Turn-Based Tactics
What it is. Grid combat where the enemy shows you their whole plan, then you act. No hidden rolls, no fog — the fun is rewriting a telegraphed future with positioning: push an enemy so its attack hits another enemy, block a spawn, bump a unit off a ledge. Each fight is a solvable puzzle, not a dice game.
Player fantasy / why it’s fun. “I saw the disaster coming and turned it against them.” Perfect information makes you feel like a chess master: the win is found, not rolled, and a clean multi-enemy redirect is pure satisfaction.
Pillars
- Rewrite the telegraph. Store telegraphs as directions on units, not target tiles — so pushing a unit re-resolves what it hits. The telegraph is a live plan you edit, which is the entire game.
- Perfect information, perfect honesty. What’s shown is exactly what resolves. No hidden variance. The genre’s trust is that the display is the truth (law 6).
- Both proofs, every scenario. A scenario needs a winning line AND a losing do-nothing. If passivity survives, there’s no threat; if no line exists, it’s unfair. The first cut’s bug is usually “can’t even reach the objective.”
The loop stack
| Scale | The beat |
|---|---|
| Moment | Read a telegraph → move/push a unit → watch the future re-resolve in your favor. |
| Encounter | A scenario: protect an objective from telegraphed threats over N turns using positioning. |
| Session | A mission chain / island: escalating scenarios, limited resources carried between them. |
| Meta | Unlock squads, mechs, and mechanics; learn the redirect vocabulary. |
Essential systems
| System | Why it’s load-bearing |
|---|---|
system-telegraphs | The whole loop. Stored as directions, resolved after your move — the editable future. |
system-enemy-archetypes | The threat roles whose telegraphs you redirect — melee, artillery, spawner, boss. |
system-encounter-design | Composing archetypes + objectives into a scenario with a findable line and a losing null. |
system-combat-model | Push, bump, redirect resolution — the deterministic rules a scenario is built on. |
system-difficulty-and-dda | Threat density and turn count as the ramp; harder = tighter lines, not hidden rolls. |
Content & difficulty model
- Content is scenarios, not stats. A scenario is a board + objective + a telegraphed threat sequence with a provable solution. Hand-authored or generated, each must pass the two proofs.
- Difficulty = tighter solution space. More threats, fewer turns, harder objectives — the line still exists but demands more precision. Never add hidden variance to raise difficulty.
- Every mechanic proven in isolation. Push, bump, redirect, spawn-block each get a unit test before they combine (law 3 / derive-don’t-vibe).
- Bots are real defenders. A 1-ply greedy planner over
structuredClonescoring is a genuine baseline — it should perfect-clear a solvable scenario and the do-nothing bot should lose everything.
Signature-mechanic seeds
- Tactics but you command the enemies too (perspective) — set both sides’ telegraphs, then a rival edits yours; a duel of redirects.
- Tactics but the grid is a deck you draft (structure) — Into-the-Breach
meets Slay-the-Spire; your moves are cards. Bridges
genre-deckbuilder. - Tactics but one push ripples through a chain (mechanic-swap) — dominoes: every shove can cascade, turning the board into a physics puzzle.
- Tactics but you protect a city that remembers (theme + structure) — a town rebuilt between missions; buildings saved persist as bonuses.
- Tactics but the future is shown only for one turn (constraint) — a memory variant; you plan several moves against a telegraph you can no longer see.
Common pitfalls
- Telegraphs stored as tiles. If a tell targets a tile, pushing the attacker doesn’t rewrite it — and the core verb (redirect) silently breaks. Store directions on units.
- Hidden variance. Any concealed roll violates the perfect-information trust; a “miss” the player couldn’t foresee feels like a betrayal.
- Passive survives. A scenario a do-nothing bot lives through has no threat. Always assert the losing null.
- No line exists. An unsolvable scenario is unfair. Prove the winning line before shipping (greedy bot perfect-clears).
- Objective unreachable. The classic first-cut bug — the plan can’t even touch the goal. Test reachability.
Anchors
anchor-into-the-breach— the definitive reference: perfect information, telegraphed grid, redirect-the-future. Steal the whole model.genre-deckbuilder— a natural blend (moves-as-cards); shares the telegraph-honesty and skill-delta discipline.
Verify
Prove it in FUN.md §12 · Turn-based tactics: the greedy bot perfect-clears, the do-nothing bot loses everything, each mechanic (push/bump/redirect) is proven in isolation, and a golden end-state replays.
Composes with
system-telegraphs— the editable future that IS the loop.system-enemy-archetypes— the threat roles you redirect.system-encounter-design— scenarios with a findable line and a losing null.
See also
examples/sokoban— the purePuzzle<State,Move>logic/view split and clone-and-score bot; tactics is that pattern with telegraphs (laws 6, 7).- design/FUN.md Part 1 — laws 3, 6, 7 underwrite this genre’s determinism and honesty.