LEARNING TO BUILD — Research edition 1.1.0

🏗️♾️ LEARNING TO BUILD ✨

How Intelligence Learns to Create What Makes New Possibilities Real

Build an ability worth having. Then test what the experience changes.

Research edition 1.1.0 · 7 November 2026
Vincent Boucher · MONTREAL.AI / QUEBEC.AI

Read the illustrated companion and complete paper.

This versioned continuation connects complete productive means to a fair test
of construction learning. It develops the connection among Transformational
Intelligence, Invent the Economy, and ALPHA GENESIS FLYWHEEL Ω, credits
systems engineering and program-learning foundations, and makes the proposed
contribution and its limits explicit.

What the edition adds

• A practical microscopy construction, from raw channels to instance masks,
  counts and pixel measurements, with exact input and execution identities.
• A preserved six-arm comparison separating fixed construction, ordinary
  artifact reuse, a learned ordering, the same content used by an ordinary
  empirical method, and full enumeration.
• A separately registered CellposeDINO practical alternative, operated with
  its frozen source, weights, runtime and parameters.
• Independent decoding, scoring, classical replay, adversarial integrity checks,
  checkpoint recovery, retained failures and a complete compact evidence archive.
• Frozen follow-up diagnostics explaining convergence under 33 order interventions
  and separating seed/foreground branches on four exposed calibration fields.
  Reference foreground improves the numerical diagnostic; it does not qualify a
  raw-input method or supply fresh confirmation.
• One separately frozen practical repair using the actual specialist foreground
  with classical seeds. It failed the calibration boundary gate (0.296809 versus
  0.60), and its stop rule prevented test evaluation. Its masks, separate scoring,
  failed verdict and measured costs are preserved.
• A portable residual-file verification/reconstruction path, an explicit unknown
  cost qualification for unresolved residual provenance, and preserved evidence
  of the extracted-installer failure and its checked repair.
• A clearer introduction, research programme, operator guide and French overview.

The measured claim verdicts are in construction/RESULTS.md and
construction/RESULTS.json. The classical source fails its complete validation
contract; all six classical successor searches exhaust the 24-recipe bank and
fail the boundary criterion. No constructor advantage is established. Software
operation, measurement qualification, an improved constructor, an enabled
successor and economic activity are reported separately.

The separately registered pretrained alternative qualifies on both nucleus
cohorts (mean instance F1 0.953878 on validation and 0.949103 on test). Its
whole-cell boundary F1 is 0.386561 on calibration and 0.505568 on test, below the
0.60 acceptance threshold. This inherited specialist competence establishes a
useful, scoped research operation; it is not evidence of locally learned
constructor improvement.

Evidence boundary

This is a reproduction of an already exposed registered comparison. The
earlier full unpushed result package was unavailable; the retained records are
identified as new executions, not a reconstruction of missing historical bytes
or a fresh confirmation. The evidence retains the original format failures,
explicit amendments, measured work, unknown cost boundaries and the observed
classical checkpoint projection with its residual-file provenance.

The original finite companion remains unchanged: four emitted programs, 4,000
declared cases, 288 authored scenarios, 864 complete-route decisions, and an
equal-content conventional tie. The original 1.0.0 publication, source and
evidence remain available.

Downloads

• LEARNING_TO_BUILD.pdf — complete research paper.
• LEARNING_TO_BUILD_SOURCE.zip — editable LaTeX, bibliography, both executable
  companions, registered sources, independent checks, complete generated
  reproduction evidence, operating instructions and the canonical PDF.
• PUBLICATION.json — exact canonical PDF and packaged-source identities.
• SHA256SUMS — release asset checksums.
• OPERATION_RECEIPT.json — added only after the actual downloaded source
  reproduces the finite companion and operates the fixed microscopy instrument;
  binds that operation to the four qualified assets. It also preserves the
  candidate scientific/browser/setup/operation receipts and downloaded setup
  and instrument timings. These are additional qualification costs after the
  bound ledger snapshot; nested intervals must not be double-counted.

Public image archives and third-party model weights are fetched separately by
fixed identity. The source package can be operated independently of the rest of
the repository. The publication workflow qualifies the extracted candidate
package before publication, preserves failure logs, and separately downloads
and operates the actual published source.
The resulting operation receipt is retained with the release. A later rerun
preserves an existing compatible receipt rather than replacing its history.

No laboratory or customer acceptance, revenue, profit, clinical utility,
commercial rights or general autonomous construction learning is established.
The value of the release is an inspectable research proposition and executable
evidence that can support a better next experiment.

When we build something new, we change what can be built next.

AGI ALPHA connected enterprise 014 — simulated operating release

Connected 014 released simulated enterprise

Immutable source commit: de9546ef9083f64a25186f1c5163e3a3a4eaeae5.

The source package passed clean operation, completed-account resume, independent native/operating audit and key-free public-package audit. The accompanying research edition states the achieved commercial and growth evidence and its conditions. Engineering smoke qualification is separate from fresh commercial confirmation. All customers, payments and earned reinvestment remain simulated. No real spending, outreach or deployment occurred.

Install the pinned requirements from the source ZIP, then follow docs/connected014/OPERATOR.md. Source/asset hashes are in RELEASE_MANIFEST.json and SHA256SUMS.txt. The publication workflow downloads and operates this actual release; its post-publication receipt is retained as a workflow artifact.

The unchanged paper describes historical source b0eae06dabfdf09197a60b60d3680f7f67946936631fafe431dc6be1f9b56983, NOT this composed financial implementation. COMPOSED_PUBLICATION.json separately binds actual current-source engineering qualification and the complete declared source delta. No fresh economic confirmation or recovered original-account continuity is claimed.

