21-slide public overview Open white paper

Platinum, Aurora, And The Conformance Project

A browser-arcade platform, evidence-led game ingestion program, and release discipline for AI-assisted software.

White paperv0.4.1-draft

Updated2026-06-07

Lanedevelopment

Build1.4.1.1

Slides21

Open white paper Open white-paper PDF Open slide metadata Open project guide Open conformance dashboard

01Project overview

Platinum, Aurora & The Conformance Project

A real game platform, a playable first game, and an evidence program for making AI-assisted software accountable.

Proof 1 Aurora release gate: 8.8/10

Proof 2 Challenge set pieces: 4.3/10

Proof 3 Measured runs: 904

Proof 4 Artifact growth: 1.48GB

Project overview v0.4.1-draft · 2026-06-07 · 1.4.1.1

02Thesis

The Claim: Fast AI-Assisted Work Can Still Be Accountable

The project forces model-assisted iteration to leave evidence, checks, and public release artifacts behind.

Proof 1 Real releases, not only demos

Proof 2 Reference-driven fidelity work

Proof 3 Visible platform boundaries

Proof 4 Local CPU/browser harnesses for repeated measurement

Thesis v0.4.1-draft · 2026-06-07 · 1.4.1.1

03History

The Arc Moved From One Game To A Conformance Program

Each release family made the project harder to fake and easier to review.

Proof 1 1.0.0: first public Aurora

Proof 2 1.2.0: Platinum platform framing

Proof 3 1.4.0: multi-game and conformance baseline

Proof 4 Next: conformance scale across games

History v0.4.1-draft · 2026-06-07 · 1.4.1.1

04System model

The Operating Model Is Five Layers, Not One Blob

Quality improves when reference evidence, ingestion, games, harnesses, and release claims have separate ownership.

ReferenceVideos, manuals, audio packs, sprite sheets

IngestionManifests, windows, event logs, tracks

RuntimeGame-owned implementation and captures

HarnessScorecards, personas, correspondence checks

ReleaseDocs, gates, lanes, public claims

System model v0.4.1-draft · 2026-06-07 · 1.4.1.1

05Operating roles

Recurring Roles Keep The Work Assignable

The project uses named human, agent, machine, and build-process roles so quality work can be assigned, checked, and handed off without becoming ad hoc.

Proof 1 Manager sets direction and stop/go constraints

Proof 2 Developer executes scoped changes and evidence

Proof 3 Architect turns repeated pain into reusable mechanisms

Proof 4 Release authority controls hosted lane claims

Proof 5 Automated gates refresh review, security, conformance, and documentation evidence

Proof 6 Player-visible review decides whether measured changes actually improve the game

Operating roles v0.4.1-draft · 2026-06-07 · 1.4.1.1

06Architecture

Platinum Hosts; Game Packs Carry Game-Specific Authority

The platform scale bet only works if Aurora-specific truth does not leak into Platinum or the next game.

Proof 1 Platform owns shell, services, and lane model

Proof 2 Aurora owns Galaga-like conformance

Proof 3 Guardians owns Galaxian-like preview truth

Proof 4 Architecture review says dev/beta separation is good but still worth hardening

Architecture v0.4.1-draft · 2026-06-07 · 1.4.1.1

07Portfolio

Current Game Portfolio: First Game, Second Proof, Third Intake

The platform matters because it can host genuinely different game instances, not just skins.

Proof 1 Aurora: first shipped application

Proof 2 Galaxy Guardians: second-game proof

Proof 3 Windigo / third game: intake lane

Proof 4 Conformant reference modes first, unique public themes later

Portfolio v0.4.1-draft · 2026-06-07 · 1.4.1.1

08Ingestion

Ingestion Turns Examples Into Implementation Contracts

External examples become source manifests, windows, annotations, runtime captures, scores, and release gates.

Proof 1 Preserved source package

Proof 2 Extracted clip or frame window

Proof 3 Semantic annotation

Proof 4 Runtime capture

Proof 5 Score and gate

Ingestion v0.4.1-draft · 2026-06-07 · 1.4.1.1

09Evidence

The Reference Corpus Expanded Beyond One Remembered Target

Recent source recovery makes claims more defensible and exposes gaps more sharply.

Proof 1 Galaga challenge videos, audio cues, sprite/walkthrough bundles

Proof 2 Galaxian no-voiceover and full-session gameplay

Proof 3 Space Invaders early preserved-source lane

Proof 4 Cabinet, manual, rule, and history sources

Evidence v0.4.1-draft · 2026-06-07 · 1.4.1.1

10Conformance

The Score Is A Measurement Surface, Not A Trophy

Scores carry confidence and resolution; a better scorer can lower a score while improving honesty.

Proof 1 Overall quality gate: 8.8/10

Proof 2 Audio identity: 7.3/10

Proof 3 Challenge set pieces: 4.3/10

Proof 4 Runtime sprite conformance: 6.2/10

Proof 5 Visual look and feel: 8.6/10

Conformance v0.4.1-draft · 2026-06-07 · 1.4.1.1

11Aurora state

Aurora Is Playable And Much More Honest Than It Was

The candidate is strong enough for review, but not yet for an easy 1.4.1 production promise.

Proof 1 Good overall roll-up

Proof 2 Audio near the next gate

Proof 3 Challenge-stage scorer now strict

Proof 4 Highest-value next gameplay step is movement grammar

Aurora state v0.4.1-draft · 2026-06-07 · 1.4.1.1

12Case study

Challenge-Stage Work Improved The Process Before It Improved The Game

This is a measurement and ingestion success, but still a gameplay blocker.

