04 · In the build
A cozy five-minute browser toy that, over twelve days, became a phase-gated commercial city-builder: a solarpunk hex world with a specialization economy, an endgame worth optimizing forever, and a headless simulator that balances it.
The living city builder you play for years, with your friends, for the price of a coffee a month.
The seed was deliberately small: a snackable toy with one number to watch and no way to lose. The vision written the same night was not: a browser-first, always-online builder where your city is a node in a trading network, free to start and $5/month for the economy that makes it social.
Every decision since has been governed by one line, held above even the design pillars: no online element → not worth building at all. Solo depth is scaffolding toward the shared world, not the destination.
The first prototype cut everything a city-builder usually hoards (utilities, traffic, budgets, fail states) down to zone, grow, watch. It shipped in an evening and grew its first city from 0 to 1,291 people. Then the direction inverted: this would be the studio's first commercial product, resourced and de-risked like a real business, and the cut systems came back one at a time, but only when each earned its place against a single test: does this add strategic depth, or just clicks?
Utilities returned as coverage networks. Money returned as stakes. Traffic returned as a pressure that transit relieves. Each was a deliberate re-addition, not a default.
The name was chosen by a five-judge panel (The Player, The Storefront Skimmer, The Cryptograph Keeper, The Multiplayer PM, and The Trademark Adversary) across three scored rounds.
The final round scored eighteen words as Metropolis Dawn ___. GRID, GLOBE, and GARDEN tied at 38 of 50, each carried by a different judge; the next word trailed by six. GATHERING and GALAXY were struck on trademark collisions, one of them with a game the studio had literally shipped.
The Operator picked GRID for its tech-showcase read, and kept the losers anyway: the globe is the world-select, the grid is the build surface, the garden is the city greening as it matures.
Not a straight line: a series of pivots, each one visible in the save-format number. Square grid to hexagons. Single materials to a six-resource basket. Money as the endgame to throughput as the endgame.
The 100-hour engine is one idea: a region can't make everything well, so it trades. Terrain hands each region a different slice of six resources; a two-step chain turns raw into finished; and what a region lacks, it must import, or export its own surplus to afford.
Trade is a real exchange, not free inflow: you barter your Goods reserve for a neighbour's raw surplus, and a region with nothing to sell can import nothing. That single rule is the online market, rehearsed inside one city.
The hard part of a systemic economy is knowing whether it actually works. So the last stretch built a headless tool that runs the real game logic (no graphics, no reimplementation to drift out of sync) as AI cities across tiers, regions and tuning knobs, and prints stats.
It caught three things a human never would have at a glance: AI cities whose Works sat on the wrong terrain, producing zero finished goods; cities pinned at 100% gridlock because their roads never reached the freeway; and trade-dependent regions quietly going insolvent. Each was then fixed and re-measured: a producing economy, flowing traffic, and towns sized to what their land can sell.
| region | tier | pop | net/mo | tp | cong |
|---|---|---|---|---|---|
| A · ore | City | 22,470 | +170 | 57.6 | 25% |
| B · stone | City | 21,971 | +154 | 76.8 | 34% |
| C · oil | District | 16,572 | +278 | 0.0 | 0% |
| D · grain | District | 16,140 | +219 | 0.0 | 0% |
| E · lumber | City | 11,148 | +65 | 0.0 | 16% |
goods regions (A,B) manufacture & export · resource regions (C,D,E) extract & sell raw · every region self-sustaining
Easy to learn, deep to master. Every hour-100 system passes one test: does this add strategic depth, or just clicks? If it's clicks, cut it.
The 100-hour engine is specialisation and trade, not fifty micromanaged meters. Depth and the social hook are the same system.
Pull, never punishment: your friends' cities thrive when you show up.
No timers, no energy meters, no dark patterns.
The subscription buys access, content and convenience, never resources or trade advantage. The integrity of the market is the whole product.
The online mandate: if there's no online element, it's not worth building at all. Solo depth exists to prove the fun is real before a dollar is spent on infrastructure.
Each phase must be validated as fun before the next is funded, especially before any backend. As of this session, the deep solo game is done to MVP: the gate to Phase 2.
The business is a $5/month no-brainer by construction: against a $40 box that goes stale and a $0 game full of timers, an always-growing, never-pay-to-win social builder is an easy yes for a genre fan. The subscription only survives if players stay happy for months, so the whole model points at long-term delight.
The solo game holds up end to end, and every AI city already wants to trade.
Cities are solvent in every region, producing where the land allows and trading where it doesn't, with traffic that flows. That's not an accident of tuning. It's what the simulator measured into place. The surpluses and shortfalls the online market needs already exist inside single cities.