Game Genre Map - Omega War

1.1 Must-prove-first slice items

1. Cover / suppression / capture combat feels tactical at company scale.
2. Build-order phasing through Forward Command Post tiers reads as authored escalation, not a timer or a chore.
3. A baseline-competent AI that does not get "steamrolled or slaughtered."
4. Mid-battle save / pause works.
5. The chosen unit ceiling holds 60fps in a browser.

Everything else - campaign map, multiplayer, wonder-weapon balance, refit culture, the other six decks - is deferred behind these.

"Must-have" = the slice is not the slice without it. "Prototype" = a scoped version whose job is to test an open hypothesis. "Defer" = out of slice scope, recorded for later.

The novel fusion is deck-as-roster gated by a build-order-driven command post. Rules of thumb, tied to the approved three-band deck contract (roster §6.1):

1. Deck defines the roster; the base decides when each part arrives. Cards are authored into Band A (Vanguard), Band B (Main Body), Band C (Commitment). The deck is identity; the build order is tempo. This replaces Steel Division's phase timer.
2. Each command-post surface unlocks a band, not a unit. Forward Command Post → Band A; Motor Pool / Field Workshop / Artillery Battery → Band B; Advance Command Post / Air Liaison / Army Group Liaison → Band C (roster §7.2). Advancing a band is an explicit, expensive, interruptible action - escalation is authored, never a countdown.
3. Legible counterplay before power. Recon/screens (Band A) precede armor (Band B) which precede heavy/commitment (Band C), so an opponent can read a build and respond - the "if you see 82nd Airborne you know their gameplan" principle expressed in tempo.
4. Force deck identity; punish convergence. Each deck needs a forced strength and a forced weakness (armoured = thin infantry; rifle = low mobility; artillery = maneuver-weak; airborne = low staying power, hit-and-run via smoke/stealth). The documented punishment is shallow/homogenized customization, not the presence of decks. These forced strengths/weaknesses can be surfaced as named trait tags (shock / commando / militia / leader) - currently a Prototype-tier convention (§3).
5. One wonder per deck, Band C only, anti-rush. A late-game pacing valve, not a tech-rush goal: expensive command-post prerequisite, resource cost, visible telegraph, long cooldown/one-use, and a battlefield-control or campaign-progress gate (roster §6.3, §8.3).
6. No free off-map power. Artillery/air must be invested through Artillery Battery / Air Liaison surfaces, so those decks visibly commit rather than receiving universal-button power.
7. Deployment position is earned. Following WiC reinforcement-point logic, holding capture points improves where/when reinvested force arrives - tying the capture loop to the band loop.

Guarded by G-A1: serve the slice, not a full campaign spec. This records the principles the later war map must honor; it does not design the campaign.

5.1 Strategic-layer handoff requirements (the "layer handoff" beat)

1. Battles must matter (anti layer-divorce). The genre's signature failure is the strategic layer feeling like "an entirely other game" (SD2 AG) or offering "zero incentive to actually fight the battles" (WARNO AG). The map and battle must be one war.
2. Auto-resolve must be fair but never strictly superior to manual. In WARNO AG players "play on auto-resolve only" because manual fighting is a casualty tax. Fighting manually should reward (fewer losses, intact captures, veterancy, objective bonuses); auto-resolve is the convenience option, not the optimum.
3. Persistence surfaces should be visible and personal. EndWar's Theater of War keeps an upgradeable persistent army on a persistent map; Omega War's equivalent is veteran units, deck refit state, and front position the player sees carry forward. The slice-level mechanic that feeds this is in-battle veterancy (a must-have, §3): units earn experience in combat and carry it forward, while the broader refit/upgrade economy stays a campaign-layer deferral.
4. Front emerges from formation positions, not region flips. Front lines form where formation tokens stand (EndWar/Wargame:RD style); collisions create battles.

5.2 Explicit deferrals (recorded, not designed now)

Map representation (regions vs node graph vs free movement), formation-token granularity (company vs battalion), real-time vs turn-based map pacing, the collision→battle handoff rules, air/naval mission mechanics, narrative deployment events, and the persistence/upgrade economy - all deferred to a dedicated campaign-layer design pass after the slice.

INF = inference flag (conclusion drawn from evidence rather than directly stated). Confidence: High / Med / Low.

9.1 Assumption & confidence register

Inherited risky assumptions still open: A-R1 (desktop wargamers try a browser deep RTS), A-R3 (alt-history fans convert to real-time play), A-R4 (deck x base reads as depth not homogenization - the central prototype risk).

