Faction Asymmetry — different but fair

What it is. Sides that share almost nothing — different units, different economies, different ways to win — yet remain balanced against each other. The StarCraft achievement: Terran, Zerg, and Protoss have no unit in common and play like three different games, and the matchups are still fair. This is the hardest, highest-value system in competitive design, and the flagship companion to world-faction-identity (the fiction) and system-counter-systems (the math).

Player fantasy / why it’s fun. Mastery of a way of being. Picking Zerg isn’t picking a color — it’s committing to swarm-and-macro, to a whole thesis about how war works. Asymmetry turns faction choice into an identity and every matchup into a genuinely new problem. Mirror factions give you variety; asymmetric factions give you depth times N.

When to use / when NOT

The honest cost: N asymmetric factions is not N× a mirror — it’s ~N² matchups to balance (each pair, from both sides). Budget for it or ship fewer, deeper sides.

Variants — where the asymmetry lives

The strongest identities bend two axes at once — a distinct economy and a distinct win condition — so the faction feels like a different game, not a different army. That’s the process-the-twist “X but Y” applied per side.

The balance doctrine — how “different” stays “fair”

Symmetry makes fairness free; asymmetry makes you earn it. Four disciplines:

1. Balance on outcomes, not stats. You cannot compare a Zergling to a Marine — they don’t share a scale. You compare matchups: does faction A, played well, win ~50% vs faction B, played well? Fairness is a win-rate window across the matrix, not stat parity (FUN.md §11’s window logic, applied to sides).
2. Every faction needs an answer to every threat — a different answer. Asymmetry is not “A has no counter to air.” Each side must be able to respond to each strategy, through its own tools (system-counter-systems across the mirror). A hole isn’t identity; it’s a broken matchup.
3. Trade strengths, don’t gift them. A faction strong early must be weak somewhere the opponent can exploit — a cheap early economy pays with a fragile mid-game. Symmetric power with asymmetric timing is the safest asymmetry.
4. Design the cross-matchups, not the units. Balance TvZ, not the Marine. The unit is a means; the matchup is the product. First-cut asymmetry always balances each faction in a vacuum and ships a broken pairing.

The StarCraft lesson. Three rosters, zero shared units, one of the most enduring competitive balances ever — because Blizzard tuned matchups over years, not units in isolation, and because every faction could answer every strategy in its own idiom. Asymmetry is a balance process, not a balance state.

Tuning levers

How it wires to Hayao

• Matchups are cross-faction duels. Pure state + world.rng lets you clone the world and run faction-A-comp vs faction-B-comp headlessly (FUN.md law 7). Assert the intended line of each side wins its share; assert every counter edge across the mirror wins its NvN duel (FUN.md §9). This is the direct fairness proof.
• Mass pathing is shared, rosters aren’t. Flow fields (one BFS per goal tile, cached outside world.state) give every faction wall-aware mass movement for free (FUN.md §9). Grep docs/API.md for astarGrid/floodFill before citing pathing.
• Strategy is the balance test (FUN.md §9): the intended line (e.g. turtle→counterpush) beats attack-move — assert per faction, so a side isn’t secretly all-in-only.
• Economy asymmetry wires through system-economy — different faucets/sinks, same solvency requirement; pace each with its own inequality (FUN.md law 3).
• The fiction of the sides is world-faction-identity; the unit vocabulary is system-unit-rosters. This module owns only the balance.

Fails when…

• Palette-swap “asymmetry.” Same units, different colors — you paid nothing and gained nothing. If the twist is cosmetic, it’s not a faction (process-the-twist cosmetic trap).
• A capability hole. “Faction A just can’t deal with air” isn’t flavor, it’s a lost matchup. Every side answers every threat in its own way.
• Balanced-in-a-vacuum. Each faction tuned alone, matchups never simulated → one pairing is 80/20 and the meta collapses to a single faction.
• Unteachable distance. So different the player can’t transfer any skill, so most players main one side and the others rot.
• The strongest side is also the easiest. Power and low skill floor on one faction kills the ladder. If a side is strong, make it demanding.

Verify

• FUN.md §9 — every counter edge wins its NvN duel; commander bot wins with the intended line; walled units route around; ms/step at peak.
• FUN.md §11 window logic — matchup win-rates sit inside a fairness window (both edges break CI: 80/20 and an unlosable 99/1).
• FUN.md law 2 — per faction, the intended strategy beats a null (attack-move / do-nothing) strategy.
• Determinism: golden hash of a scripted cross-faction match.

Composes with

• world-faction-identity — the fiction/voice/look that makes the asymmetry read; design the two together (mechanics + meaning).
• system-counter-systems — asymmetry balances through near-hard counters that cross the mirror.
• system-unit-rosters — each faction’s roster is its vocabulary of pieces.
• system-economy — asymmetric resource flow is a primary identity axis.
• process-the-twist — each faction is an “X but Y” applied to the base kit.

See also

• design/FUN.md §9 (RTS-lite duels), §11 (fairness as a window).
• anchor-starcraft — the reference: three rosters, one balance, tuned as matchups.
• anchor-age-of-empires — asymmetry via civ bonuses on a shared unit base (a gentler dose).
• sandboxes/pathfinding-demo — mass pathing all factions share.