-claims, do not rewrite**. Each finding repairs in place — the +sentence keeps its position and its purpose: + +- D1: replace with the paper's ruled formulation (fill in the actor slot the results "hold largely + fixed"; drop "missing parameter" and "changes their predictions" — what is supported is that the + framework *indexes* their questions, §11's wording). +- D2: "None of them makes that actor explicit" → the paper's "each holds it largely fixed". +- D3, D4, D6, U1, U2: add the status in place (the primer's device: assert-with-grade), and + "results" → "claims"/"the framework's central claim" where projected. +- D5: "proven, exercised, or still projected" → the actual vocabulary; drop "harder to knock down". +- D7: repair with I-3's pin advance (one edit, the correct version, or point at the file without + naming a version so it cannot go stale again). +- D8: "shown invariant" → the identity holds under the projected identification; "paid" → scoped to + what measure-02 carries. +- U3: mark as the exposition it is ("suggests a mechanical reading of one class of hallucination", + or carry `DDD-floor-01`'s status and region). +- U4: align the two front pages — either the upstream section adopts the demoted register + ("suggestive parallel, disanalogies recorded") or Emil rules the demotion itself superseded by + core/12; the two files must say the same thing. ("Tested" → what core/12 actually does, at its + actual grade.) +- B1: Emil rules the sentence-form question once; the same ruling covers its instances. + +**HOLD at GATE 2 — awaiting Emil's ruling on the findings and the repair shape.** + +--- + +## RULING (Emil, 2026-09-03): all twelve repair, plus B1 + +- **D1 is two defects and both go** — the absence claim withdraws with the paper's ruled + formulation; "changes their predictions" goes entirely: §11 explicitly declines the move, so the + README claimed something the paper refuses. +- **D5 takes the real vocabulary** with §5's glosses, "harder to knock down" goes with it — the + highest-priority repair on the page. +- **D3, D4, D6, U1, U2: status marks in place.** D4 specifically must stop leaning the projected + identification on the established chain rule — the identity-as-evidence move the measure note's + review caught must not survive on a front page after the paper repaired it. +- **D2 takes "largely/usually held fixed"** — the absolute is falsified by any counterexample and + the survey supplies candidates. +- **D7 repairs with I-3's advance. U3 becomes marked exposition** — the nearest node is toy-actor + reported, and "a mechanical account" claims more. +- **B1: repair the sentence-form** — "applies to any arrangement" asserts universal applicability; + "is written to apply across arrangements" states the intent, which is what a purpose sentence is + for. One ruling, all instances. +- **U4: the v3-era demotion governs, and this is the sharper finding.** The immune-system material + is a suggestive parallel with known disanalogies, and no claim carries the instance. core/12 + being a document is not supersession: prose does not overturn a recorded demotion, and if it + were meant to, that would need a decision node nobody filed. The upstream README loses the + tested-against and survived-falsification framing. Recorded at its true size: **two front pages + contradicting each other, one recording the demotion the other ignores — the strongest single + argument that this session had to happen before publication.** +- Repair shape confirmed: converge on the register already present in both files, in place, no + rewriting. Frame-08's `changed: v5.14` correctly flagged now; the close proposes the version and + the field either matches or is corrected there. + +**Executed:** downstream `3a575bb` (D1–D6, D8), upstream `6e14a83` (U1–U4, B1); D7 rides I-3. diff --git a/meta/sessions/2026-09-02-pre-transfer/gate3-pin-advance.md b/meta/sessions/2026-09-02-pre-transfer/gate3-pin-advance.md new file mode 100644 index 0000000..7795582 --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/gate3-pin-advance.md @@ -0,0 +1,149 @@ +# GATE 3 — the pin advance to v5.13.0 (I-3): the prediction, before the operation + +**Status: draft-pending-ruling. Nothing advanced yet.** The `DDD-dec-29`/`DDD-dec-34` pattern: +prediction stated and committed before the operation; ref advanced first and hashes re-instrumented +second, so a pass is distinguishable from a skip; verified after; divergence recorded, never +reconciled. + +## The operation + +`graph/upstream.yaml` `ref: v5.12.0` → `ref: v5.13.0`. Tag verified at session start: `v5.13.0` = +`5c7fe46`, exactly the head of the upstream default branch. + +## Baseline, run fresh before anything moves + +`python3 validate-core-order.py core/` at `ref: v5.12.0`, this session, matches the migration +record's in-session observables exactly: + +> upstream: 71 pins resolved against the pinned ref, 0 basis-loss, 0 content-drift, 1 shadowed id(s) +> 5 documents, 5 terms, 5 graph objects, 4 embedded, 0 errors, 0 warnings + +## The prediction + +Quoted from the ground-migration execution record +(`meta/sessions/2026-08-31-ground-migration-exec/gate3-prediction.md` §2), stated there before the +upstream edits were made, and owed its verification by "a session that can verify its own +prediction, which is this one": + +> **Prediction: when the downstream pin next advances past these commits, exactly two W6 fire and +> no more, both from W1:** +> +> | id | occurrences in hashed fields | pinned hash (= live, verified) | predicted hash after edit | +> |---|---|---|---| +> | `DDD-measure-01` | 1 (statement) | `sha256:66b01ede631c0173d86696dca945ade53c9ffe37462bbe290a6df6b6de4b41ea` | `sha256:d78274d2350cb867c9ec3497aaf13caeafb4792f9e20c20697ba54114a58c198` | +> | `DDD-measure-16` | 2 (statement) | `sha256:2d763d1adf9e1c68c88d1f5c3e30550a5079e274d84e688c31f5f5bb51f1d94e` | `sha256:6828d06649045bfdf374f985a7984d337d01fdb42be29ef8f051a185411fe674` | +> +> **Zero W5** (no status moves), **zero W7** (no id added or shadowed), **zero E12/E13.** Of the +> 71 pins, exactly these two digests move; the other 69 are verified unchanged after the edits. +> The edit is the phrase `ground distribution` → `deployment distribution`, verbatim, nothing else +> in any hashed field. + +Restated for this operation's observables: at step (b) below the validator must report **71 pins, +0 basis-loss, exactly 2 content-drift — `DDD-measure-01` and `DDD-measure-16`, old and new digests +exactly as tabled — 0 status movement, the one standing W7 unchanged**. Any other id firing, any +other digest, or a silent pass is divergence, to be recorded and held, not reconciled. + +**Non-interference, stated so it is checkable:** this session's I-1 change to `DDD-frame-08` lives +on the upstream session branch only; the pin resolves against the `v5.13.0` tag, which does not +carry it. Predicted contribution of I-1 to this operation: **nothing**. (`DDD-frame-08`'s pinned +hash `cf73b307…` stays verified at the tag; its supersession fires at a *later* advance, after the +close's version proposal is cut.) + +## Order of operations + +a. This file commits (the prediction is a repo object before the operation — `DDD-dec-20`'s logic + applied to `DDD-dec-29`'s pattern). +b. **Ref advances alone** (`ref: v5.13.0`, no hash touched), one commit; validator runs; the two + predicted firings are **observed**, not assumed. +c. **Re-instrumentation