Cities: Skylines — the city-builder that finally beat SimCity

Colossal Order is a studio of maybe a dozen people in Tampere, Finland, and in March 2015 they released a city-builder that had no business landing as cleanly as it did. Cities: Skylines arrived two years after Maxis and EA’s SimCity reboot had promised the genre’s future and delivered a smaller one instead — tiny city plots dressed up as “regions,” a simulation engine called GlassBox that turned out to be faking more than it modelled, and a launch so dependent on EA’s servers that the single-player game was frequently unplayable because a server in another country had fallen over. Colossal Order had made small transport and city simulators before — Cities in Motion was their prior release — and Skylines reads like a studio that watched a bigger competitor stumble and simply built the game that competitor had advertised.

The comparison is worth making explicit because Skylines isn’t a reaction in tone — Jay doesn’t remember it being marketed as a rebuttal — but every design decision in it answers a specific complaint about the 2013 game. Where SimCity capped city size to force multiplayer regional trading, Skylines gave you one tile you could tile outward from, buyable in a 5x5 arrangement, entirely offline if you wanted. Where SimCity’s traffic simulation was, by Maxis’s own later admission, substantially smoke and mirrors, Skylines built genuine agent-based pathfinding: every citizen is a simulated individual with a home, a job, a car or a bus seat, and a route calculated against the actual road network you’ve built. Get an intersection wrong and you will watch it fail, exactly and repeatedly, until you fix it.

Why the traffic AI is the whole game

The single mechanical decision that made Skylines the genre’s new baseline is that traffic works as the feedback signal for almost every other system, well beyond a cosmetic layer on top of a zoning game. Zone too much industry too far from housing and you get a commute problem before you get a pollution problem. Build a single-lane road as your only artery between two districts and you’ll watch a jam form at the exact junction where two lanes narrow to one, because the pathfinding is genuinely finding the shortest route and genuinely getting stuck doing it. Cims — Skylines’s citizens — don’t obey a scripted flow field the way many contemporaries fake it; they’re solving a routing problem against a network you keep changing, which is why the same road layout can work fine at ten thousand population and collapse at forty thousand once feeder traffic compounds.

That legibility is what separates a good city-builder from a pretty one. SimCity 2000 and SimCity 4 modelled traffic more abstractly, as a load value flowing along a graph rather than as individuals actually driving; it’s a valid design choice for a game with slower base hardware to answer to, but it means the player never quite trusts the number on screen. Skylines makes the failure visible: you can watch the exact car stuck at the exact junction and understand why. That single decision — simulate the individual unit, don’t fake the aggregate — is the same instinct that makes a strategy game’s economy trustworthy rather than merely plausible, and it’s the reason modders were able to build on top of it rather than around it. The community-made Traffic Manager: President Edition, which lets players override junction priority and lane assignment by hand, only works because the underlying pathfinding is real enough to be worth correcting.

Modding as the second act

Skylines shipped with a full asset editor and Steam Workshop support from day one, and that decision did more for the game’s longevity than any single expansion. Within its first year the Workshop held tens of thousands of custom buildings, props, and maps; within a couple of years it held total conversions of the UI, alternate traffic AIs, and entire policy systems the base game never shipped with. Paradox — who published rather than developed the game, which matters, because Paradox’s post-launch model for its grand-strategy line (build a base game cheap, sell years of expansions) transferred cleanly to a city-builder — kept the official expansion cadence going: After Dark added day-night cycles and tourism, Snowfall added winter cities and trams, Mass Transit deepened public transport, Green Cities added pollution-conscious late-game options, Parklife turned parks into their own zoned system, and several more followed. Few of those expansions were essential on their own, but they extended a base simulation solid enough to keep absorbing new systems without buckling.

The lineage worth tracing here runs further back than SimCity, though SimCity is the game Skylines was most directly answering in 2015. The idea of the simulated management game — a system you tend rather than a level you clear — is older, and Jay would put a chunkier ancestor in the same family: Theme Hospital, Bullfrog’s 1997 management sim, ran on the same basic contract of watching a system you’d built produce emergent, sometimes comic failure states, dressed in satire rather than urban planning. Further back still sits the strange, half-remembered lineage of the simulation genre’s earliest attempt at modelling a life rather than a level: Little Computer People on the Commodore 64 gave you a single simulated inhabitant to watch rather than a city to build, but the underlying promise — a system that runs whether or not you’re actively steering it — is the same promise Skylines makes at city scale, just multiplied by tens of thousands of individually pathfinding citizens instead of one.