9.2 Alternatives considered + rejected / lower-confidence findings

• Comparable-teardown deliverable shape - rejected per GMQ2 = decision-map. Teardowns are an input here, not the output.
• World in Conflict no-base / reinforcement-point model - elegant pacing, but deletes the base x deck wedge (E8). Borrowed only as capture→deployment logic (§4.7).
• Steel Division timer-driven phases - rejected for player-authored build-order phasing (E12); the timer is the convention being bent.
• US 2nd Armored vs 2nd Guards Tank slice - a strong "more American first impression" alternate, but UK 11th Armoured vs 2nd Guards Tank matches the recorded slice suggestion and the Comet-vs-IS-3 "1946 arsenal gets its war" promise (roster §4.1). Carried as the recorded alternate.
• Men of War direct-control sim depth - rejected for readability/scope (E9).
• Lower-confidence / inference-flagged: E8, E9 (INF - design judgments); E10, E11, E13, E14 (Med - depend on unproven browser-depth and legibility hypotheses). Kept visible, not laundered into the decision map as facts.

9.3 Source coverage by category + coverage gaps

All five evidence families swept (GMQ4 exhaustive): (1) Tactical RTS/RTT - CoH1-3, World in Conflict, Men of War/Gates of Hell; (2) Deck/division wargames - Wargame:RD, WARNO, SD2, Broken Arrow; (3) Operational campaign layers - WARNO AG, SD2 AG, Wargame campaigns, EndWar; (4) Browser/low-friction - OpenFront, Three.js perf, CrazyGames (carried); (5) Approved internal artifacts - idea-brief, game-audience, game-fantasy, arsenal-1945, divisions-1945, roster.

Coverage gaps (recorded, not papered over): no first-party SP-vs-MP splits exist for any comparable; no direct browser-delivery sentiment quotes from desktop wargamers (A-R1); no shipped precedent for build-order-authored phasing fused with deck-as-roster (the wedge is genuinely novel - E12/E13 rest on prototype testing); exact browser unit ceiling is unmeasured (E11); campaign conventions (E14) are sketched from EndWar/AG analogs only - a dedicated campaign genre pass is deferred.

10.1 What this hands to the tactical vertical-slice spec

• A prioritized convention list (§3): build the must-haves; scope the prototypes as explicit hypothesis tests; ignore the defers.
• The deck x base rules of thumb (§4) as the spec's core mechanic contract.
• The anti-pattern register (§7) as a "what not to do" checklist - AI competence, save, onboarding, avoiding layer divorce when the campaign arrives.
• The browser/solo scope controls (§6), including the perf-spike dependency.

10.2 Why the chosen route is unit-data-schema (recorded, not executed)

The schema inputs are already gathered - arsenal §4A (unit-type taxonomy, WARNO-card style, ACCEPTED) and §3 (per-faction equipment tables), plus divisions §4 (To&E patterns → card-class proportions, explicitly labeled "schema inputs"). A unit/division data schema is the shared substrate for the tactical layer, the deck builder, and the campaign map (idea-brief next steps), so drafting it next unblocks the most downstream work with inputs in hand. This packet only records the route; it does not author the schema. (The plan referenced "arsenal §5.3"; the actual schema-input sections are arsenal §4A/§3 and divisions §4 - corrected here.)

10.3 Proposed canonical changes (Stage-3 outline only)

On Stage-3 approval, research/game-genre-map.md would contain: the genre-family map with stances; the must-have/prototype/defer convention table; deck x base rules; operational handoff principles + deferrals; browser/solo scope controls; the anti-pattern register; the evidence matrix + confidence/assumption register; alternatives, lower-confidence findings, source coverage + gaps; and the downstream implications including the unit-data-schema route rationale. It would not set combat stats, prices, exact card counts, the data schema itself, build/resource numbers, or final balance.

Proposal only. Terms are appended to research/glossary.md only if approved at Stage 3. layer handoff already exists and defines "layer divorce" inline, so a standalone entry is offered as a cross-reference convenience.

Changed Updated 2026-06-13: added unit trait tag (the Prototype-tier buff/debuff convention from §3) - now four proposed terms.

All five evidence families swept; GMQ1-GMQ6 each answered with a dedicated section; every load-bearing claim carries a source + confidence; assumptions carried and extended; alternatives, lower-confidence findings, and gaps recorded. No novel genre family or convention class surfaced in the final batch beyond the internal artifacts → exhaustive-sweep stopping rule honored for the decision-map scope.

Totals: 10 searches, 0 verification fetches (claims grounded in search result excerpts + carried internal evidence). Internal artifacts supplied guardrails and carried CoH3/Broken Arrow/SD2/Stormgate/OpenFront evidence.

14. Compile Approval YAML

Answer the required gates above, add optional section feedback, then compile and copy the YAML into the next execution message. You may compile partial responses.