What it is. You author a system — a factory, a park, a colony, a household — then tune its inputs, throughput, and demand until it runs without you. The verbs are place, connect, price, staff, cut. Play is diagnosis and adjustment, not reflexes.
Player fantasy / why it’s fun. The dopamine is legibility earned: a mess of red numbers you slowly turn green, then watch tick over on its own. You are the invisible hand that finally understands its own machine.
Pillars (exactly 3)
- Legible systems. Every number traces to a cause the player can see and touch. If a queue backs up, the bottleneck is findable. Opacity is the enemy — see
pattern-readability. - Meaningful optimization. There is always a better configuration, and the delta is felt. Choices trade off, not dominate — dodge
antipattern-boring-optimalwhere one build wins forever. - Escalating load. Demand, complexity, or scale ratchets up so a solved layout becomes tomorrow’s bottleneck. Pressure, not a plateau.
The loop stack
| Layer | Beat |
|---|---|
| Moment | Read a gauge, place/reprice one node, see the number move. |
| Encounter | Diagnose one bottleneck or shortfall and rebalance around it. |
| Session | Absorb a demand spike or unlock; re-plan the whole flow to hold margin. |
| Meta | Master the system’s shape — the intuitions that make the next map trivial. |
The loop is observe → hypothesize → adjust → watch it settle. Keep the settle-time short enough to close the feedback fast; see pattern-feedback-loops and pattern-mastery-and-flow.
Essential systems
system-economy— the spine. Money or a proxy currency that flows in from output and out through upkeep; the ledger must balance to survive.system-resource-loops— raw → refined → sold, with buffers and spoilage. This is where bottlenecks live.pattern-feedback-loops— the demand curve. Success raises expectation, cost, or scrutiny so the target keeps moving.system-progression— an unlock spine (new buildings, tiers, districts) that gates complexity sosystem-onboardingcan land one concept at a time.system-difficulty-and-dda— scenario pressure: quotas, deadlines, disasters that force replanning.system-inventory-and-ui— the interface is the game. Graphs, heatmaps, and overlays are your primary telegraph; a hidden number is a dead mechanic.
Content & difficulty model
- Scenario, not level. Ship hand-authored maps with a starting condition, a win/loss ledger, and a twist constraint (scarce water, hostile winter, a fixed footprint). Reference
anchor-frostpunk‘s scenario framing. - Difficulty is demand, not damage. Raise quotas, shrink margins, add a competing sink. The curve is economic tension — pace it per
pattern-pacing-and-tension. - Sandbox unlock late. A free-build mode is the reward for beating the authored pressure, not the default first experience.
- Determinism buys trust. A fixed-seed economy means a spike is diagnosable, not luck. Route every random event through a deterministic RNG; see
system-procgen-designfor how to seed varied-but-fair scenarios.
Signature-mechanic seeds
- Tycoon but you can never expand, only rebalance (constraint). The footprint is fixed at frame one — no new land, no new buildings. Every gain comes from reconfiguring what exists: repurpose, re-route, re-price. Optimization goes deep instead of wide, and the puzzle is legibly bounded.
- Management but the thing you run is a single family (perspective). Zoom the tycoon lens onto one household — budget, chores, moods, aging. Currencies are money and trust and time; a bad quarter isn’t bankruptcy, it’s a strained relationship. Human weight replaces the spreadsheet.
- The demand curve talks back (agency). Your customers/citizens are named agents with memory — see
anchor-rimworld. Cut a corner and a specific person suffers, and remembers; optimization acquires a conscience. - Optimize toward shutdown (goal-inversion). Victory is winding the machine down cleanly — a factory you must decommission, a fund you must spend to zero on the right things by a deadline. The tycoon loop, run in reverse.
- You inherit a broken machine (starting-state). Open on someone else’s mess — miswired, over-leveraged, badly zoned. The first session is pure archaeology: understand it before you touch it.
Common pitfalls
- Spreadsheet without stakes. Numbers move but nothing is at risk that the player cares about — the core trap. Attach human or moral weight: a person, a place, a promise. Without it you get
antipattern-fake-choicedressed as depth. antipattern-boring-optimal— one layout dominates and every map collapses to it. Make trade-offs genuine and contexts differ.antipattern-currency-spaghetti— six currencies that all convert to money. Keep sinks and sources few and meaningful.antipattern-false-depth— knobs that look like choices but never change outcome. Cut them.antipattern-endless-tutorial— over-gated onboarding that explains the machine you’d rather discover. Teach one system, then step back.antipattern-solved-metagame— once players find the build order, escalation must outrun it or the game is over.
Anchors
anchor-frostpunk— the moral-weight exemplar: every efficiency choice is also an ethical one; the primary reference for this genre.anchor-rimworld— named-agent management where the story emerges from the system; see alsopattern-emergence.anchor-factorio— the throughput-optimization purest form; bottleneck-hunting as the whole game.anchor-mini-metro— legibility and constraint distilled: minimal inputs, a rising demand curve, no fat.anchor-stardew-valley— the cozy end, where management wears the skin of agenre-farming-sim.
Neighbors worth studying: genre-city-builder, genre-incremental (the demand curve without the map), genre-4x (management at civilizational scale).
Verify
Prove the loop against design/FUN.md#17-·-city-builder — is the bottleneck findable, does the demand curve keep the target moving, does a tuned system visibly run itself? Judge looks via process-refine-and-handoff.