Confirmed Game Prototype Test — Tactical Vertical Slice
Alignment date: 2026-06-15. Status: confirmed. The page is current for the completed alignment cycle; later research may amend it only by archiving this confirmed page and highlighting the amendment.
Confirmation Status
confirmedprototype tiercanonical artifact writtenrepo-only evidence
Canonical artifact: research/unthinkable/game-prototype-test.md. Previous review page archived at docs/history/archive/2026-06-15/193916/alignment/game-prototype-test-tactical-vertical-slice.html.
Approved route: directly to prototype implementation. No task-file mutation was made because the approved file-change gate limited this pass to the artifact, review page, and index update.
Table of Contents
Approval Record
| Gate | Decision | Impact |
|---|---|---|
| Scope / non-goals | Approve as proposed | Use the fixed tactical vertical-slice scope without reopening the battle contract. |
| Evidence coverage | Approve repo-only source coverage | Use confirmed internal artifacts as the evidence base. |
| Assumptions / confidence | Accept all assumptions as the Stage-2 baseline | AI, save/pause, deck x command-post legibility, and browser performance stay mandatory validation gates. |
| Coverage checkpoint | Approve the Stage-2 review format | Use the proposed sections: scope, questions, script, checklist, criteria, evidence, and file changes. |
| Artifact destination | Approve working and canonical paths | Write research/unthinkable/game-prototype-test.md. |
| Proposed file changes | Approve Stage-2 working packet, review page, and index updates only | Canonical artifact, confirmed page, archive, and index only; no task file changes. |
| Post-approval route | Route directly to prototype implementation | Next work is a runnable tactical prototype, not another research step. |
alignment_page: "alignment/game-prototype-test-tactical-vertical-slice.html" response_status: "complete" approval_status: "ready-for-agent-review" required_gate_status: "complete" unanswered_required_questions: []
Prototype Thesis
Omega War's first playable prototype should prove one claim: a browser-first tactical battle is fun and readable when a fixed division deck expresses identity, while command-post construction decides when that identity arrives.
| Field | Prototype decision |
|---|---|
| Scenario | Conquest of Helmstedt working title |
| Mode | Single-player tactical battle against baseline AI |
| Player / AI | UK/Commonwealth 11th Armoured Division vs Soviet 2nd Guards Tank Army |
| Map | One Central Germany tactical map with village edge, road junction, tree lines, low ridges, and five capture points |
| Required systems | Band A/B/C exposure, one telegraphed wonder, pause/save/autosave/reload/restart, and perf logging |
Smallest Valid Scope
Include
- One fixed map with five capture points.
- Open, light-cover, and heavy-cover states.
- Normal, pressured, suppressed, and pinned infantry states.
- Fixed UK and Soviet Band A/B/C deck fixtures.
- Visible command-post surfaces that unlock bands and support categories.
- Baseline AI capture, escalation, defense, and retreat behavior.
- Pause, manual save, visibility autosave, reload, and restart.
Exclude
- Campaign map, formation tokens, auto-resolve, and strategic persistence.
- Full deck builder or all playable divisions.
- Final combat balance, card counts, prices, cooldowns, and trait magnitudes.
- Production art, audio, VFX, cinematic intro, and final UI styling.
- Multiplayer, matchmaking, network determinism, and cloud saves.
Core Test Questions
| Question | Measurement | Failure signal |
|---|---|---|
| Is the tactical fight readable and fun at company scale? | Player uses cover, reacts to suppression, contests capture points, and explains outcomes. | Combat reads as visual noise or unexplained attrition. |
| Does command-post escalation make the deck feel authored? | Player recalls what unlocked Band B/C and why build order mattered. | Bands feel like a hidden timer or the base feels like busywork. |
| Does AI create pressure without feeling unfair? | AI captures, contests, escalates, defends, preserves units, and uses prerequisites honestly. | AI is irrelevant, one-lane predictable, or opaque and unfair. |
| Does the session survive browser reality? | Pause/save/autosave/reload preserve exact tactical state. | Reload corrupts command posts, captures, resources, entities, or cooldowns. |
| Is replay desire present before campaign context? | Player wants to retry a build order or counter the AI differently. | Player says the battle is flat or only worth playing with campaign rewards. |
| Does the browser unit ceiling hold? | Perf log captures frame time, entity count, and worst spikes during Band C pressure. | Responsiveness collapses or the entity ceiling stops feeling company-scale. |
Playtest Script
Use one observer/moderator and one player. Explain only the scenario objective: capture territory, build command posts, escalate the force, and win the battle. Do not explain the hidden design hypothesis unless the player is blocked.
| Stage | Player task | Moderator observes |
|---|---|---|
| Opening | Deploy Band A, scout, capture nearest points, find cover. | Can the player read cover, suppression, capture ownership, and resources? |
| First escalation | Choose and build an early command-post surface. | Does the player see a deck/band unlock rather than generic base clutter? |
| Mid fight | Use Band B units to contest center and answer the AI. | Does Band B feel like authored identity arriving through player decision? |
| AI pressure | Let the AI attack, contest, and escalate. | Is pressure readable, fair, and recoverable? |
| Save drill | Pause, save, reload, and resume; trigger visibility autosave if practical. | Is restored state faithful? |
| Late escalation | Complete Advance Command Post and trigger or counter one wonder. | Does the event feel earned, telegraphed, and answerable? |
| Finish | Play to win/loss or hard stop at 25 minutes. | Does the session resolve cleanly and create replay intent? |
Post-Session Interview
- What were you trying to do in the first five minutes?