second:** the two `content_hash` values move to the predicted digests, one + commit; validator returns to baseline shape (71 pins, 0 basis-loss, 0 content-drift, 1 shadowed + id). +d. The advance files as a decision, `core/decisions/DDD-dec-35` (the DDD-dec-10/16/18/25/28/29/34 + convention: an advance is governed, not mechanical), and the `upstream.yaml` comment block takes + the advance record with prediction and verification, as every prior advance has. +e. **The primer regenerates at the new pin** (`generate.py --upstream` on the local clone at + `v5.13.0`): generated blocks re-draw, the stamp's `pin=` moves to v5.13.0. The hand-written + sections carry the pin in their own text by design, so their `v5.12.0` references move by hand, + and §6's "the migration's changes arrive at the next advance" paragraph updates — **this is + that advance**, as §6 itself anticipates. +f. **Paper A's three checkers re-run at the new pin** (`check-quotations.py`, `check-appendix.py`, + `check-status.py`). Expected: the sixteen deferred manuscript occurrences were deferred *for* + this advance ("they quote canon at the pinned ref and regenerate when the pin next advances"). + Whatever the checkers report is **recorded verbatim at the verification hold**; which passages + move now versus carry is Emil's ruling, not this session's reconciliation. +g. **D7 repairs here** (GATE 2 ruling): the front page's "(`v5.5.0` at time of writing)" — the + version parenthetical is dropped in favour of pointing at `graph/upstream.yaml`, which is + authoritative and cannot go stale. +h. The successor item "passages quoting a live claim verbatim — `DDD-measure-12` and + `term:verdict` still read `ground distribution` at `v5.12.0`" **discharges at this advance**: + at v5.13.0 both nodes read `deployment distribution`, so the blocker the deferral named is + gone. Recorded in the successor file at the verification step. + +**HOLD — awaiting Emil's ruling on the prediction before the ref moves.** + +--- + +## RULING (Emil, 2026-09-04): prediction approved; advance + +As stated, with two additions: (f) none of the sixteen deferred occurrences moves here — a pin +advance is not a licence to reopen a deferral; what the checkers report is evidence, recorded for +whichever session next touches that manuscript; this session repairs no manuscript prose. (g) the +pointer repair is the better one — a stale version number recurs, a pointer does not. + +## VERIFICATION (executed 2026-09-04) + +**Step (b) — ref alone (`a2676cf`), observed:** + +> upstream: 71 pins resolved against the pinned ref, 0 basis-loss, **2 content-drift**, 1 shadowed id(s) +> W6 `DDD-measure-01`: pinned `sha256:66b01ede…` → resolves `sha256:d78274d2…` +> W6 `DDD-measure-16`: pinned `sha256:2d763d1a…` → resolves `sha256:6828d066…` +> total warnings: exactly 2; errors: 0 + +Both ids and all four digests exactly as tabled. Zero W5, zero E12/E13, the standing W7 unchanged +at 1. **The prediction held in every limb.** Divergence: none. + +**Step (c)–(d) — re-instrumented (`d8d0306`):** validator back to baseline shape — 71 pins, 0 +basis-loss, 0 content-drift, 1 shadowed id; 0 errors, 0 warnings. The advance filed as +`core/decisions/DDD-dec-35` (23 decisions valid) and recorded in `graph/upstream.yaml`'s comment +block in the house pattern. Non-interference verified: `DDD-frame-08`'s pinned hash unchanged at +the tag — predicted nothing, observed nothing. + +**Step (e) — primer (`571b557`):** 13 generated regions re-drawn; stamp `pin=v5.13.0`; `--check`: +"OK: stamp present, pin v5.13.0, all regions current." Hand-written sections verified to hold at +v5.13.0 (zero pre-migration vocabulary) and their pin references moved; §6's "arrives at the next +advance" paragraph discharged. Roster unchanged: 89 claims, `established` still the same four. + +**Step (f) — Paper A checkers at v5.13.0, recorded as evidence, nothing repaired:** + +| Checker | Target | v5.12.0 | v5.13.0 | +|---|---|---|---| +| `check-quotations` | paper-a.md | 30 verbatim, 0 failing | **30 verbatim, 0 failing** (rc 0) | +| `check-status` | paper-a.md | 96 assertions, 0 wrong | **96 assertions, 0 wrong** (rc 0) | +| `check-appendix` | paper-a.md + supplement | 75 rendered, 75 cited, 0 discrepancies | **4 discrepancies** (rc 1): `DDD-measure-01`, `DDD-measure-11`, `DDD-measure-16` statements; `term:verdict` canonical wording | +| `check-quotations` | supplement | 1 failing (UNCITED, the boundary rule) | 1 failing, **identical** | + +The four appendix discrepancies are the sixteen-deferral's expected expiry — the rows render canon +at v5.12.0 wording and canon moved. Per the ruling they are **recorded, not repaired**; noted in +the successor file for whichever session next touches the manuscripts. The supplement's one +failing quotation is identical at both refs — ref-independent and pre-existing, not this advance's +divergence; booked for the I-4 sweep (the supplement quotes the session's own boundary rule, which +is not a canon node, and the supplement was not a declared target of that checker). + +**Step (g) — D7:** the front page now points at `graph/upstream.yaml` as authoritative, no version +named. **Step (h):** the successor item's blocker recorded as discharged at `DDD-dec-35`. + +**Also recorded, an instrument note for I-4:** all three checkers print their docstring and exit +**0** when invoked with wrong arguments (e.g. an unrecognised `--upstream` flag) — a bad +invocation is indistinguishable from a pass by exit code alone. Found by making exactly that +mistake in this session; the counts above are from corrected invocations with directly captured +exit codes. + +> **ERRATUM (I-4, 2026-09-07): the instrument note above is RETRACTED.** Re-tested with no pipe +> in the measurement path, every bad invocation of all three checkers exits **1** — including +> the exact two-argument form this session used. The "exit 0" readings were the session's own +> measurement artefact: `script | tail -N; echo $?` reports `tail`'s exit status, always 0. The +> checkers satisfy the exit-code rule; the session's harness did not. The checker RESULTS table +> above is unaffected — it was produced with directly captured exit codes. Full test matrix and +> the recounted unwatched-surface class: `gate4-readiness.md` §0. + +**HOLD at GATE 3 (verification) — awaiting Emil's ruling on the verification record.** diff --git a/meta/sessions/2026-09-02-pre-transfer/gate4-readiness.md b/meta/sessions/2026-09-02-pre-transfer/gate4-readiness.md new file mode 100644 index 0000000..ce85dba --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/gate4-readiness.md @@ -0,0 +1,201 @@ +# GATE 4 — the publication readiness sweep (I-4) + +**Status: draft-pending-ruling. A report, not a repair pass.** Emil rules what files now and what +carries to the transfer. Led, as ruled, with the invocation defect — which leads as an erratum. + +--- + +## 0. ERRATUM first: the invocation defect retracts, and so does half of a second finding + +**The GATE 3 instrument note was wrong, and the defect was this session's, not the checkers'.** + +Tested directly, with exit codes captured with no pipe in between, every bad invocation of all +three Paper A checkers exits non-zero: + +| Invocation | check-quotations | check-appendix | check-status | +|---|---|---|---| +| no arguments | 1 | 1 | 1 | +| unrecognised flag (`--bogus`) | 1 | 1 | 1 | +| two arguments (`--upstream ` — the session's own mistake) | 1 | 1 | 1 | +| swapped argument order | — | 1 | — | + +`gen-appendix.py` (1), `generate.py` (2), `validate-claims.py` (1) and `validate-core-order.py` +(1) also exit non-zero on bad invocations. **Emil's exit-code rule is already satisfied by every +instrument tested.