LEARNING TO BUILD — Research edition 1.0.0

🏗️♾️ LEARNING TO BUILD ✨

How Intelligence Learns to Create What Makes New Possibilities Real

Build what makes the next breakthrough buildable.

Research edition 1.0.0 · 7 November 2026
Vincent Boucher · MONTREAL.AI / QUEBEC.AI

Read the illustrated browser companion and complete paper.

This paper connects Transformational Intelligence, Invent the Economy, and ALPHA GENESIS FLYWHEEL Ω through the construction of complete productive means: knowledge, instruments, processes, resources and voluntary arrangements. It distinguishes the current frontier from the construction frontier and the reuse of an artifact from construction of a new capability and improvement of the constructor.

Change the game → build what the new game requires → learn to build better.

Included evidence

The finite companion emits four standalone measurement programs and checks them over 4,000 declared input cases. It evaluates 288 supplied cash/deadline/consent/maintenance cells across three states, totaling 864 complete-route decisions. The equally informed conventional control ties all 288 corresponding constructor-state decisions. Maintenance can reverse the apparent expansion of reachable outcomes.

The constructor, task specifications, cost parameters and participation conditions are authored. The programs and executions are computational artifacts; they do not establish autonomous discovery, empirical constructor learning, real customer acceptance, actual receipts or physical deployment. Separate checker source remains under common research authorship.

Download

• LEARNING_TO_BUILD.pdf — complete research paper.
• LEARNING_TO_BUILD_SOURCE.zip — editable LaTeX, bibliography, build instructions, finite study, evidence, source map, review and the canonical PDF.
• PUBLICATION.json — reviewed source and PDF identities.
• SHA256SUMS — release-asset checksums.

The archive can be reproduced independently of the rest of the repository. The README explains the complete commands and the distinction between canonical PDF identity, text-equivalent recompilation and same-toolchain byte reproducibility. The release workflow verifies the downloaded assets and operates the source companion from the downloaded archive. Existing historical papers, source editions and experimental outcomes are preserved.

When we build something new, we change what can be built next.

AGI ALPHA connected enterprise 014 — simulated operating release

Connected 014 released simulated enterprise

Immutable source commit: 2d1b3a2c1dab8235f2c7bef40bf9a36da1d62d7d.

The source package passed clean operation, completed-account resume, independent native/operating audit and key-free public-package audit. The accompanying research edition states the achieved commercial and growth evidence and its conditions. Engineering smoke qualification is separate from fresh commercial confirmation. All customers, payments and earned reinvestment remain simulated. No real spending, outreach or deployment occurred.

Install the pinned requirements from the source ZIP, then follow docs/connected014/OPERATOR.md. Source/asset hashes are in RELEASE_MANIFEST.json and SHA256SUMS.txt. The publication workflow downloads and operates this actual release; its post-publication receipt is retained as a workflow artifact.

The unchanged paper describes historical source b0eae06dabfdf09197a60b60d3680f7f67946936631fafe431dc6be1f9b56983, NOT this composed financial implementation. COMPOSED_PUBLICATION.json separately binds actual current-source engineering qualification and the complete declared source delta. No fresh economic confirmation or recovered original-account continuity is claimed.

SURF THE EXPONENTIAL — The Missing Move

SURF THE EXPONENTIAL — The Missing Move

Find what the answer missed. Build what the world is missing.

Research edition 1.2.0 by Vincent Boucher · MONTREAL.AI / QUEBEC.AI.

A research synthesis of Transformational Intelligence, ALPHA GENESIS FLYWHEEL Ω, and Invent the Economy: turn evidence-grounded criticism into missing capabilities, viable productive arrangements, and better discovery on fresh work.

Read the accessible browser edition, download the PDF, or rebuild the complete editable LaTeX and executable evidence from the source archive.

What should survive the answer? Edition 1.2 adds a costed choice among preserving a sufficient record, preserving access to its source, and relying on ordinary reconstruction. Cheaper reconstruction can reverse the best retained asset; rare opportunities can favor ordinary access. A stronger deferred-access control wins when identical access can be bought later. The browser edition includes an accessible interactive exploration with exact ties and an explicit late-access alternative.

The 32-page manuscript retains the verified three/six-label construction, changed-task successor, full-refund counterexample and 2,304-point review audit. Five new retention cases contain 16 policy outcomes. Separate checkers and 35 focused tests verify the finite models, including conventional ties and wins. These are authored conditional calculations, not observed frontier-model superiority or realized commerce. The stronger empirical programme remains prospective.

Earlier edition 1.0.0 and edition 1.1.0 remain available with their original artifacts.

PUBLICATION.json binds the reviewed PDF to its source inputs. SHA256SUMS verifies the release assets. The archive is self-contained; its README explains reproduction and the scope of byte reproducibility across TeX environments.

CONTINUITY Ω / CONTINUITÉ Ω — Bilingual edition / Édition bilingue 1.1.1

CONTINUITY Ω / CONTINUITÉ Ω — Bilingual research edition / Édition de recherche bilingue 1.1.1

English · Français

English

Title translation correction

Edition 1.1.1 gives the French paper its translated main title, CONTINUITÉ Ω, throughout its current cover, running headers, text and publication metadata. The English title remains CONTINUITY Ω. Both manuscripts carry the new shared edition number; their scientific content, equations, source identities, results and limitations are unchanged. The prior 1.1.0 release remains immutable. Historical cited work titles and machine filenames retain their authentic forms. The French cover series label and running footer now read INTELLIGENCE TRANSFORMATIONNELLE ∞.