The economy underneath the traffic

Skylines’s budget layer is easy to overlook next to the traffic simulation, but it’s doing real work of its own. Each service — road maintenance, healthcare, fire coverage, education — carries its own budget slider, and underfunding one doesn’t just reduce a background stat; it produces visible, specific failure states, the same way an underbuilt road network does. Cut fire budget too far and a burning building’s local station genuinely can’t muster the trucks to reach it in time, and the fire spreads to adjacent buildings exactly as slowly or quickly as the truck count and travel distance dictate. Education funding determines the literal skill level of the workforce your industrial and office zones can draw from, which means a city that never builds schools will cap its own commercial growth regardless of how much land it zones, because the simulated workforce simply isn’t qualified for the jobs on offer. None of that is a deep economic model by the standards of a dedicated management sim, but it’s consistent with the game’s central discipline: model the individual unit — the truck, the student, the worker — rather than an abstracted city-wide average, and let the visible failures do the teaching.

Immigration works the same way. Cims arrive at the city edge as individuals looking for a home and a job that matches their education level, not as an abstract population counter ticking upward, which means a city that’s zoned nothing but low-density housing next to zero industry will simply stop attracting new residents once the existing job pool fills, no matter how much appealing housing sits empty. That’s the same underlying design conviction across every one of the game’s systems, budget included: the aggregate number on a graph is a symptom, not a lever, and the actual lever is always some individually simulated unit’s decision about where to live, work, or drive.

The parts that don’t hold up

Skylines is not a flawless simulation, and the seams show at scale. The road-building tools in the base game are fussier than they need to be — curves snap awkwardly, and building an interchange by hand before mods streamlined the process was a genuine chore. The economic model is shallower than the traffic model: budget sliders and a handful of policies stand in for anything like the layered tax-and-service trade-offs a genuinely deep economic sim would model, and once a city crosses roughly one hundred thousand population the simulation’s frame rate becomes the real limiting factor on ambition rather than any design ceiling. And a city-builder that leans this hard on player-made content has an uncomfortable dependency: a great deal of what makes late-game Skylines sing — the traffic overrides, the growable buildings, the UI improvements — was never built by Colossal Order at all.

None of that changes the verdict. Skylines didn’t just outsell and outlast its rival; it reset what “the city-builder genre” meant for the better part of a decade, because it was willing to simulate the individual commuter rather than fake the citywide average, and because it opened the game up to a community willing to fix what it hadn’t finished. That combination — genuine simulation plus genuine extensibility — is a harder thing to build than a flashier launch trailer, and it’s why Skylines was still the default answer to “what city-builder should I play” for years after its release, long enough that whatever comes next from Colossal Order has an unusually tall shadow to step out of.

Spoilers below

There’s no story to spoil in a city-builder, but the late-game systems are worth flagging for anyone who wants to discover the traffic-collapse problem themselves rather than be warned about it. Past roughly sixty thousand population, the base game’s road hierarchy tools stop being merely fussy and start actively fighting you: highway interchanges you built early for a town a quarter of that size will bottleneck catastrophically as feeder traffic compounds, and the honest fix is usually to bulldoze and rebuild rather than patch. The Cims’ pathfinding also has a specific, well-documented failure mode around “dumb parking AI” in the pre-Parklife version of the game, where citizens will park a full simulated street away from their actual destination and then walk the distance, which reads at first as a bug and is in fact the router correctly solving for the parking-space shortage you created by underzoning parking. Discovering that a jam three districts away is being caused by one broken junction back near the start of your map is the game’s genuine “aha,” and it’s better hit blind the first time.

If you’ve exhausted Skylines and want the same “systems you tend rather than levels you clear” feeling with more overt satire and no traffic to fight, Theme Hospital is the shorter, funnier detour; if you want the same genre pushed toward a single simulated life rather than a whole city, Little Computer People is the strange eight-bit root of the whole idea.