The living city · version 1.0
A living city from open maps: zones, streets, residents and their day
How Parallel Earth turns OpenStreetMap into Delhi's zones and streets, generates the people who live there, and routes each of them through their day along real roads.
16 October 2026 · 8 min read
Parallel Earth shows real places and generated people. The streets, parks, schools, places of worship, markets, hospitals and metro stations are real, from OpenStreetMap. The people are not: they are generated, labelled as such, and they move through their day on those real streets.
Zones
Delhi is divided the way it is governed: the Municipal Corporation of Delhi's twelve zones, the New Delhi Municipal Council and the Delhi Cantonment. OpenStreetMap does not hold these boundaries, so each zone is approximated as the catchment of its office: every mapped neighbourhood goes to the nearest zone office, with a few clear cases pinned by name. Pune works the same way with its fifteen ward offices and two cantonment boards. The maps say they are approximate until official boundaries replace them.
Streets and places
For each zone we take its roads by class (from expressways down to residential lanes), its parks, water and land use, and its places: schools, colleges, places of worship with their religion, markets and malls, hospitals and clinics, police and fire stations, metro stations. Everything is clipped to the zone's cell and simplified, about 250 KB to 1 MB per zone. Rohini, for example, has thirteen metro stations, over a hundred schools and some eighty parks and grounds.
Residents
Each resident belongs to a household: joint, nuclear, a couple, someone living alone, or a single parent, with a head and a sense of who decides money, rituals and schooling. A household's tradition is drawn from Census 2011 shares for its state; its people have professions, ages and languages. Tradition decides which festivals they keep and how. It never sets anyone's opinions or personality. The household and work mixes are placeholders until they are fitted to NFHS-5 and PLFS tables, and every person says so.
A day as a set of trips
Each person has a routine for the kind of day it is: a workday, a weekend, a holiday. Every block of that routine is taken to a real place: children to the nearest school, a teacher to a school, a nurse to a hospital, a shopkeeper to the market, an office worker to the metro, worship to the nearest place of their own tradition. On a festival evening, people who keep that festival go to the biggest ground nearby, which is where the crowds form on the map. Life changes the plan too: a parent with a sick child stays home.
Routes along real roads
Trips follow the shortest path on the road network, with a slight preference for bigger roads, as people have. The server builds each zone's network once and keeps the shortest-path trees it grows toward each destination, so a whole zone's day is planned in well under a second once warm.
cost = length × f(class) f = 0.8 for main roads, 0.9 for tertiary roads, 1.0 for lanes
- The path minimises total cost from the starting point to the destination
| What the map shows | Where it comes from |
|---|---|
| Streets, parks, water, places | OpenStreetMap, real |
| Zones | Office catchments from OpenStreetMap points, approximate |
| Residents and households | Generated, with labelled priors |
| Their trips | Their routine, the calendar and their life, routed on real roads |
| Local issues | Simulated from season, geography and festivals |
What we checked
- ✓ Every Delhi zone has a street map, and every place on it lies inside its zone
- ✓ On 16 October 2026, a Navratri and Ramleela evening, most of Rohini's simulated residents head to Swarn Jayanti Park after 7 pm
- ✓ Toddlers stay home; children go to schools near their homes; nobody goes to a place of a tradition they do not keep
Limits, for now
- Zone boundaries are approximate until official ones are published
- The simulation follows a sample of households per zone, not the whole population, and reports shares, not head counts
- Building footprints and traffic are not modelled yet
Written with AI assistance from the colony's working notes and code, and reviewed by a person. Simulated results are labelled as simulated.