The strengthened language checker rejects regressions in the French cover, running header, body macro and PDF metadata title. PUBLICATION.json records the new title QA; the older publication and translation QA JSON files remain unchanged and retain their original scope. This is an editorial correction, not new operational, commercial or model evidence.

Keep the promise.

Independently verified cyber resilience
Cybersecurity Resilience — Invent the Economy of Continuity

Vincent Boucher · MONTREAL.AI / QUEBEC.AI · 6 November 2026.

AI that finds the vulnerability is a breakthrough. Intelligence that helps businesses survive the vulnerability—and invents a viable market for continuity—is a different game.

Read the complete English browser edition or the complete French browser edition.

Downloads: CONTINUITY_OMEGA.pdf · CONTINUITY_OMEGA_FR.pdf · CONTINUITY_OMEGA_SOURCE.zip.

Complete bilingual reading paths

This is a complete bilingual publication: English and French manuscripts, LaTeX sources, browser reading paths, claim registers, citation metadata, evidence explanations, reviews and source-lineage guides. The current prototype's entry guide, quick start and proposed pilot also have complete French counterparts with reciprocal language links. Both languages disclose the same known checker weaknesses.

The English media gallery and French media gallery present four pairs of conceptual animations with still previews, play/stop controls and complete transcripts, alongside localized social cards and French citation metadata; these illustrations add no operational evidence.

The language contract preserves claim IDs, source anchors, numeric values, equations, status tokens, code identifiers and source identities. Raw recorded evidence stays shared. Translation does not create a new experiment, new customer acceptance or a qualification of the separately owned hardening candidate.

The source bundle contains both complete manuscripts, canonical PDFs, bibliographies, bilingual paper documentation and claim registers, compact evidence and build/reproduction tools. Each bibliography is embedded in its LaTeX source. The two README language editions provide complete instructions. To build and check both manuscripts:

python3 build.py --language all
python3 verify.py --check-build

build.py defaults to both languages; --language en and --language fr select an individual manuscript. The source ZIP and both PDFs belong to this new edition. The original 1.0.0 release, its tag, notes and assets remain unchanged.

What the paper contributes

CONTINUITY Ω makes the business commitment the unit of evaluation. A challenger exposes a consequential omission; evidence or an authorized test adjudicates it; construction changes the permitted route; and separate acceptance checks whether the original promise was kept.

The paper connects Transformational Intelligence, ALPHA GENESIS FLYWHEEL Ω and Invent the Economy to protected commitment records, portable production requirements, common dependencies, reserved capacity and complete costs. Its conditional pooling result shows why lower shared costs alone cannot establish viability: in the authored illustration, economics needs at least three buyers while simultaneous capacity permits at most two.

The proposed first service is a controlled continuity rehearsal for a regional packaging-label printer, one product and one qualified fallback provider. The commercial assumption concerns acceptance and funding of the complete qualified offer. Payment for preparation alone tests that stage only.

Executable evidence and its limits

Existing source	Reproduced result	Evidence boundary
PR 278 source at a755d0375693	29 tests pass; 12 CLI cases produce 3 synthetic handoff passes and 9 refusals; two outstanding commitments total 1,900 units.	Authored records and supplied constraints; a handoff pass does not repair the restored snapshot.
PR 277 source at 84c4288b4726	9 tests pass; a competent conventional reconciliation control ties all four authored decisions.	Implementation evidence, with no demonstrated model superiority.

The handoff example's US$650 contribution is modeled arithmetic, not collected money. The separate earlier GENESIS Commitment 001 central case's US$1,080 modeled loss remains unchanged. The proposed target of up to 20,000 units within 48 agreed business hours requires real provider qualification; it is not demonstrated by the 1,900-unit synthetic record.

The verifier was also challenged. Five separate probes against the unchanged handoff source retain a positive control, reproduce a Boolean monetary false pass, a malformed-input exception and stale inspect output, and demonstrate the already-disclosed coherent-reseal trust boundary. They were motivated by the later hardening review at its inspected revision. counterexamples.py and evidence/COUNTEREXAMPLES.json preserve the observations; this edition does not qualify the proposed later repair. Passing the original 29-test suite does not erase the counterexamples.

No frontier-model comparison, live cyber recovery, physical manufacturing, external customer acceptance, payment, market formation or causal successor advantage is established by this edition. The manuscript's separate AI review tasks shared one assistant workflow; they are not third-party certification.

Publication provenance

PUBLICATION.json binds both reviewed PDFs and the complete paper-source inventory. The publication workflow checks those identities and language consistency, downloads the two pinned public prototype sources, verifies their Git blob and SHA-256 identities, reruns their synthetic evidence and the separate counterexample probes, and rebuilds both manuscripts without replacing their canonical PDFs. The main-only release job publishes the two PDFs and source bundle from the qualified commit and checks their downloaded identities.

An existing edition's tag, notes and assets are preserved. Different publication bytes require a new edition or explicit investigation of the mismatch. This is a research release; it does not change the repository's enterprise operating authority or certify operational resilience or commercial success.

Français

Correction de la traduction du titre

L’édition 1.1.1 donne à l’article français son titre principal traduit, CONTINUITÉ Ω, sur sa couverture, dans ses en-têtes, son texte et ses métadonnées de publication courantes. Le titre anglais reste CONTINUITY Ω. Les deux manuscrits portent le nouveau numéro commun d’édition ; leur contenu scientifique, leurs équations, leurs identités de sources, leurs résultats et leurs limites sont inchangés. La publication antérieure 1.1.0 demeure immuable. Les titres historiques cités et les noms de fichiers techniques gardent leur forme authentique. Le libellé de série sur la couverture française et le pied de page indiquent désormais INTELLIGENCE TRANSFORMATIONNELLE ∞.