- What did building command posts change for your army?
- When did you feel in control? When did you feel lost?
- Which unit or card felt most like your division?
- Did the AI pressure feel fair, too weak, or unfair?
- What did the save/reload interruption do to your confidence in the session?
- Would you replay this scenario to try a different build order or preserve units better?
- What one thing should be cut, kept, or amplified before adding campaign context?
Observation Checklist
Tactical Readability
- Moves infantry into cover without prompting.
- Notices suppression or pinning and changes behavior.
- Can explain at least one unit loss or survival.
Deck x Command Post
- Notices Band A/B/C availability changes.
- Makes an intentional escalation tradeoff.
- Recognizes an AI escalation cue.
AI Pressure
- Captures nearby points.
- Contests center.
- Builds Band B and Band C paths.
- Defends damaged command posts.
Browser Resilience
- Pause stops simulation.
- Manual save restores exact state.
- Visibility autosave does not corrupt state.
- Late battle remains responsive.
Success Criteria
| Level | Criteria |
|---|---|
| Pass | Most players finish or nearly finish in 15-25 minutes; explain cover, suppression, and capture; understand command-post band unlocks; express replay intent; do not call AI unfair or irrelevant; save/reload succeeds; browser responsiveness holds. |
| Concern | Players like combat but ignore escalation; understand bands only after explanation; AI works in one lane but collapses elsewhere; save works but players distrust it; performance works only below company-scale feel. |
| Fail | Players cannot explain unit outcomes; command-post escalation reads as timer or busywork; AI is steamrolled or unfair; save/reload corrupts state; campaign rewards are the only stated reason to keep playing. |
Cut / Keep / Amplify
| Outcome | Decision |
|---|---|
| Combat readable, command-post loop readable, replay intent present | Keep the fused deck x command-post loop and harden implementation. |
| Combat readable, command-post loop confusing | Amplify unlock telegraphs, command-post UI, build-order contrast, and enemy escalation tells. |
| Command-post loop understood, combat unclear | Amplify cover/suppression feedback, unit states, damage causality, and capture signposting. |
| AI weak but loop readable | Keep scope and prioritize AI capture, escalation, and retreat fixes before campaign work. |
| AI unfair or opaque | Cut hidden advantages; add telegraphs and prerequisite visibility. |
| Save/reload unreliable | Cut further feature work until state restoration is fixed. |
| Browser performance fails | Cut entity ceiling, visual fidelity, or simulation frequency first; consider wrapper only after web optimization. |
| Replay desire absent | Reopen core-loop and fantasy assumptions before expanding campaign design. |
Evidence Matrix
| Claim | Evidence | Inference | Confidence | Decision impact |
|---|---|---|---|---|
| Center deck x command-post escalation. | tactical-vertical-slice-spec.md; game-genre-map.md. | The fusion is the riskiest fun question. | High | Make command-post comprehension a primary gate. |
| AI quality is a slice gate. | game-audience.md; game-genre-map.md. | A sandbox without AI pressure cannot validate the audience promise. | High | Include baseline AI in smallest valid scope. |
| Save/pause cannot be deferred. | Audience and genre-map comparable complaints; browser interruption risk. | Interruption recovery is part of browser-first viability. | High | Require save/reload drill in every session. |
| Campaign systems stay out. | game-genre-map.md; game-fantasy.md; tactical spec. | Campaign rewards would mask whether the battle works alone. | High | Exclude campaign map and persistence economy. |
| Fixture values remain provisional. | unit-data-schema.md. | Shape can be validated without freezing balance. | High | Keep numeric tuning non-binding. |
| Browser performance is a go/no-go unknown. | game-genre-map.md; game-audience.md. | Need concrete measurement before platform commitment. | Medium-high | Log responsiveness and entity ceiling during late battle. |
Prototype Implementation Handoff
The approved route is direct prototype implementation. Begin with the smallest vertical thread that can run the playtest script:
- Map shell with capture points, cover volumes, deployment zones, and camera.
- Unit runtime from provisional card fixtures: movement, basic weapons, health, suppression, and capture.
- Command-post unlock chain for Band A/B/C and support surfaces.
- Baseline AI capture/escalation loop.
- Save/pause/reload state contract.
- Local observation/performance log.
- Moderator-ready playtest build and reset controls.
Do not add campaign systems, full deck construction, or extra factions before the above path can complete a 15-25 minute session.