** + +What actually happened: the session measured exit codes as `script | tail -N; echo $?` — which +reports **`tail`'s** exit status, always 0. The "docstring + exit 0" observation was an artefact +of the session's own measurement harness. The finding is retracted; the GATE 3 record carries an +erratum note pointing here rather than a silent edit. + +**Half of the unpinned-embeds finding retracts with it.** The three ids embedded in downstream +`core/14-maturation.md` and `core/16-calibration-ledger.md` (`term:waterline`, `term:maturity`, +`term:calibration-ledger`) are **not** unwatched: they live in the downstream repository's own +registry (`core/graph/terms.yaml`, 5 terms) and the canonical invocation +(`validate-core-order.py core/`) validates their embeds (4 embedded, 0 errors). What remains true +and reportable: **run from the repository root (`validate-core-order.py .`), the tool +misclassifies the downstream-registry ids as unpinned upstream ids and FAILS with 4 spurious E13s +on a tree the canonical invocation passes.** A stranger cloning the public repo and running the +validator the obvious way sees a failing canon gate. That is a mode inconsistency in the +instrument, not an unwatched surface. + +**The unwatched-surface class, recounted honestly:** three prior instances stand (the +column-count guard, the zero-citation render, the abstract past three green checkers). The fourth +and fifth named at GATE 3 correct to: a validator mode inconsistency, and — the genuine new +instance — **this session's own rc-through-pipe measurement**: a harness reporting success where +it did not look, exactly the class's shape, found only because the retraction check re-ran +everything without the pipe. The class finding stands, with the newest instance being the +session's, and the lesson sharpened: the checkers were sound; the checking of the checkers was +not. + +## 1. Statuses that overstate their evidence (charter item) + +`spec/claim-format.md` §5: `reported` claims "at least one evidence entry whose asset reproduces." + +- **`DDD-measure-11`** — conceptual, `reported`, evidence: one `derivation` entry + (`core/09-the-measure.md` §7, §9 caveat 1). No asset. +- **`DDD-measure-13`** — formal, `reported`, evidence: one `derivation` entry (`core/09` §5, third + bullet). No asset. + +Both are exactly the defect the charter names. Facts that price the ruling: **neither is pinned +downstream**, so a demotion fires nothing; **the primer already reads both conservatively as +`projected`** (its §6 says so in terms); the primer session's filter already found them, so this +is the second session to look and decline to leave them. Dispositions for the ruling: (a) demote +both to `projected` now — cheap, honest, aligns canon with how its own projection already reads +them; the derivation evidence stays recorded; (b) write the missing assets — out of this session's +scope; (c) record why they wait. The primer's §6 warning would need no change under (a). + +No third instance: the primer's filter swept all eleven `reported` claims and found exactly these +two derivation-only; that sweep is cited rather than redone. + +## 2. Anything asserting external validation: clean + +Both repositories swept, world-facing surfaces (`core/`, READMEs, `spec/`, `apparatus/`, +`applications/`, `projections/`, `papers/`), two pattern families (strict: "external validation", +"empirically confirmed/validated", "validated against", "field-tested", "proven in practice"; +soft: "has been shown", "we show that", "demonstrates that", "confirmed by", "verified in +practice", "real-world evidence/results"). **Every hit is a denial or the §5 definition** ("not +presented as external validation"; "Neither field claims external validation"; the DORA doc's +"was *not* validated against"). Zero assertions found. The I-2 repairs removed the nearest +approaches ("tested against", "explains", "results"). + +## 3. Claim/term disagreements beyond I-1 + +- **NEW, front page: the closure bounds count.** `term:closure` (settled), `core/03` §, and + `DDD-measure-16`'s quotation all say closure is evaluable "within declared **resource, latency, + and confidence** bounds" — three. Upstream `README.md:117` says "within the declared resource, + latency, confidence, **and assurance** bounds" — four. The same defect family as the arity, on + the front page of the settled term's own repository. (The likely source: `DDD-floor-02`'s + relation tuple carries `assurance` as a separate coordinate; the README merged the enumerations.) + **Not repaired — the repair direction is a ruling:** either the README drops "assurance" and + conforms to the settled term (one word, I-2 family), or the term is wrong at settled grade, + which is a supersession question nobody has raised. The cheap reading is the first. +- **Checked and clean: `term:floor` vs `DDD-floor-02`.** The same shape as the arity — a settled + term locating the floor in the predicate, a projected claim relocating it to the six-coordinate + relation — but the claim's own `region` already states the reconciliation: "Consistent with, not + competing against, term:floor — the acceptance predicate is a coordinate of the relation." This + is disposition C avant la lettre, and evidence the I-1 ruling generalises. +- **Checked and clean:** `term:store`, `term:commitment-level`, `term:arrangement` against their + README and claim carriages — counts agree everywhere. +- The `core/05` four-element sentence was the arity's third count — repaired at I-1. + +## 4. Dangling references to artefacts that do not exist (mechanical pass, both repos) + +Method: every markdown link and backticked repo path in every `.md`, resolved against the +referencing file's directory and the repo root. Raw total ≈ 250; ~215 are in dated records, +CHANGELOGs and session files — **correct as history, no repair proposed**. The live findings: + +**4a. The worst artefact finding of the sweep: `applications/sdlc/` does not contain the v3 +framework the front page says it contains.