Le contrôle linguistique renforcé rejette toute régression du titre français sur la couverture, dans les en-têtes, dans la macro du texte et dans les métadonnées PDF. PUBLICATION.json consigne les nouveaux contrôles du titre ; les anciens relevés JSON de publication et de traduction restent inchangés et gardent leur portée d’origine. Il s’agit d’une correction é...

CONTINUITY Ω — Édition bilingue / Bilingual edition 1.1.0

CONTINUITY Ω — Bilingual research edition / Édition de recherche bilingue 1.1.0

English · Français

English

Keep the promise.

Independently verified cyber resilience
Cybersecurity Resilience — Invent the Economy of Continuity

Vincent Boucher · MONTREAL.AI / QUEBEC.AI · 6 November 2026.

AI that finds the vulnerability is a breakthrough. Intelligence that helps businesses survive the vulnerability—and invents a viable market for continuity—is a different game.

Read the complete English browser edition or the complete French browser edition.

Downloads: CONTINUITY_OMEGA.pdf · CONTINUITY_OMEGA_FR.pdf · CONTINUITY_OMEGA_SOURCE.zip.

What is new in 1.1.0

This is a complete bilingual publication: English and French manuscripts, LaTeX sources, browser reading paths, claim registers, citation metadata, evidence explanations, reviews and source-lineage guides. The current prototype's entry guide, quick start and proposed pilot also have complete French counterparts with reciprocal language links. Both languages disclose the same known checker weaknesses.

The English media gallery and French media gallery present four pairs of conceptual animations with still previews, play/stop controls and complete transcripts, alongside localized social cards and French citation metadata; these illustrations add no operational evidence.

The language contract preserves claim IDs, source anchors, numeric values, equations, status tokens, code identifiers and source identities. Raw recorded evidence stays shared. Translation does not create a new experiment, new customer acceptance or a qualification of the separately owned hardening candidate.

The source bundle contains both complete manuscripts, canonical PDFs, bibliographies, bilingual paper documentation and claim registers, compact evidence and build/reproduction tools. Each bibliography is embedded in its LaTeX source. The two README language editions provide complete instructions. To build and check both manuscripts:

python3 build.py --language all
python3 verify.py --check-build

build.py defaults to both languages; --language en and --language fr select an individual manuscript. The source ZIP and both PDFs belong to this new edition. The original 1.0.0 release, its tag, notes and assets remain unchanged.

What the paper contributes

CONTINUITY Ω makes the business commitment the unit of evaluation. A challenger exposes a consequential omission; evidence or an authorized test adjudicates it; construction changes the permitted route; and separate acceptance checks whether the original promise was kept.

The paper connects Transformational Intelligence, ALPHA GENESIS FLYWHEEL Ω and Invent the Economy to protected commitment records, portable production requirements, common dependencies, reserved capacity and complete costs. Its conditional pooling result shows why lower shared costs alone cannot establish viability: in the authored illustration, economics needs at least three buyers while simultaneous capacity permits at most two.

The proposed first service is a controlled continuity rehearsal for a regional packaging-label printer, one product and one qualified fallback provider. The commercial assumption concerns acceptance and funding of the complete qualified offer. Payment for preparation alone tests that stage only.

Executable evidence and its limits

Existing source	Reproduced result	Evidence boundary
PR 278 source at a755d0375693	29 tests pass; 12 CLI cases produce 3 synthetic handoff passes and 9 refusals; two outstanding commitments total 1,900 units.	Authored records and supplied constraints; a handoff pass does not repair the restored snapshot.
PR 277 source at 84c4288b4726	9 tests pass; a competent conventional reconciliation control ties all four authored decisions.	Implementation evidence, with no demonstrated model superiority.

The handoff example's US$650 contribution is modeled arithmetic, not collected money. The separate earlier GENESIS Commitment 001 central case's US$1,080 modeled loss remains unchanged. The proposed target of up to 20,000 units within 48 agreed business hours requires real provider qualification; it is not demonstrated by the 1,900-unit synthetic record.

The verifier was also challenged. Five separate probes against the unchanged handoff source retain a positive control, reproduce a Boolean monetary false pass, a malformed-input exception and stale inspect output, and demonstrate the already-disclosed coherent-reseal trust boundary. They were motivated by the later hardening review at its inspected revision. counterexamples.py and evidence/COUNTEREXAMPLES.json preserve the observations; this edition does not qualify the proposed later repair. Passing the original 29-test suite does not erase the counterexamples.

No frontier-model comparison, live cyber recovery, physical manufacturing, external customer acceptance, payment, market formation or causal successor advantage is established by this edition. The manuscript's separate AI review tasks shared one assistant workflow; they are not third-party certification.

Publication provenance

PUBLICATION.json binds both reviewed PDFs and the complete paper-source inventory. The publication workflow checks those identities and language consistency, downloads the two pinned public prototype sources, verifies their Git blob and SHA-256 identities, reruns their synthetic evidence and the separate counterexample probes, and rebuilds both manuscripts without replacing their canonical PDFs. The main-only release job publishes the two PDFs and source bundle from the qualified commit and checks their downloaded identities.

An existing edition's tag, notes and assets are preserved. Different publication bytes require a new edition or explicit investigation of the mismatch. This is a research release; it does not change the repository's enterprise operating authority or certify operational resilience or commercial success.

