Demo Lab · Jabodetabek

The hour.

An hour on public transport buys a Menteng resident about a tenth of Greater Jakarta's job-dense floorspace. Sixty minutes from a desa in Kabupaten Bogor buys a walk. This is that hour, computed door-to-door over every footpath, busway, commuter rail, MRT and LRT in the region — and the hour itself is a choice, so the half-hour and the three-quarter-hour are published beside it.

Scheduled times · weekday 07:00–09:00 departures · p50
01

One network, four systems

A busway that carries a million rides a day, 98 Mikrotrans capillaries feeding it, five commuter-rail lines, one MRT and two LRTs — assembled here into a single routing graph with every OSM footpath in the metro.

2-D CANVAS · OSM ODbL

So what: the network looks continuous on a map. Access is not the network — it is what the network reaches inside a working hour, and that is a much smaller thing.

02

An hour from home

Every one of Jabodetabek's kelurahan and desa, shaded by the share of the region's job-dense floorspace it can reach on public transport within the cutoff. Click any unit for its own isochrone.

03

Who is left out

Rank every resident by the access their kelurahan has, poorest access first, and plot the cumulative share of reachable opportunity against the cumulative share of people. A fair network would trace the diagonal.

—Gini of population-weighted 60-minute access
—Palma ratio — the top tenth's access over the bottom four tenths'
—of Jabodetabek's people cannot reach a single hospital in an hour on scheduled public transport — most of the region travels by motorcycle, which this does not model

Every inequality number on this page is a property of the hour as much as of the network: a cumulative-opportunity measure counts what falls inside an arbitrary boundary, so the boundary has to be published with it.

Best served

Worst served (population over 5,000)

04

What each layer buys

Rerun the same hour with the rail layer removed, then with no public transport at all. The gap between the three is what each system is worth to the median resident — and what it does to the distribution.

05

How this was built, and where it is weak

Scheduled times, not congestion. Frequency-aware routing over a two-hour morning peak. A floorspace proxy standing in for jobs. Every one of those choices is stated here, with the gates that tested them.

Validation gates

Hand-encoded rail timetables

Sources & licences

Known limits

  • Jobs are a proxy. The measure is non-residential built-up floorspace (GHS-BUILT-S NRES, 2020) — where workplaces are, not how many people work in them. It is called "job-dense floorspace" throughout and never "jobs".
  • Scheduled, not congested. R5 routes the published timetable. Real TransJakarta buses sit in traffic; real KRL trains hold at Manggarai.
  • Angkot are largely missing. Outside the TransJakarta feed and one community Bogor angkot feed, the paratransit that actually moves the periphery has no open timetable anywhere. Access in the outer kabupaten is therefore a floor, not an estimate.
  • Rail timetables are hand-encoded from published headways (±15 %).
  • Walking access to transit is capped at 30 minutes, the standard assumption; ojek and feeder angkot are not modelled.
    06 · Review

    We had this instrument reviewed, adversarially.

    An independent read of the same data against the published literature. It shows that the inequality headline is largely a property of the sixty-minute cut-off rather than of the network; that the whole system is relatively regressive on this measure, with 94 % of what it adds accruing to the 30 % of the region inside DKI; that the one check this build passed compares two different estimators; and it computes the poverty axis this page had been carrying as pending.

    Read the review article →

    What this becomes for a client

    The same graph answers the questions a planning agency, an employer or a developer actually asks: what a new corridor would add, which kelurahan a fare reform would reach, where a campus should sit so its workforce can get there. Congestion-adjusted times, fare and affordability layers, park-and-ride and future-network scenarios are the obvious next build.