** The downstream README (twice) and +`applications/sdlc/README.md` present the four v3 design documents — `01-foundations.md`, +`02-entity-reference.md`, `03-autonomy-levels.md`, `04-implementation.md` — as "retained" and the +framework as "**intact** — it now lives in `applications/sdlc/`". None of the four files exists +anywhere in the tree; the directory holds only the README and `production-as-ground.md`. History: +the docs lived under `docs/` and were deleted at commit `219ad6c` ("rewrite"); they were never +landed at the advertised path. A stranger clicks four dead links under an existence claim. This is +simultaneously a dangling-reference finding and an I-2-class front-page over-claim ("That +framework is intact"). Dispositions for the ruling: restore the four docs from `219ad6c^` (content +landing — a projection nobody in this session ruled); or repair the claims to match the tree (the +links point at history, the "intact" sentence weakens to what is true: retained in history, +`production-as-ground.md` and the reference implementation carry the projection today). The second +is this session's shape; the first is real work someone must own. + +**4b. The cross-repo reference class (the split's residue), world-facing members:** + +| Referencing file (downstream) | Dangles to | Note | +|---|---|---| +| `README.md` (front page, as **links**) | `core/01`, `core/03`, `core/04`, `core/09`, `meta/lineage-and-limits.md` | 404 on the public site; the objects live upstream. The reading-order section *says* the theory lives upstream, then links it repo-relatively | +| `CONTRIBUTING.md`, `LICENSE.md`, `meta/way-of-working.md`, `meta/the-declaration.md` | `meta/lineage-and-limits.md` | lives upstream only; the declaration is embedded in the primer, so the primer inherits the reference textually | +| `spec/claim-format.md` | `spec/claim-format-2-addendum.md` | the downstream spec names an addendum "in force" that exists only upstream — a reader of the public downstream repo cannot find the rules the spec says bind it | +| `apparatus/closure-principle.md` | `core/00-determination.md` | doubly stale: cross-repo AND the pre-renumber document name that no longer exists upstream either | +| `apparatus/adversarial-ground.md`, `papers/measure-note/*` (4 files), `paper-a-supplement.md` | various `core/*`, `core/assets/*` | same class | +| `core/13-cost-projection.md`, `core/14-maturation.md` | `meta/mdl-cost-manufacturing-assessment-2026-08-08.md` | the record exists upstream only | + +One repair shape covers the class: cross-repo references become explicit upstream URLs (or a +stated "upstream:" prefix convention), starting with the front page's five links. The full-corpus +rewrite is a session of its own; the front-page links are publication-critical now. + +**4c. Upstream, world-facing:** `spec/claim-format.md` → `meta/conversion-protocol.md` — the file +lives downstream; the upstream spec references its dependent layer's process file, which is both +dangling and against the spirit of the no-references-to-dependents rule (the letter covers +`core/`). One sentence to scope ("the projection repository's conversion protocol") or drop. + +**4d. Resolved during the pass (no action):** upstream CHANGELOG's `DDD-cost-04.yaml` — relocated +whole downstream under `DDD-dec-09`/`DDD-dec-10`, governed, the history is correct. +`meta/research-program.md` — `way-of-working` marks it "[to create]", a declared debt rather than +a dangling reference. + +**4e. Housekeeping found at arrival:** `meta/sessions/README.md` lacks index rows for +`2026-08-31-ground-migration-exec` and `2026-09-01-primer`. + +## 5. The known collisions, reported with costs, not repaired + +- **`verdict`** (C10, ground-migration): `term:verdict` names the induced assignment (the function + whose entropy is `D`); `term:act-individuation` and `term:outcome` use the per-act verdict (a + value of that function) — a function against its value across three settled entries established + by one document. `core/09`'s contract already disambiguates (`verdict|verdict function`) and the + registry carries the alias; only the bare `term:` display field collides. Cost to repair: a + settled registry entry's display field moves (the id `term:verdict` — the pin key — need not), + with a supersession record; embeds unaffected (canonical_md untouched). Cost to carry: an + outside reader meets one word naming two objects, with the disambiguation one contract away. +- **`projection`** (C10): `term:projection` (the compound's two axes) against "the engineering + projection" / the projection layer — and the colliding sense appears **inside `term:verdict`'s + settled canonical text**. The `denominations:` field repaired half at v5.12.0. Cost to repair + the rest: prose-sense disambiguation in a settled canonical_md → canonical text moves → pin + fires, embed re-projects. Cost to carry: same as above, and the collision sits inside a settled + definition a paper quotes. +- **Publication test, honestly applied:** both are documented, disambiguated-by-contract, and + quotable only as untidiness, not as error. They read as *untidy waits* unless Emil weighs the + settled-text location of `projection`'s collision as quotable-against-us. + +## 6. Also carried to the close + +- `DDD-frame-08`'s `changed: v5.14` awaits the close's version proposal (flagged at GATE 1). +- The four `check-appendix` discrepancies at v5.13.0 stand recorded for the next manuscript + session (GATE 3 ruling). +- The supplement's one failing quotation (the boundary rule, UNCITED, ref-independent) — the + supplement quotes a session artefact, not canon; either the quotation declares its source or the + checker's scope statement names the supplement out. One line either way; a ruling picks which. +- Out of scope, untouched, as chartered: W4's local items, Q40–Q46, the measure note's missing + checkers, the delivery layer, the G-track, the licence split and `meta/lineage-and-limits.md` + as public register (the transfer session's). + +## 7. Proposed dispositions, for the ruling (the session proposes; Emil disposes) + +**Files now (unfixable-or-expensive-once-public):** +1. `DDD-measure-11`/`-13` demote to `projected` (1a) — two claims move upstream, nothing fires. +2. The closure-bounds word on the upstream front page (§3) — one word, pending the direction + ruling. +3. The sdlc front-page existence claim and the four dead links (4a) — the repair-to-match-tree + variant; restoring v3 content is its own ruled session. +4. The downstream front page's five cross-repo links become upstream URLs (4b, first row). +5. The `spec/claim-format-2-addendum.md` reference gains its upstream location (4b) and the + upstream spec's `conversion-protocol` sentence is scoped (4c). +6. The two index rows (4e). + +**Carries to the transfer or later:** the collisions (§5); the full cross-repo reference rewrite +beyond the front pages; the v3 content restoration question; the validator's root-mode +inconsistency (§0) — an instrument fix with its own tests, wrong to rush inside a canon session; +the supplement-quotation scope line (§6). + +**HOLD at GATE 4 — awaiting Emil's ruling on the dispositions.** diff --git a/meta/sessions/2026-09-02-pre-transfer/manifest.md b/meta/sessions/2026-09-02-pre-transfer/manifest.md new file mode 100644 index 0000000..0bd75d8 --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/manifest.md @@ -0,0 +1,101 @@ +# Manifest — the pre-transfer session (2026-09-02 … 2026-09-07) + +Interactive canon curation; Emil ruling at every gate; nothing merged by the session. Branch +`claude/pre-transfer-publication-readiness-dh0dqo` in both repositories; PRs upstream-first. +The session's test throughout: *does this get more expensive, or become unfixable, once someone +outside can read it or fork it?* + +## The erratum, first (as ruled) + +The GATE 3 instrument note — "all three checkers print their docstring and exit 0 on a bad +invocation" — is **retracted**. Re-tested with no pipe in the measurement path, every bad +invocation of every instrument tested exits non-zero; the exit-0 readings were the session's own +`| tail; echo $?