4.3/10Strict challenge-stage set-piece score.

40/40Reference-backed challenge groups.

8.6/10Target trajectory-control readiness.

0Release-ready challenge contracts.

3.6/10Target-video object-track fit.

NextMovement grammar plus human-perfect guard.

Case study v0.4.1-draft · 2026-06-07 · 1.4.1.1

13Gameplay grammar

The Next Unlock Is A Reusable Movement Grammar

Challenge stages and normal-stage entries should use the same contract language for group arrivals, paths, timing, and player opportunity.

Proof 1 Group contracts

Proof 2 Path primitives

Proof 3 Timing contracts

Proof 4 Human-perfect guard

Proof 5 Motion identity

Proof 6 Game-specific variation

Gameplay grammar v0.4.1-draft · 2026-06-07 · 1.4.1.1

14Audio

Audio Is Valuable, But The Economics Tell Us To Be Disciplined

Audio clarity matters, but current tuning loops are expensive for modest score movement.

Proof 1 Audio identity: 7.3/10

Proof 2 Audio consumed large local-compute blocks

Proof 3 Audio loop around 39 minutes per positive score point

Proof 4 Next gain should come from better evaluators or more stable cue windows

Audio v0.4.1-draft · 2026-06-07 · 1.4.1.1

15Sprites

Sprites Need Three Separate Conformance Concepts

One static sprite score hides the work players actually perceive.

Proof 1 Catalog proxy score

Proof 2 Runtime canvas crop score

Proof 3 Active motion planning

Proof 4 Scale and formation context

Proof 5 Frame-labeled motion windows needed

Sprites v0.4.1-draft · 2026-06-07 · 1.4.1.1

16Personas

Personas Turn Gameplay Assessment Into Repeatable Evidence

Beginner, intermediate, professional, and expert runs help expose score, survival, and learnability changes.

Proof 1 Generic across games

Proof 2 Distribution matters more than one run

Proof 3 Challenge-tour/watch mode accelerates human review

Proof 4 Capture/rescue persona behavior still needs richer strategy

Personas v0.4.1-draft · 2026-06-07 · 1.4.1.1

17Economics

The Economics Ledger Keeps Us Honest About Spend Versus Return

The project tracks CPU, browser, Codex/GPU-equivalent time, artifact growth, and score movement.

Proof 1 904 measured runs

Proof 2 971.3 tracked wall minutes

Proof 3 576.5 CPU wall minutes

Proof 4 429.6 browser wall minutes

Proof 5 630.8 GPU/Codex-equivalent wall minutes

Proof 6 1.48GB artifact evidence

Economics v0.4.1-draft · 2026-06-07 · 1.4.1.1

18Release

Release Discipline Is Part Of The Product

The lane model prevents a passing build from becoming an exaggerated public promise.

Proof 1 Local: fast iteration

Proof 2 Dev: hosted review

Proof 3 Beta: authority-published candidate lineage

Proof 4 Production: stable public claim

Proof 5 MacBook is release authority but production remains conservative

Release v0.4.1-draft · 2026-06-07 · 1.4.1.1

19Review

Professionalism Also Means Review Records And Issue Closure

Code, architecture, documentation, security, and storage reviews are now artifacts rather than vibes.

Proof 1 Code review packets

Proof 2 Architecture review

Proof 3 Security/auth/storage gates

Proof 4 Documentation freshness

Proof 5 Review learning ledger

Review v0.4.1-draft · 2026-06-07 · 1.4.1.1

20Near-term path

Next 90 Days: Make Quality Movement Visible And Release-Worthy

The near-term plan is quality conversion, not just more artifacts.

Proof 1 Challenge movement grammar

Proof 2 Sprite motion windows

Proof 3 Guardians first-class evidence

Proof 4 1.4.1 candidate only when improvements are player-visible

Proof 5 Production story only when release notes can defend the claim

Near-term path v0.4.1-draft · 2026-06-07 · 1.4.1.1

21Longer path

The Longer Goal: A Conformance Factory, Then Original Variations

First prove true-to-reference modes; then make public games unique with confidence.

Proof 1 Ingest examples

Proof 2 Conform reference mode

Proof 3 Vary themes and movement

Proof 4 Scale across games

Proof 5 Publish an honest AI-assisted software story

Longer path v0.4.1-draft · 2026-06-07 · 1.4.1.1

Source Artifacts

WHITE_PAPER.md
REFERENCE_MEDIA_INVENTORY.md
GAME_CONFORMANCE_CATALOG.md
white-paper/PROJECT_ROLES.md
RELEASE_CONFORMANCE_DASHBOARD.md
CONFORMANCE_ECONOMICS.md
PROJECT_STATE_AND_CONFORMANCE_PROGRAM.md
GALAGA_TARGET_ARTIFACT_COVERAGE.md
LEVEL_VISUAL_TIMING_ALIGNMENT.md
AURORA_SPRITE_MOTION_CORRESPONDENCE.md
reference-artifacts/analyses/release-conformance-dashboard/latest.json
reference-artifacts/analyses/conformance-economics/latest.json
reference-artifacts/analyses/challenge-stage-conformance/latest.json
reference-artifacts/analyses/challenge-setpiece-contracts/latest.json
reference-artifacts/analyses/challenge-trajectory-controls/latest.json