Français

Tenir parole.

Cyberrésilience vérifiée de manière indépendante
Résilience en cybersécurité — Inventer l’économie de la continuité

Vincent Boucher · MONTREAL.AI / QUEBEC.AI · 6 novembre 2026.

Une IA qui découvre la vulnérabilité représente une percée. Une intelligence qui aide les entreprises à y survivre — et qui invente un marché viable de la continuité — change la nature du jeu.

Lire l’édition française intégrale dans le navigateur ou l’édition anglaise intégrale dans le navigateur.

Téléchargements : CONTINUITY_OMEGA_FR.pdf · CONTINUITY_OMEGA.pdf · CONTINUITY_OMEGA_SOURCE.zip.

Nouveautés de la version 1.1.0

Il s’agit d’une publication intégralement bilingue : manuscrits anglais et français, sources LaTeX, parcours de lecture dans le navigateur, registres d’affirmations, métadonnées de citation, explications des éléments probants, revues critiques et guides de filiation des sources. Le guide d’entrée du prototype courant, son démarrage rapide et son pilote proposé disposent également de versions françaises complètes avec des liens de langue réciproques. Les deux langues exposent les mêmes faiblesses connues du vérificateur.

La galerie française et la galerie anglaise présentent quatre paires d’animations conceptuelles avec aperçus fixes, commandes de lecture et d’arrêt et transcriptions complètes, accompagnées de visuels de partage localisés et de [métadonnées de citation en français](https://github.com/MontrealAI/AGI-Alpha-Invention-Flywheel-Lab/blob/continuity-omega-v1.1.0/docs/...

CONTINUITY Ω — Research edition 1.0.0

CONTINUITY Ω — Research edition 1.0.0

Keep the promise.

Independently verified cyber resilience
Cybersecurity Resilience — Invent the Economy of Continuity

Vincent Boucher · MONTREAL.AI / QUEBEC.AI · 6 November 2026.

AI that finds the vulnerability is a breakthrough. Intelligence that helps businesses survive the vulnerability—and invents a viable market for continuity—is a different game.

Read the browser edition, download CONTINUITY_OMEGA.pdf, or use CONTINUITY_OMEGA_SOURCE.zip for the editable LaTeX, evidence, claim register and reproduction tools. The source bundle's README.md gives the build and verification commands.

What this paper contributes

CONTINUITY Ω makes the business commitment the unit of evaluation. A challenger exposes a consequential omission; evidence or an authorized test adjudicates it; construction changes the permitted route; and separate acceptance checks whether the original promise was kept.

The paper connects Transformational Intelligence, ALPHA GENESIS FLYWHEEL Ω and Invent the Economy to protected commitment records, portable production requirements, common dependencies, reserved capacity and complete costs. Its conditional pooling result shows why lower shared costs alone cannot establish viability: in the authored illustration, economics needs at least three buyers while simultaneous capacity permits at most two.

The proposed first service is a controlled continuity rehearsal for a regional packaging-label printer, one product and one qualified fallback provider. The commercial assumption concerns acceptance and funding of the complete qualified offer. Payment for preparation alone tests that stage only.

Executable evidence and its limits

Existing source	Reproduced result	Evidence boundary
PR 278 source at a755d0375693	29 tests pass; 12 CLI cases produce 3 synthetic handoff passes and 9 refusals; two outstanding commitments total 1,900 units.	Authored records and supplied constraints; a handoff pass does not repair the restored snapshot.
PR 277 source at 84c4288b4726	9 tests pass; a competent conventional reconciliation control ties all four authored decisions.	Implementation evidence, with no demonstrated model superiority.

The handoff example's US$650 contribution is modeled arithmetic, not collected money. The separate earlier GENESIS Commitment 001 central case's US$1,080 modeled loss remains unchanged. The proposed target of up to 20,000 units within 48 agreed business hours requires real provider qualification; it is not demonstrated by the 1,900-unit synthetic record.

The verifier was also challenged. Five separate probes against the unchanged handoff source retain a positive control, reproduce a Boolean monetary false pass, a malformed-input exception and stale inspect output, and demonstrate the already-disclosed coherent-reseal trust boundary. These were motivated by the later pinned hardening review. counterexamples.py and evidence/COUNTEREXAMPLES.json preserve the observations; this edition does not qualify that later proposed repair. Passing the original 29-test suite does not erase the counterexamples.

No frontier-model comparison, live cyber recovery, physical manufacturing, external customer acceptance, payment, market formation or causal successor advantage is established by this edition. The manuscript's separate AI review tasks shared one assistant workflow; they are not third-party certification.

Publication provenance

PUBLICATION.json binds the reviewed PDF and complete source inventory. The publication workflow checks these identities, downloads the two pinned public prototype sources, verifies their Git blob and SHA-256 identities, reruns their synthetic evidence and the separate counterexample probes, and rebuilds the manuscript without replacing its canonical PDF. The main-only release job publishes the PDF and source bundle from the qualified commit and checks their downloaded identities.

An existing edition's tag, notes and assets are preserved. Different publication bytes require a new edition or explicit investigation of the mismatch. This is a normal research release; it does not change the repository's enterprise operating authority or certify operational resilience or commercial success.