` harness reporting `tail`'s status. The erratum is appended to the GATE 3 record +(never silently edited), with the test matrix in `gate4-readiness.md` §0. **The unwatched-surface +class recounts at three prior instances** (the column-count guard, the zero-citation render, the +abstract past three green checkers) **plus the genuinely new one: the session's own measurement +harness** — an instrument passing where it did not look, found by re-checking the checker. + +A second stale-fact correction of the same discipline: GATE 4 carried C10's "the colliding sense +sits inside `term:verdict`'s settled canonical text" without re-verifying it against the live +registry. At v5.11.0 that was true; **the v5.12.0 `denominations:` repair already removed it** — +no canonical text but `term:projection`'s own contains the word. The GATE 4 amendment's premise +was therefore already discharged; see the `projection` row below. + +## Item dispositions, complete + +| Item | Disposition | +|---|---| +| **I-1** the accountability arity | **Ruled C and executed** (upstream `b0d603f`): different objects — the settled term defines the relation, the claim asserts design-time completeness; frame-08 re-scoped by supersession, prior statement in notes; verification corrected to the pair (the completeness sentence + the sanctionability line); Bovens lineage in full (four of five his; delta = authority linkage + tense, stated as a tense change); core/05 §2 carries five attributed and marked projected, the distinction and its test beside the embed; no term text moved; no schema field. Successor item 1 discharged | +| **I-2** the front pages (D1–D8, U1–U4, B1) | **All ruled repair and executed** (downstream `3a575bb`, upstream `6e14a83`): the withdrawn absence claim and "changes their predictions" gone; statuses marked in place; "proven"/"harder to knock down" replaced with §5's vocabulary; the identity-not-evidence sentence on the front page; the immune section demoted to the recorded register ("a reading to be broken, not a test that was passed"); purpose sentences state intent. D7 rode I-3 as the pointer repair | +| **I-3** the pin advance v5.12.0 → v5.13.0 | **Predicted, executed, verified** (`a2676cf`, `d8d0306`, `571b557`; downstream decision `DDD-dec-35`): ref first, hashes second; exactly two W6 observed — `DDD-measure-01` and `DDD-measure-16`, all four digests as the migration session wrote down before its edits; zero W5/E12/E13, W7 unchanged; divergence none. Non-interference predicted as nothing and observed as nothing (frame-08's pin verified unchanged at the tag). Primer regenerated, stamp `pin=v5.13.0`, `--check` green, hand-written stamps re-read; §6's arrival paragraph discharged | +| Paper A checkers at v5.13.0 | paper green (30/30 quotations, 96/96 status assertions); `check-appendix` 0 → 4 discrepancies — the sixteen-deferral's reason expiring on schedule; **recorded, not repaired** (ruled: a pin advance is not a licence to reopen a deferral); carried for whichever session next touches the manuscripts | +| Supplement's one failing quotation | ref-independent and pre-existing (identical at v5.12.0/v5.13.0; quotes the session's boundary rule, not canon); **carries** — one line either way (disclose the source or scope the checker), a ruling picks which | +| **I-4.1** `DDD-measure-11`, `DDD-measure-13` | **Ruled demote and executed** (upstream `081c303`): `reported` → `projected`, derivation-only evidence under §5's asset requirement; derivations stay recorded; unpinned, nothing fires; canon now agrees with how the primer already read them | +| **I-4.2** external validation | **Clean** — both repos, strict and soft patterns; every hit a denial or the §5 definition | +| **I-4.3** closure bounds | **Ruled conform and executed** (upstream `081c303`): the README returns to the settled three (resource, latency, confidence); the fourth bound was the front page's own addition | +| `term:floor` vs `DDD-floor-02` | checked, clean — the claim's own region carries the reconciliation (disposition C avant la lettre) | +| **I-4.4a** `applications/sdlc/` | **Ruled repair-to-match-tree and executed** (downstream `6da379d`): the four v3 documents named as history (deleted at `219ad6c` under `docs/`), de-linked; the front page says what is there rather than promising what will be; **restoring the v3 content is a separate ruled session** — carried, explicitly open | +| **I-4.4b** cross-repo references | Front page's five links (plus three same-class in the same file) → upstream URLs; `spec/claim-format-2-addendum.md` reference scoped with its upstream URL; CLAUDE.md's stale `v5.7.0` → pointer. **Zero cross-repo links remain in world-facing files**; the remaining backtick *mentions* (papers, apparatus, primer doc-names) carry to the full reference-rewrite session | +| **I-4.4c** upstream spec → `meta/conversion-protocol.md` | **Scoped** (upstream `081c303`): "the projection repository's" | +| **I-4.4e** session index rows | **Landed** (2026-08-31, 2026-09-01) | +| Validator root-mode defect | **Documented now, fixed later** (ruled addition): the README names the canonical invocation and the spurious root-mode E13s; the instrument fix carries with its own tests | +| `verdict` collision | **Carries** (ruled): the bare `term:` display field; disambiguated by `09`'s contract and the alias; genuinely untidy | +| `projection` collision | **Amendment premise already discharged at v5.12.0** — the `denominations:` repair removed the colliding sense from `term:verdict`'s canonical text; no canonical text but `term:projection`'s own contains the word. Remaining surface: the pervasive layer-sense in prose (way-of-working's layers, "software projection", `projections/`), corpus-wide vocabulary work — **carries at that size, put to Emil at this gate for confirmation** | +| Historical dangling mass (~215) | records stay as history; no repair | +| `DDD-cost-04`, `research-program.md` | resolved in-check: relocated under `DDD-dec-09`/`10`; a declared "[to create]" debt | +| Out of scope, untouched, as chartered | the transfer (licence split, editorship, fork paragraph, lineage register); the freight remainder; Q40–Q46; W4's local items; the measure note's checkers; the delivery layer; the G-track | + +## Instruments and validators, final state + +| Check | Result | +|---|---| +| upstream `validate-core-order.py core/` | exit 0, zero W4, 66 warnings (W1 class, standing) | +| upstream `validate-claims.py core/claims/` | 63 claims valid, **32 warnings — count unchanged** | +| upstream `validate-claims.py core/decisions/ --decisions` | 12 decisions valid | +| upstream `validate-releases.py releases/` | **10 descriptors valid** (v5.14.0 included) | +| downstream `validate-core-order.py core/` | 0 errors, 0 warnings; upstream block: 71 pins, 0 basis-loss, 0 content-drift, 1 shadowed id | +| downstream `validate-claims.py` (claims; decisions) | 26 claims valid, **6 warnings — count unchanged**; 23 decisions valid | +| primer `generate.py --check` | OK: stamp present, pin v5.13.0, all regions current | +| Paper A checkers | as recorded at GATE 3 verification (appendix divergence carried by ruling) | + +## Basis-impact: what the next pin advance fires, by id and hash + +Stated now so the next session verifies rather than assumes. When the downstream pin advances +past v5.14.0: **exactly one W6** — `DDD-frame-08`, `sha256:cf73b307…` → `sha256:ef5d5faf…` +(statement re-scoped at GATE 1, status held; digest computed with the validator's own +normalisation, method sanity-checked by reproducing `DDD-measure-01`'s verified pin digest +exactly). **Zero W5** — the demoted claims are unpinned. The standing W7 unchanged. Downstream +consequences at that advance: the primer's generated roster re-draws (measure-11/13 rows read +`projected`), and its hand-written §6 sentence "reads them conservatively as projected" becomes +redundant rather than wrong. + +## Version proposal + +**`releases/v5.14.0.yaml` is in the upstream PR** — merging it cuts the tag (no manual step). +Minor, not patch: a projected claim's statement moves by supersession and two statuses move; not +major: every format-1 claim valid unchanged, no id moves, no settled node moves. `DDD-frame-08`'s +`changed: v5.14` and the demotions' `changed:` fields match the proposal; if Emil renames the +version, those three fields are the corrections. + +**Open question for this gate:** whether ruling C also files as an upstream decision node (the +`DDD-dec-15` precedent for a ruled re-scoping) or rests, as now, in frame-08's notes plus the +session record. Not filed unilaterally. + +--- + +## RULING (Emil, 2026-09-07) — GATE 5: accepted + +- **`projection`:** already-filed-at-v5.12.0 plus remainder-carries. No settled text moves; the + remaining layer-sense in prose is corpus-wide vocabulary work and does not meet the + pre-transfer test now that the canonical text is clean. +- **Ruling C files as an upstream decision node** — the `DDD-dec-15` precedent governs: a + stranger reading frame-08 should reach why it was re-scoped through the graph rather than + through `meta/`, and the Bovens lineage constraint is a governing decision that outlives the + session. Notes carry the text; a decision carries the ruling. **Filed as `DDD-dec-36`** + (upstream `fa82c83`, in PR #22): the ruling, the two refused dispositions with their reasons, + the test, the lineage constraint, the executed consequences. Added to v5.14.0's basis; + frame-08's notes point at it. Notes are unhashed — the next-advance prediction is unmoved + (frame-08's digest verified still `ef5d5faf…`). +- **Version:** v5.14.0 as proposed; the three `changed:` fields match. +- **Both PRs accepted** — Emil merges #22 (cutting the tag), then #35. diff --git a/meta/sessions/2026-09-02-pre-transfer/prompt.md b/meta/sessions/2026-09-02-pre-transfer/prompt.md new file mode 100644 index 0000000..afb8037 --- /dev/null +++ b/meta/sessions/2026-09-02-pre-transfer/prompt.md @@ -0,0 +1,119 @@ +# Session prompt — pre-transfer: what must be right before anyone else can read it + +Repositories: `actor-indexed-determination` (upstream, head — canon at **v5.13.0**; verify) and +`decision-driven-design` (downstream, head — the primer merged). Fetch both. +Session type: **interactive canon curation.** Hold at every gate. Supersession, never rewriting. +British spelling; one idea per sentence. +**First act, before Gate 1:** commit this prompt and bootstrap to +`meta/sessions/2026-09-02-pre-transfer/` in `decision-driven-design`, per DDD-dec-20. + +## What this session is + +**The last work before the repositories become public.** Four items, chosen by one test: *does this +get more expensive, or become unfixable, once someone outside can read it or fork it?* + +Everything else waits. This session is not a freight sweep — the freight list keeps its remaining +items, and the transfer session that follows is governance, not canon. + +**The standing reason:** after publication, a repaired defect becomes a *visible correction* against +a public record, a superseded statement may already be quoted elsewhere, and a false originality +claim on a front page is the most expensive kind of error the framework can make. All four items +below are cheap now and dear later. + +## I-1 — the accountability arity *(the one that reaches canon)* + +**Ruled at Paper A's Gate 3 as both-not-either:** the paper flags that canon disagrees with itself +about the relation's arity, and a canon session owes the ruling. This is that session. + +`DDD-frame-08` (projected) states a five-element accountability relation. `term:accountability` +(settled) states "borne consequence" as one element where the claim refines it into stake plus +sanction path, and the claim adds **authority linkage**, which the term does not carry. + +**So a projected claim adds an element to a settled term.** That is a supersession question, not an +elaboration, and it is exactly the shape that must not be published unresolved: an outside reader +finds a claim and a term that disagree, with nothing saying which governs. + +**Report both texts verbatim first.** Then propose — do not assume — one of: the term supersedes to +carry five; the claim narrows to the term's three; or the two are reconciled as different objects +(the term defining the relation, the claim asserting what a *complete* instance requires) with the +distinction stated in both. + +**Note the survey result that bears on it:** Paper A's related-work section found four of the five +elements are already Bovens's, and what is not his is authority linkage plus a change of tense from +retrospective assessment to design-time constitution. Whatever files must not claim more than that. + +**GATE 1 — hold on both texts and the three dispositions.** + +## I-2 — the README over-claim, and the front page generally + +`README.md:32` reads that *supplying the missing parameter changes their predictions* — an absence +claim the related-work survey no longer supports, and the same over-claim the paper's retitle +withdrew. It is the transfer's front door. + +**Scope it wider than one line, and only here:** read both repositories' READMEs as a stranger +would, against what the graph carries. Report every sentence that asserts more than a status +supports, or claims an absence the survey does not support. **The primer's own §6 is the model** — +a front page may say what the framework is for; it may not say what the framework has shown unless +a status carries it. + +**Constraint:** repair over-claims, do not rewrite the READMEs. A front page that becomes a new +document in this session is a projection nobody ruled. + +**GATE 2 — hold on the findings before any repair.** + +## I-3 — the pin advance to v5.13.0 + +The downstream pin sits at v5.12.0. The migration's two W1 firings are predicted and named by id and +hash in the ground-migration session's record; they belong to a session that can verify its own +prediction, which is this one. + +**Prediction stated before the operation, verified after, divergence recorded rather than +reconciled** — the `DDD-dec-29`/`DDD-dec-34` pattern. Advance the ref first and re-instrument hashes +second, so a pass is distinguishable from a skip. + +**Consequences to carry:** the primer pins v5.12.0 in its generated header and §6 and says the +migration's changes arrive at the next advance — **this is that advance**, so the primer regenerates +and its stamp updates. Paper A's checkers re-run at the new pin. The successor item noting that +`DDD-measure-12` and `term:verdict` still read `ground distribution` at v5.12.0 discharges here. + +**GATE 3 — hold on the prediction, then on the verification.** + +## I-4 — the publication readiness sweep + +Not a repair pass. **A report**, and the last chance to find what only matters once the repositories +are public: + +- **Statuses that overstate their evidence.** `DDD-measure-11` and `DDD-measure-13` are already + found — derivation-only evidence under a status `spec/claim-format.md` §5 defines as requiring a + reproducing asset. Rule them here or record why they wait; the primer already treats them as + projected. +- **Anything asserting external validation.