Wiki / Running a network
Demand
This is the part of the game everything else serves. Demand is not a number you influence from outside; it is a simulation of choices, and your network is one of the choices.
Bubbles are homes
The coloured bubbles on the map are groups of homes, not stations and not people standing on the street. Each bubble's colour is how the people living there mostly travel:
- Green — mostly walking
- Red — mostly driving, or otherwise not on transit
- Blue — mostly on transit (which, before you build, means hardly anywhere)
Click a bubble to see where its trips go. Click a destination to open that trip in the trip explorer. The explorer is the most underused screen in the game: it breaks a single trip into walking, waiting, riding and the fare, and compares it with the same trip by car.
Ten seconds of looking
Before drawing anything, click three bubbles in different parts of a city. You will usually find that demand is not evenly spread: it runs along corridors between a few dense places. Those corridors are your first line, and no amount of clever track will beat choosing them correctly.
What the model compares
Every trip is a choice between modes, and the model compares them in effective time — a single number in minutes that mixes in everything that makes a mode feel costly.
| Mode | What counts |
|---|---|
| Transit | Walking to the station, waiting, riding, walking at the far end — and the fare, counted as minutes |
| Car | The drive, parking, and the cost of owning the car, also counted as minutes |
| Walking | The walk itself; for short trips this is often the winner and no line will beat it |
Two consequences follow, and they explain most of the game:
- A cheap fare can beat a faster train. Because the fare is in the comparison, lowering it makes transit more attractive even if nothing else changes.
- Access is half the journey. If walking to the station adds 12 minutes at each end, the train has to be dramatically faster than driving to win. That is why stations in the right place matter more than fast trains in the wrong place.
Crowding changes the answer
The model does not treat a full train as a free win. Once a vehicle is 70% full, each minute on board counts for more than a minute:
| Load | What a minute on board feels like |
|---|---|
| Below 70% | 1 minute |
| At capacity (100%) | 3 minutes |
| At 120% of capacity | 5.4 minutes |
Riders who cannot board at all wait for a train with room, and that wait is added to their journey — up to an hour at a stop. So an overcrowded line does not just fail to grow; it actively pushes people back into cars, and the next day's share drops.
The line that is too popular
A line at 120% load is not a success to be left alone. It is losing riders it could have carried. The cheapest fix is usually longer trains, not more of them — but check the capacity list first, because the limit may be track rather than fleet.
Purposes
Trips are not all the same shape. The model splits them by purpose:
| Purpose | Typical pattern |
|---|---|
| Work | Peak, directional, the same people every day |
| School | Sharp peaks, very directional |
| University | Peak-ish, more scattered |
| Errands | Midday, short, local |
| Leisure | Evening and weekend |
| Medical | Spread through the day, often to specific places |
| Airports | Long, luggage-laden, willing to pay |
| Attractions | Seasonal, midday, tourist-heavy |
You can filter the map and the panels by purpose. This is the single best way to understand a line that "should" be busy: a line that only carries work trips looks full at 8am and empty at 11am, and that is not a failure — it is a schedule to fix.
Five periods, not one day
Demand is solved in five periods:
- Morning peak
- Midday
- Evening peak
- Evening
- Night
The top bar tells you which one you are in and when the next starts. The schedule panel's three demand levels (high, medium, low) map onto these periods, which is why a line can be crushed at 8am and fine at 2pm without any change to its trains.
Weekends and holidays are modelled separately: the game will tell you that today is a weekend, a holiday, or a "holiday travel day", and the pattern changes accordingly. A network tuned only on a Tuesday will feel wrong on a Sunday, and that is expected.
Intercity trips
Cities are not islands. The model also produces trips between cities — visits, shopping and business, counted in both directions. These are the long-distance market: they are why a high-speed corridor can be worth building, and why an airport link can pay.
You can inspect them per city: the city inspector shows trips within the city and trips to and from other places, with the top destinations ranked. Intercity demand is what the fares and vehicles pages are really about — on a long trip, speed and price dominate.
Reading demand correctly
Five things that surprise people:
- The bubbles are homes. Destinations appear when you open the demand view for destinations; the default view is one side of the story.
- Nothing is modelled outside the filled area. Empty map is empty demand; building there is a donation.
- Demand is per city. A station just outside a city's modelled area is nearly worthless, even if it looks close on the map.
- The mode bar is the score. Transit share is the number the game's own goals are set in.
- Crowding feeds back. The demand you serve changes tomorrow's demand. A line that is full is losing riders, quietly, every day.
Where the numbers come from
The game solves demand city by city, from mapped homes, workplaces, schools, hospitals, airports and attractions. Its panels state which sources cover which purpose, and which purposes are not modelled in a given city — the phrase "not modeled here" is a real limit, not a bug. A city with no airport data has no airport trips to capture.