CONTINUITY Ω / CONTINUITÉ Ω — Reading companion / Compagnon de lecture 1.0.1

CONTINUITY Ω / CONTINUITÉ Ω — Reading companion / Compagnon de lecture 1.0.1

Release tag / Étiquette : continuity-omega-reading-v1.0.1
Download / Téléchargement : CONTINUITY_OMEGA_BILINGUAL_COMPANION.zip
Included paper / Article inclus : immutable edition / édition immuable 1.1.1

Download the companion / Télécharger le compagnon · English reading guide · Guide de lecture français · Paper release / Publication de l’article

English — CONTINUITY Ω

A complete offline reading path in both languages

Edition 1.0.1 corrects the displayed French title to CONTINUITÉ Ω, including the guides, help and generated reports. The English title remains CONTINUITY Ω. Technical identifiers, paths, filenames and historical source records retain their original forms.

This separate companion brings the prototype’s English and French guides, a display-only passport reader, representative raw synthetic records, and the complete paper edition 1.1.1 into one archive. Open START_HERE.md after extraction. The nominal example and the overbooked refusal each have an English and a French report beside the original JSON records.

The companion preserves the six prototype source files, the original 29-test file and all paper files byte for byte. The earlier paper release continuity-omega-v1.1.0 and companion release continuity-omega-reading-v1.0.0 remain unchanged; this new companion includes the corrected paper edition 1.1.1. This reading release does not integrate the separately owned checker-hardening work or claim that the historical checker’s known weaknesses have been repaired.

Run from the downloaded archive

From the extracted archive’s root, with Python 3.10 or later:

python3 -B scripts/package_continuity_reading.py --check .
python3 -B scripts/continuity_passport.py --in-dir examples/nominal --language en
python3 -B scripts/continuity_passport.py --in-dir examples/overbooked --language en

No package installation, network or API key is needed for these commands. The self-check verifies the complete file inventory against the included byte counts and hashes, checks the frozen source and paper identities, and executes four presentations from the extracted code: the two examples in both languages. It compares the rendered text exactly with the packaged reports. It does not run a new audit or replay the synthetic exercise.

The presenter reads saved records. Its exit code 0 means a supported report was rendered, including a recorded failure or refusal. Exit code 2 means presentation was refused. Neither a successful render nor an unchanged hash establishes current evidence validity, authentic source custody, checker correctness, live recovery, supplier availability, customer acceptance or commercial qualification. The documented historical counterexamples and source-authenticity limits still apply.

Archive boundary and reproducibility

The explicit source inventory contains:

• Six unchanged files in continuity_omega/.
• scripts/continuity_passport.py and scripts/package_continuity_reading.py.
• tests/test_continuity_omega.py and tests/test_continuity_passport.py.
• The paired README, QUICKSTART, REAL_PILOT and OUTPUTS guides, plus these bilingual release notes, in docs/continuity-omega/.
• All 31 files listed in the paper’s immutable PUBLICATION.json, plus that manifest, under docs/continuity-omega-paper/.

The generated inventory contains START_HERE.md, CHECKS.json, MANIFEST.json, and the nominal and overbooked examples under examples/. Machine identifiers, filenames, raw JSON keys and verdict tokens retain their original forms. The legacy English-only PASSPORT.md is excluded; the new reports are explicitly named PASSPORT.en.md and PASSPORT.fr.md. Website pages, social artwork and unrelated repository components are outside this archive. Paper compilation retains its separately documented requirements; it is not required to read or check this companion.

Assembly runs the original 29 tests and the included presentation tests, exercises all 12 synthetic cases, and renders them in both languages (24 presentations). The native decisions are three synthetic passes and nine refusals. CHECKS.json keeps compact counts and content identities rather than bulky execution logs. The two representative cases retain their raw records and four complete readable reports.

The archive has a fixed file order, timestamp and permissions and uses uncompressed ZIP entries to avoid compression-library differences. Identical included source bytes produce identical archive bytes regardless of the build directory or current Git commit. Output must be a new absolute directory outside the source tree, with an existing parent. For example:

python3 -B scripts/package_continuity_reading.py --output /tmp/continuity-reading-001

MANIFEST.json covers every archive file except itself. The extracted check rejects missing files, additional paths, symbolic links and changed contents. These are local consistency checks; a manifest cannot authenticate its own origin. Keep the extracted archive unchanged for checking, and write any new demos or exported reports outside it.

Français — CONTINUITÉ Ω

Un parcours de lecture complet hors ligne dans les deux langues

L’édition 1.0.1 corrige le titre affiché en français pour utiliser CONTINUITÉ Ω, dans les guides, l’aide et les rapports générés. Le titre anglais demeure CONTINUITY Ω. Les identifiants techniques, chemins, noms de fichiers et enregistrements sources historiques conservent leur forme d’origine.

Ce compagnon distinct réunit les guides du prototype en français et en anglais, un lecteur de passeport limité à la présentation, des données synthétiques brutes représentatives et l’édition complète 1.1.1 de l’article dans une seule archive. Après extraction, ouvrez START_HERE.md. L’exemple nominal et le refus pour capacité déjà réservée disposent chacun d’un rapport français et d’un rapport anglais, à côté des fichiers JSON d’origine.

Le compagnon conserve, octet par octet, les six fichiers sources du prototype, le fichier des 29 tests originaux et tous les fichiers de l’article. Les publications antérieures de l’article continuity-omega-v1.1.0 et du compagnon continuity-omega-reading-v1.0.0 demeurent inchangées ; ce nouveau compagnon inclut l’édition corrigée 1.1.1 de l’article. Cette publication de lecture n’intègre pas les travaux de renforcement du vérificateur qui relèvent d’une initiative distincte et ne prétend pas corriger les faiblesses connues du vérificateur historique.