** The framework has none, and §5 says so; a sentence + implying otherwise is the defect most likely to be quoted against it. +- **Unresolved disagreements between a claim and a term**, beyond I-1. The arity was found by a + paper; the sweep asks whether it is the only one. +- **Dangling references to artefacts that do not exist**, in either repository. Seven arrival + failures say this is worth one mechanical pass. +- **The remaining known collisions**, `verdict` and `projection`, with their contract findings — + reported with costs, not repaired here. + +**GATE 4 — hold on the report; Emil rules what files now and what carries to the transfer.** + +## Close + +**GATE 5** — basis-impact sweep; instruments; validators both repos; reference closure; branch per +repo; PRs upstream-first; manifest with every item dispositioned; version proposal with reasoning. + +## Out of scope + +The transfer itself — the licence split, the editorship decision, the fork paragraph, and +`meta/lineage-and-limits.md` as the public attribution register are the next session, and it is +governance rather than canon. The remaining freight list. Q40–Q46. W4's local items. The measure +note's missing checkers. The delivery layer and the G-track. Do not bundle. + +## Standing note + +Commit drafts before reporting at each gate, bodies marked draft-pending-ruling. **The test for +everything in this session is publication, not tidiness.** An item that is merely untidy waits; an +item that becomes unfixable, or embarrassing, or quotable-against-us once a stranger can read it +lands now. Where the two are hard to tell apart, report and let Emil rule — that judgement is the +session's real work. diff --git a/meta/sessions/README.md b/meta/sessions/README.md index 54f289b..f1f9f93 100644 --- a/meta/sessions/README.md +++ b/meta/sessions/README.md @@ -35,3 +35,6 @@ session's later work exceeds the committed prompt, the excess is visible as exce | `2026-08-25-item4/` | Interactive canon curation — item 4: status, kind, the validator and the appendix (I-1…I-4) | | `2026-08-27-ground-migration/` | Interactive canon curation with an execution phase — item 5: the ground migration, on the ground audit's rulings (G1…G6) | | `2026-08-30-paper-a-revision/` | Interactive paper revision — Paper A answers its external review, advances its pin to `v5.12.0`, and carries W1 (R-1…R-6) | +| `2026-08-31-ground-migration-exec/` | Interactive canon curation with an execution phase — the ground migration executed upstream; the two-W6 firing predicted for the next pin advance (G1…G6) | +| `2026-09-01-primer/` | Interactive drafting — the primer at `projections/primer/`, generated where it can be, pinned always (gates 1–4) | +| `2026-09-02-pre-transfer/` | Interactive canon curation — publication readiness before transfer: the accountability arity, the front pages, the pin advance to `v5.13.0`, the readiness report (I-1…I-4) | diff --git a/meta/successor-items-paper-a-revision.md b/meta/successor-items-paper-a-revision.md index c110d80..ccf665b 100644 --- a/meta/successor-items-paper-a-revision.md +++ b/meta/successor-items-paper-a-revision.md @@ -10,6 +10,12 @@ canon has to leave as an item rather than as an edit. That is the whole reason t ## 1. `term:accountability` and `DDD-frame-08` disagree about the relation's arity +**DISCHARGED — pre-transfer session, GATE 1 ruling C (Emil, 2026-09-03).** The two are different +objects: the settled term defines the relation; the claim asserts what a complete instance requires +at design time. `DDD-frame-08`'s statement re-scoped by supersession; the term untouched; the +distinction and its test stated in `core/05` §2 beside the embed; the Bovens lineage filed in the +claim's notes. Record: `meta/sessions/2026-09-02-pre-transfer/gate1-arity.md`. + **Ruled at GATE 3 (Emil): the paper flags, and a canon session gets the issue — both, not either.** The graph carries two counts of accountability's elements. @@ -66,6 +72,11 @@ judgement call: agreement the pin does not have; - **passages quoting a live claim verbatim** (4) — `DDD-measure-12` and `term:verdict` still read `ground distribution` at `v5.12.0`, so a rename would make the note misquote a live node; + *(the blocker discharged at the pre-transfer session's pin advance to `v5.13.0`, `DDD-dec-35`: + both nodes now read `deployment distribution` at the pinned ref. The deferral itself was NOT + reopened there — Emil's GATE 3 ruling: a pin advance is not a licence to reopen a deferral, and + the four discrepancies check-appendix now reports at `v5.13.0` are recorded for whichever + session next touches the manuscripts.)* - **one filename** (`measure-nonuniform-ground.py`, upstream) and **three occurrences ruled ambiguous** between the population sense and the sense naming conditions in the case. diff --git a/projections/primer/primer.md b/projections/primer/primer.md index d88ca74..0704450 100644 --- a/projections/primer/primer.md +++ b/projections/primer/primer.md @@ -6,7 +6,7 @@ nobody made, and know where this framework stops. It is not the argument for the that lives in canon and in Paper A, and §6 says exactly what this manual does not carry. -This primer describes canon at **`actor-indexed-determination v5.12.0`** — the version `graph/upstream.yaml` pins. Generated blocks are drawn from that tag; hand-written sections say so and carry the same pin in their own text. +This primer describes canon at **`actor-indexed-determination v5.13.0`** — the version `graph/upstream.yaml` pins. Generated blocks are drawn from that tag; hand-written sections say so and carry the same pin in their own text. This file is a projection. The marked blocks are generated by `generate.py` from the pinned @@ -17,7 +17,7 @@ generation ran, not merely that nothing looks wrong. ## 1. What this is for -*Hand-written against `actor-indexed-determination v5.12.0`; the quotation below is generated +*Hand-written against `actor-indexed-determination v5.13.0`; the quotation below is generated from `meta/the-declaration.md`.* @@ -56,7 +56,7 @@ determination comes from — and escape-prevention: making sure none is supplied ## 2. Filing a decision -*Hand-written against `actor-indexed-determination v5.12.0`; term boxes generated at the pin.* +*Hand-written against `actor-indexed-determination v5.13.0`; term boxes generated at the pin.* ### 2.1 Admit it @@ -65,7 +65,7 @@ determination comes from — and escape-prevention: making sure none is supplied > > **A fact is ground iff varying *the world* moves the outcome past tolerance.** > -> — `term:admission-test` (registry status *settled*, established by `00-primitives.md`, at `v5.12.0`) +> — `term:admission-test` (registry status *settled*, established by `00-primitives.md`, at `v5.13.0`) Run the test before filing anything. If varying the choice does not move the outcome past @@ -78,7 +78,7 @@ against: > everything: a choice is a governing decision, and a fact is ground, only relative to a > declared tolerance. Without one, the decision set is not well-formed. > -> — `term:tolerance` (registry status *settled*, established by `00-primitives.md`, at `v5.12.0`) +> — `term:tolerance` (registry status *settled*, established by `00-primitives.md`, at `v5.13.0`) Declare the tolerance first. A register whose entries carry no tolerance cannot be admitted or @@ -99,7 +99,7 @@ following it buys you is a claim, and claims live in §6. > **{rule, check, actor, nothing}.