Interurban / 79 metros / Philadelphia

Pennsylvania · NortheastCensus 2020 tractsFree rotationSteam · Mac · Windows

Philadelphia

A pre-war city with a subway, an elevated, trolleys and regional rail already in place. The bones are extraordinary and the money never is.

The ground no playable rail · water and major roads only
Philadelphia: the metro's water and major roads; the pack starts with no transit25 km

No playable rail in the pack The pack has no playable rail for Philadelphia: you start from bus, on these roads.

Water: Census TIGERweb hydrography. Roads: FHWA HPMS, the busiest classes. 3,683,718 residents in 926 census tracts, Census 2020.

#Facts, figures, history

Philadelphia, as the pack records it.

Population by decennial census, as recorded in the pack.
YearPopulation
19202,278,649
19302,593,756
19402,639,052
19502,983,527
19603,380,915
19703,582,499
19803,366,448
19903,352,513
20003,416,146
20103,510,118
20203,683,718

The pack's census series in Philadelphia, Delaware, Montgomery and Bucks runs from 2,278,649 residents in 1920 to 3,683,718 in 2020.

The census tracts cover 3,640 km², about 1,012 residents per km², with 1,756,678 jobs at the workplace.

The pack has no inherited transit; the map is water, roads and the ground.

Difficulty is Forgiving.

✓What winning here takes

Six ways to win Philadelphia, each set by Philadelphia.

There is no score to beat. Each condition is computed from the metro as you inherit it, with the engine's own arithmetic; these are the bars the game will set on day one in Philadelphia, in 2025. The blue tick is where the city stands today; the red band is where it has to get to. The run continues after you win.

  1. Transit Capitalheld for a year

    3.5%today · the real system carries
    3.5%needed · carry yourself, for a year

    The region's existing system carries about 3.5% of all trips. Those lines are not in the pack as playable rail, so the bar is to carry that much on a network of your own.

  2. Roads Unclogged

    0%today · relief before you build
    10%needed · cut in peak congestion

    Against what the region would have without your network, not against the day you started — the region grows, and holding congestion flat through growth is the harder achievement.

  3. Carbon Cut

    0ttoday · net tonnes avoided today
    73,674tneeded · net tonnes of CO₂ a year

    Scaled to this metro's 3,683,718 residents at 0.02 tonnes each — roughly what moving a few percent of the region's driving onto rail saves, after the trains are charged for their own power.

  4. Held the Lineheld for a year

    0%today · on time, before a train runs
    80%needed · on time · 65% satisfied · a year in surplus

    All three at once. The cheapest route to a surplus is cutting the service that produces the other two, so this is the condition that cannot be bought.

  5. Within Reach

    0%today · within a walk of your stops
    10%needed · of the metro within a kilometre of a stop

    There is no playable rail to inherit, so coverage starts at nothing and the bar is the absolute one: a tenth of the metro within a walk. Two long rail lines through the core, roughly.

  6. Ten Minutes Back

    0%today · cut so far
    10%needed · off the average regional commute

    Measured on every resident, including the people who never board a train — a network that takes cars off the road shortens the drive for everyone left on it. The day-one average is the model’s first solve.

Carbon Cut's bar is 0.02 net tonnes a year for each resident, floored at 20,000. Transit Capital, coverage and the commute are taken from the network you inherit: where the pack seeds operating rail, from that system's first solve; where it does not, Transit Capital uses the published NTD share (floored at 1% and capped at 6%) and coverage starts from nothing. The gauges' Transit Capital tick is the published share either way.

What you are given

What you are given

Every one of Philadelphia's 926 census tracts arrives with its real numbers attached: resident population, population 18 and over, occupied households, jobs by sector at the workplace, resident workers, and median household income — each one hoverable, each one carrying its GEOID so you can go and check it against the Census yourself.

Work trips are not guessed. They are seeded from the observed LEHD LODES home-to-work flows between those tracts, so the real spatial structure of the region — a century of geography, zoning and history that no gravity model reproduces — is preserved, and the model predicts only the change your network causes.

Philadelphia's pack has no present-day playable rail, so a 2025 start is bus on these roads. A 1920 start inherits the dated streetcar, elevated and interurban map from OpenHistoricalMap; Philadelphia Parks & Recreation.

What makes it hard

What makes Philadelphia hard

Forgiving. Dense and centred — forgiving as a city even when the pack has no playable rail to inherit. There is no network to inherit and no habit of using one, which the mode choice model expresses as a constant working against you before you have drawn anything.

And the decade changes the problem, not the vehicle list. Philadelphia in 1960 is a different city from Philadelphia in 2025: smaller, denser, fewer cars per household, no federal transit money at all, and an expectation that you cover your own operating cost. The population distribution is back-cast by pivoting the real 2020 distribution rather than invented, so the rivers and the rail yards and the university stay where they are. The inherited map for that decade is the archival alignment that was open then — OpenHistoricalMap; Philadelphia Parks & Recreation — not a modern line that had not been built.

Philadelphia is in the free rotation. One of thirteen metros that take turns as the week's free assignment — the whole demand model, the operating budget, the events, the strike, the land use and the full progression ladder, in 2025, for five years. Not a cut-down demo: the same simulation.

Play Philadelphia

Sources

How the model turns these into riders, with every mechanic named →