Exécuter le lecteur depuis l’archive téléchargée

À la racine de l’archive extraite, avec Python 3.10 ou une version ultérieure :

python3 -B scripts/package_continuity_reading.py --check .
python3 -B scripts/continuity_passport.py --in-dir examples/nominal --language fr
python3 -B scripts/continuity_passport.py --in-dir examples/overbooked --language fr

Ces commandes ne nécessitent ni installation de paquets, ni réseau, ni clé d’API. Le contrôle autonome compare l’inventaire complet aux tailles en octets et aux empreintes incluses, vérifie les identités figées des sources et de l’article, puis exécute quatre présentations avec le code extrait : les deux exemples dans les deux langues. Il compare exactement les textes produits aux rapports inclus. Il ne relance aucun audit et ne réexécute pas l’exercice synthétique.

Le présentateur lit les données enregistrées. Son code de sortie 0 signifie qu’un rapport pris en charge a été présenté, y compris un échec ou un refus consigné. Le code 2 signifie que la présentation a été refusée. Ni une présentation réussie ni une empreinte inchangée n’établit la validité actuelle des preuves, l’authenticité de leur conservation, la correction du vérificateur, une reprise réelle, la disponibilité d’un fournisseur, l’acceptation par un client ou une qualification commerciale. Les contre-exemples historiques documentés et les limites d’authenticité des sources demeurent applicables.

Périmètre de l’archive et reproductibilité

La liste explicite des sources comprend :

• Les six fichiers inchangés de continuity_omega/.
• scripts/continuity_passport.py et scripts/package_continuity_reading.py.
• tests/test_continuity_omega.py et tests/test_continuity_passport.py.
• Les guides README, QUICKSTART, REAL_PILOT et OUTPUTS dans les deux langues, ainsi que les présentes notes bilingues, dans docs/continuity-omega/.
• Les 31 fichiers énumérés dans le manifeste immuable PUBLICATION.json de l’article, ainsi que ce manifeste, sous docs/continuity-omega-paper/.

Le contenu généré comprend START_HERE.md, CHECKS.json, MANIFEST.json, ainsi que les exemples nominal et de capacité déjà réservée sous examples/. Les identifiants machine, noms de fichiers, clés JSON brutes et verdicts techniques conservent leur forme d’origine. L’ancien PASSPORT.md unilingue anglais est exclu ; les nouveaux rapports portent les noms explicites PASSPORT.en.md et PASSPORT.fr.md. Les pages du site, les visuels sociaux et les autres composants du dépôt ne font pas partie de cette archive. La compilation de l’article conserve ses exigences documentées séparément ; elle n’est pas nécessaire pour lire ou contrôler ce compagnon.

L’assemblage exécute les 29 tests originaux et les tests de présentation inclus, exerce **le...

CONTINUITY Ω — Reading companion / Compagnon de lecture 1.0.0

CONTINUITY Ω — Reading companion / Compagnon de lecture 1.0.0

Release tag / Étiquette : continuity-omega-reading-v1.0.0
Download / Téléchargement : CONTINUITY_OMEGA_BILINGUAL_COMPANION.zip
Included paper / Article inclus : immutable edition / édition immuable 1.1.0

Download the companion / Télécharger le compagnon · English reading guide · Guide de lecture français · Paper release / Publication de l’article

English

A complete offline reading path in both languages

This separate companion brings the prototype’s English and French guides, a display-only passport reader, representative raw synthetic records, and the complete paper edition 1.1.0 into one archive. Open START_HERE.md after extraction. The nominal example and the overbooked refusal each have an English and a French report beside the original JSON records.

The companion preserves the six prototype source files, the original 29-test file and all paper files byte for byte. The paper’s existing continuity-omega-v1.1.0 release, PDFs and source archive remain unchanged. This reading release does not integrate the separately owned checker-hardening work or claim that the historical checker’s known weaknesses have been repaired.

Run from the downloaded archive

From the extracted archive’s root, with Python 3.10 or later:

python3 -B scripts/package_continuity_reading.py --check .
python3 -B scripts/continuity_passport.py --in-dir examples/nominal --language en
python3 -B scripts/continuity_passport.py --in-dir examples/overbooked --language en

No package installation, network or API key is needed for these commands. The self-check verifies the complete file inventory against the included byte counts and hashes, checks the frozen source and paper identities, and executes four presentations from the extracted code: the two examples in both languages. It compares the rendered text exactly with the packaged reports. It does not run a new audit or replay the synthetic exercise.

The presenter reads saved records. Its exit code 0 means a supported report was rendered, including a recorded failure or refusal. Exit code 2 means presentation was refused. Neither a successful render nor an unchanged hash establishes current evidence validity, authentic source custody, checker correctness, live recovery, supplier availability, customer acceptance or commercial qualification. The documented historical counterexamples and source-authenticity limits still apply.

Archive boundary and reproducibility

The explicit source inventory contains:

• Six unchanged files in continuity_omega/.
• scripts/continuity_passport.py and scripts/package_continuity_reading.py.
• tests/test_continuity_omega.py and tests/test_continuity_passport.py.
• The paired README, QUICKSTART, REAL_PILOT and OUTPUTS guides, plus these bilingual release notes, in docs/continuity-omega/.
• All 31 files listed in the paper’s immutable PUBLICATION.json, plus that manifest, under docs/continuity-omega-paper/.

The generated inventory contains START_HERE.md, CHECKS.json, MANIFEST.json, and the nominal and overbooked examples under examples/. Machine identifiers, filenames, raw JSON keys and verdict tokens retain their original forms. The legacy English-only PASSPORT.md is excluded; the new reports are explicitly named PASSPORT.en.md and PASSPORT.fr.md. Website pages, social artwork and unrelated repository components are outside this archive. Paper compilation retains its separately documented requirements; it is not required to read or check this companion.