** There is no fifth source. > -> — `term:store` (registry status *settled*, established by `01-the-principle.md`, at `v5.12.0`) +> — `term:store` (registry status *settled*, established by `01-the-principle.md`, at `v5.13.0`) Name which of the four supplies this decision at the acts it governs: a rule (standing, @@ -132,7 +132,7 @@ author. ## 3. Reading an act -*Hand-written against `actor-indexed-determination v5.12.0`; term boxes generated at the pin.* +*Hand-written against `actor-indexed-determination v5.13.0`; term boxes generated at the pin.* > The **act** — the bounded episode of determination running from its first governing @@ -140,7 +140,7 @@ author. > an act is decisions resolved against ground. Its exact individuation is earned in `09` — > one verdict, one act, where the measure exists. > -> — `term:act` (registry status *settled*, established by `00-primitives.md`, at `v5.12.0`) +> — `term:act` (registry status *settled*, established by `00-primitives.md`, at `v5.13.0`) To read an act, answer three questions in order: what governed it, what was delivered, what @@ -160,7 +160,7 @@ List the decisions whose applicability predicates hold at this act — evaluated > alone: the same decision can be act-triggered at one act-site and judgment-mediated at > another, and a path from a decision to an act is act-triggered only if **every** edge on it is. > -> — `term:delivery` (registry status *draft*, established by `13-delivery.md`, at `v5.12.0`) +> — `term:delivery` (registry status *draft*, established by `13-delivery.md`, at `v5.13.0`) For each decision in the governing set, name the channel that carried it to this act-site. @@ -195,7 +195,7 @@ ran. ## 4. The floor, operationally -*Hand-written against `actor-indexed-determination v5.12.0`; term boxes generated at the pin. +*Hand-written against `actor-indexed-determination v5.13.0`; term boxes generated at the pin. This section asserts more than the others; each assertion carries its status.* @@ -205,7 +205,7 @@ run, in judgment. > **The intrinsic floor is a property of the acceptance predicate, not of the decision.** > -> — `term:floor` (registry status *settled*, established by `03-the-floor.md`, at `v5.12.0`) +> — `term:floor` (registry status *settled*, established by `03-the-floor.md`, at `v5.13.0`) The definition does the work: the floor is a property of the acceptance predicate, not of the @@ -217,7 +217,7 @@ per predicate: ground is observable and adequacy can be evaluated within declared resource, latency, and confidence bounds. **Decidable** is reserved for the formal special case. > -> — `term:closure` (registry status *settled*, established by `03-the-floor.md`, at `v5.12.0`) +> — `term:closure` (registry status *settled*, established by `03-the-floor.md`, at `v5.13.0`) **Does the predicate close for *this* arrangement, over material it can inspect, within the @@ -250,13 +250,13 @@ check the worker. One object, two predicates, two correct arrangements. ## 5. Finding escape -*Hand-written against `actor-indexed-determination v5.12.0`; term boxes generated at the pin.* +*Hand-written against `actor-indexed-determination v5.13.0`; term boxes generated at the pin.* > **Escaped** — determined *never*, by nobody: decided-by-nobody as a first-class category. > Latent defect exposure. **The only forbidden state.** > -> — `term:escape` (registry status *settled*, established by `01-the-principle.md`, at `v5.12.0`) +> — `term:escape` (registry status *settled*, established by `01-the-principle.md`, at `v5.13.0`) Escape is supply-general: however the failure happened, the state it names is a determination @@ -294,7 +294,7 @@ held, the excess escapes: > actor's residual escapes without overflowing anything: escape is supplied-by-nobody > (`term:escape`), and capacity shortfall is one generator of it. > -> — `term:escape-mechanism` (registry status *settled*, established by `11-the-floor-mechanism.md`, at `v5.12.0`) +> — `term:escape-mechanism` (registry status *settled*, established by `11-the-floor-mechanism.md`, at `v5.13.0`) In a register this looks like an actor assigned acts whose predicates do not close, at a @@ -308,7 +308,7 @@ reached the act: > decision at the act, so it was determined by nobody: escape, with a distinguishing feature — > **the ledger shows coverage.** Escape that presents as governance. > -> — `term:undelivered` (registry status *draft*, established by `13-delivery.md`, at `v5.12.0`) +> — `term:undelivered` (registry status *draft*, established by `13-delivery.md`, at `v5.13.0`) This is the generator a healthy-looking register hides best, because the ledger shows @@ -336,12 +336,12 @@ generator, open and unexamined in canon (§6). ## 6. What this does not do -*Hand-written against `actor-indexed-determination v5.12.0`; the roster is generated.* +*Hand-written against `actor-indexed-determination v5.13.0`; the roster is generated.* -**The pin, plainly.** This primer describes canon at `v5.12.0`, the version this repository -pins. The ground migration's prose changes landed upstream after that tag; they reach this -document at the next pin advance, which is a governed decision and not this document's to -make. +**The pin, plainly.** This primer describes canon at `v5.13.0`, the version this repository +pins. The ground migration's prose changes arrived at this pin: the advance was taken as a +governed decision (`DDD-dec-35`, pre-transfer session), and this document regenerated at it — +the hand-written sections were re-read against the new tag in the same act. **The procedures outrank their justifications here, deliberately.** §§2, 3 and 5 rest on terms — which define and do not assert — and on normative filing rules. The claims that say @@ -350,7 +350,7 @@ has met. A practitioner taught speculation as settled discards the lot on discov manual would rather you knew. -89 claims at `v5.12.0` (principle) and this repository's head (projection). `established` is 4 claims, all `formal`: `DDD-frame-06`, `DDD-measure-02`, `DDD-measure-10`, `DDD-measure-16`. A status is not a confidence score and does not aggregate. +89 claims at `v5.13.0` (principle) and this repository's head (projection). `established` is 4 claims, all `formal`: `DDD-frame-06`, `DDD-measure-02`, `DDD-measure-10`, `DDD-measure-16`. A status is not a confidence score and does not aggregate. | Area | established | reported | projected | retired | |---|---|---|---|---| @@ -414,4 +414,4 @@ document's own stamp, which asserts that generation ran at the pin, and nothing **Open and unexamined.** The empty-option-set escape generator is recorded in canon as open. This manual's §5 does not sweep it. - + diff --git a/spec/claim-format.md b/spec/claim-format.md index fd123ba..52b69d1 100644 --- a/spec/claim-format.md +++ b/spec/claim-format.md @@ -98,7 +98,9 @@ not checked prevents nothing; a rule checked by a proxy that overreaches prevent Rule 1's real enforcement is an adjudication with rulings. Rules added since format 1 — the falsifier condition at every live status, and `retired_from` on -retired claims — are stated in `spec/claim-format-2-addendum.md` with their migration notes. Both +retired claims — are stated in the principle repository's +[`spec/claim-format-2-addendum.md`](https://github.com/Hafeok/actor-indexed-determination/blob/main/spec/claim-format-2-addendum.md) +with their migration notes. Both are enforced now and both have empty hit lists. ## 5. What the statuses mean, and what they do not