A feature list with no citations is a wishlist. Every entry below names the module it lives in, because the whole argument of this project is that you do not have to take its word for anything — the engine is open source, and if a line here and the code disagree, the code is right and this page is a bug.
Trip generation by purpose. Pivot-point gravity distribution seeded on observed commute flows rather than on a fitted curve. Nested-logit mode choice across auto, transit, walk and bike, with car ownership in the nest. Frequency-based hyperpath assignment that reasons about headways and mean waits. Solved to equilibrium over crowding and road congestion by successive averages.
demand.ts · distribution.ts · modechoice.ts · skim.ts · assignment.ts · sim.ts
Base fare and distance charge. Free-transfer windows modelled the way ticket validity actually works. Time-of-day pricing per period, which genuinely flattens the peak rather than only suppressing it. Concessions that reach behaviour and revenue through separate factors. Monthly passes priced as a ceiling, where the holder's next trip is free to them and still earns the agency money. Every one of them is a fare rise the board has to vote on.
economy.ts · skim.ts · riders.ts — measured elasticity −0.52, held in a band by a test
Service level, wage ratio and maintenance funding, the last split across guideway, rolling stock, facilities and systems. Staffing targets, missed runs, grievance that accumulates, and a strike hazard convex in the shortfall. Per-period headways across six periods including evening, night and weekend. Consist lengths. Express and skip-stop patterns.
policy.ts · economy.ts — labour content from CRS R47900 and APTA
Five alignments from tunnel to mixed traffic, each with its own capital cost, speed ceiling and reliability. Cost and schedule risk drawn once at commit from the run's own seeded stream, so reloading cannot escape a bad draw. Named setbacks that fire at most once per project. Track that splits automatically at any station it crosses, because track may not run past a stop without serving it.
construction.ts · network.ts · land.ts
Second entrances that widen the walk catchment by a platform length — about eighteen per cent more reachable ground, and the reason agencies spend nine figures on a doorway. Retail concessions for non-fare revenue. Longer platforms, step-free access, park-and-ride. Each one changes a number the demand model reads.
upgrades.ts — accessRadiusKm()
Vehicles ordered with real delivery lead times, ageing on condition and mean distance between failures, overhauls that reset part of that, and a spare ratio that decides how much of the fleet is actually available on any given morning. Reliability feeds unpunctuality, which feeds perceived wait, which feeds ridership.
fleet.ts · transitgraph.ts
Operating cost split the four ways the National Transit Database splits it. Farebox recovery. Federal, state and local subsidy with era-correct shares, including the withdrawal of federal operating assistance from large urbanised areas in 1998. Debt with a credit rating and a spread. A state-of-good-repair backlog that grows when you defer maintenance and bites when you do not.
economy.ts — NTD four-function split
Public approval and council support as separate things that move differently. A board that votes on fare rises and borrowing, and refuses both when it will not carry them. Election cycles. Institutional trust that changes your borrowing spread, your funding, and how much scrutiny an overrun attracts. Local opposition that is specific to a place rather than a global difficulty slider.
politics.ts · trust.ts
17 fiscal, 25 political, 19 operational, 9 social, 27 exogenous and 15 opportunity. Some are decisions with real branches; some simply happen to you. Every one that is in force is shown, named, with what it is currently doing to the model — because an invisible demand multiplier is indistinguishable from a broken simulation.
events.ts · effects.ts · ripple.ts
Population and jobs move over a run in response to accessibility. Districts derive their own character annually from their own live census and employment numbers, so a residential neighbourhood can become a secondary centre while you are playing. The counterfactual is tracked too, so the game can say how much of the growth was yours.
landuse.ts · districts.ts
When ridership moves, the attribution panel says what moved it — service, fares, shocks, the city itself — and prints the part it could not explain rather than folding it into whichever term was largest. There is also a per-station answer to “why does nobody board here?”, decomposed into the terms of the utility function.
explain.ts — attribution, not decomposition
A career ladder from Draughtsman upward, a blended score with the arithmetic shown, six victory conditions each computed from the city you inherited, and rider sentiment broken out by segment so you can see which group you have been failing. Runs are recorded across cities.
ranks.ts · score.ts · victory.ts · career.ts · riders.ts
A six-step walkthrough and Field Notes, each note verified against the code that produces it. Undo and redo. Standing policies that answer routine decisions the way you would. Keyboard shortcuts and an escape ladder that closes exactly one thing at a time. Saves, and a seed you can share so somebody else can play your city.
Coach.tsx · dismiss.ts · autopilot.ts · save.ts
54 metros from Census 2020 PL94-171, ACS 2018–2022, LEHD LODES8 2022, FHWA HPMS and TIGER/Line — 127.8 million residents across 31,521 census tracts. Real transit from each agency's own published GTFS feed where one exists: 480 real routes and 19,033 real stops. One script, no API key, reproducible.
tools/ — every source a US federal dataset or an agency feed
Everything on the page above is in the free game. That is not a marketing position, it is how the code is built: the model does not know which tier it is running under, and a free run is not a degraded run. What is bounded is how much metro, how much history, how many kinds of vehicle, and how long.
| Base — free | Full — $20 | |
|---|---|---|
| Metros you can play | 13 | All 54 |
| Decades | 2025 only | 1920 – 2040 (8) |
| Modes | Rail only | Every mode |
| Years in office | 10 | Unlimited |
| The full four-step demand model | Yes | Yes |
| Real census & observed commute data | Yes | Yes |
| Fares: time-of-day, concessions, passes | Yes | Yes |
| Operating budget and three dials | Yes | Yes |
| Labour model and strikes | Yes | Yes |
| Fleet ageing, station upgrades, express | Yes | Yes |
| 112 events, land use, career ladder | Yes | Yes |
| Bus, bus rapid transit, ferry | No | Yes |
| Plays offline | No | Yes |
| Account and connection required | Yes | No |
| Every future expansion included | No | Yes |
| Price | Free | $20 once |
The complete simulation on thirteen real metros. Check the model on a city you know before taking anyone's word about the other forty-one.
Either one also opens all 54 metros and removes the ten-year clock. Surface adds bus, bus rapid transit and ferry; Eras adds seven more decades.
All 54 metros, eight decades, every mode, no clock, every future expansion at no further cost, and the one thing nothing else buys: it runs with no account and no connection.
A tenth of everything this earns goes to pro-urban coalitions across the United States — transit advocacy groups, street-safety campaigns and the land-use reform organisations doing the unglamorous work of showing up to public meetings. Why, and who.