Assembly runs the original 29 tests and the included presentation tests, exercises all 12 synthetic cases, and renders them in both languages (24 presentations). The native decisions are three synthetic passes and nine refusals. CHECKS.json keeps compact counts and content identities rather than bulky execution logs. The two representative cases retain their raw records and four complete readable reports.

The archive has a fixed file order, timestamp and permissions and uses uncompressed ZIP entries to avoid compression-library differences. Identical included source bytes produce identical archive bytes regardless of the build directory or current Git commit. Output must be a new absolute directory outside the source tree, with an existing parent. For example:

python3 -B scripts/package_continuity_reading.py --output /tmp/continuity-reading-001

MANIFEST.json covers every archive file except itself. The extracted check rejects missing files, additional paths, symbolic links and changed contents. These are local consistency checks; a manifest cannot authenticate its own origin. Keep the extracted archive unchanged for checking, and write any new demos or exported reports outside it.

Français

Un parcours de lecture complet hors ligne dans les deux langues

Ce compagnon distinct réunit les guides du prototype en français et en anglais, un lecteur de passeport limité à la présentation, des données synthétiques brutes représentatives et l’édition complète 1.1.0 de l’article dans une seule archive. Après extraction, ouvrez START_HERE.md. L’exemple nominal et le refus pour capacité déjà réservée disposent chacun d’un rapport français et d’un rapport anglais, à côté des fichiers JSON d’origine.

Le compagnon conserve, octet par octet, les six fichiers sources du prototype, le fichier des 29 tests originaux et tous les fichiers de l’article. La publication existante continuity-omega-v1.1.0, ses PDF et son archive de sources demeurent inchangés. Cette publication de lecture n’intègre pas les travaux de renforcement du vérificateur qui relèvent d’une initiative distincte et ne prétend pas corriger les faiblesses connues du vérificateur historique.

Exécuter le lecteur depuis l’archive téléchargée

À la racine de l’archive extraite, avec Python 3.10 ou une version ultérieure :

python3 -B scripts/package_continuity_reading.py --check .
python3 -B scripts/continuity_passport.py --in-dir examples/nominal --language fr
python3 -B scripts/continuity_passport.py --in-dir examples/overbooked --language fr

Ces commandes ne nécessitent ni installation de paquets, ni réseau, ni clé d’API. Le contrôle autonome compare l’inventaire complet aux tailles en octets et aux empreintes incluses, vérifie les identités figées des sources et de l’article, puis exécute quatre présentations avec le code extrait : les deux exemples dans les deux langues. Il compare exactement les textes produits aux rapports inclus. Il ne relance aucun audit et ne réexécute pas l’exercice synthétique.

Le présentateur lit les données enregistrées. Son code de sortie 0 signifie qu’un rapport pris en charge a été présenté, y compris un échec ou un refus consigné. Le code 2 signifie que la présentation a été refusée. Ni une présentation réussie ni une empreinte inchangée n’établit la validité actuelle des preuves, l’authenticité de leur conservation, la correction du vérificateur, une reprise réelle, la disponibilité d’un fournisseur, l’acceptation par un client ou une qualification commerciale. Les contre-exemples historiques documentés et les limites d’authenticité des sources demeurent applicables.

Périmètre de l’archive et reproductibilité

La liste explicite des sources comprend :

• Les six fichiers inchangés de continuity_omega/.
• scripts/continuity_passport.py et scripts/package_continuity_reading.py.
• tests/test_continuity_omega.py et tests/test_continuity_passport.py.
• Les guides README, QUICKSTART, REAL_PILOT et OUTPUTS dans les deux langues, ainsi que les présentes notes bilingues, dans docs/continuity-omega/.
• Les 31 fichiers énumérés dans le manifeste immuable PUBLICATION.json de l’article, ainsi que ce manifeste, sous docs/continuity-omega-paper/.

Le contenu généré comprend START_HERE.md, CHECKS.json, MANIFEST.json, ainsi que les exemples nominal et de capacité déjà réservée sous examples/. Les identifiants machine, noms de fichiers, clés JSON brutes et verdicts techniques conservent leur forme d’origine. L’ancien PASSPORT.md unilingue anglais est exclu ; les nouveaux rapports portent les noms explicites PASSPORT.en.md et PASSPORT.fr.md. Les pages du site, les visuels sociaux et les autres composants du dépôt ne font pas partie de cette archive. La compilation de l’article conserve ses exigences documentées séparément ; elle n’est pas nécessaire pour lire ou contrôler ce compagnon.

L’assemblage exécute les 29 tests originaux et les tests de présentation inclus, exerce les 12 cas synthétiques et les présente dans les deux langues (24 présentations). Les décisions d’origine comprennent trois réussites synthétiques et neuf refus. CHECKS.json conserve des décomptes et des identités de contenu compacts, sans journaux d’exécution volumineux. Les deux cas représentatifs conservent leurs données brutes et leurs quatre rapports lisibles complets.

L’archive utilise un ordre des fichiers, un horodatage et des permissions fixes, ainsi que des entrées ZIP non compressées afin d’éviter les différences entre bibliothèques de compression. Des sources incluses identiques produisent des archives identiques, quels que soient le répertoire d’assemblage et le commit Git courant. La sortie doit être un nouveau répertoire absolu à l’extérieur des sources, dont le parent existe. P...