Street report card

Cairo, measured street by street.

Districts built across six planning eras — from Ismail’s 1860s boulevards to the desert new towns. Not opinions about which is better planned: intersection density, dead-end share, circuity and orientation entropy, computed from open street data with the method written down.

Every district, one metric at a time

Where each district lands.

Colour by:
Orientation

Which way do the streets face?

Each rose shows how much street length faces each compass direction. The number underneath is orientation entropy, which measures one thing only: how concentrated those bearings are into a few directions. A four-way grid puts nearly all its length into two opposing pairs and scores about 1.39; streets facing every which way approach 3.58.

A high score is not the same as an unplanned district. Concentration is not regularity. A radial plan — avenues converging on circular midans, the Haussmannian idea Ismail brought to Cairo — is rigorously planned and spreads its bearings across many sectors by design, so it scores high for the same reason a tangle of lanes does. This number cannot tell the two apart; the rose above it can, because a star reads differently from a smear. Read them together.

Bearings are computed on unprojected coordinates because they are compass directions, and weighted by segment length so a long avenue counts for more than a short lane. Click any rose for that district’s full card.

The report card

Every metric, side by side.

Sorted by intersection density — the single best one-number summary of how connected a street network is. Click a row for the full district card.

DistrictIntersectionsDead endsStreet densityMean segmentStreets/junctionCircuityEntropy

Walking access

Can you walk to a hospital, school or pharmacy?

Walking time along the actual street network — not straight-line distance — at 80 metres a minute, roughly 4.8 km/h. The score is the share of walkable street that sits within fifteen minutes of a mapped facility.

Read this before you read the scores. Amenity coverage in OpenStreetMap is incomplete and uneven across Cairo. A low score can mean a district genuinely lacks facilities, or that its facilities simply have not been mapped — and this data cannot separate the two. It also says nothing about whether a facility is open, affordable, staffed, or any good. It is a map of what is mapped.
How this was measured

The method, in the open.

Five decisions shape every number on this page. They are written down here because a metric you cannot interrogate is a metric you should not trust.

  • Boundaries come from OpenStreetMap, not a gazetteer. Egyptian district names are inconsistent in OSM, so each district is tried against several candidate names and every resolved polygon is area-checked before use. Anything implausible is dropped and named rather than quietly kept. Most of these outlines are mapper-drawn place polygons rather than official administrative limits — real, but not a cadastre.
  • Downtown Cairo’s outline is ours, and it is the only one. OpenStreetMap has no polygon for Downtown at all. We tried nineteen name candidates across English, Arabic and qism spellings, and every one that returned a shape returned a landmark inside downtown — Abdeen Palace, the Qasr el-Nil theatre, Azbakeya Garden, the Central Bank building — rather than the district. Then we asked OSM directly what it holds over the area: twenty-five small neighbourhood polygons of the districts around khedivial Cairo, and nothing administrative between Cairo governorate at 3,006 km² and unnamed 0.5–5 km² fragments. So Downtown’s 7.1 km² extent is drawn by hand from its known edges: the Nile, Ramses, Port Said Street and Sayyida Zeinab. It is labelled as hand-drawn everywhere it appears, it is area-checked exactly like a geocoded boundary, and it is used only after every OSM lookup fails — so the day OSM gains a real Downtown polygon, this quietly stops being used.
  • Everything is projected before measurement. Lengths and circuity computed on raw latitude and longitude are meaningless; each network is projected to UTM first.
  • Counts are normalised per square kilometre. 6th of October is vastly larger than Downtown, so raw totals would compare size rather than design.
  • Driving and walking networks are kept separate. The report card uses the drive network, because that is what street design produces. The access score uses the walk network, because a fifteen-minute city is about walking. Mixing them would answer neither question.
  • Bearings are computed unprojected and weighted by length, because they are compass directions and because a two-kilometre avenue should count for more than a fifty-metre lane.
Street geometry is among the best-mapped data in OpenStreetMap, which is what makes this comparison reasonable. It is still community-mapped data describing what OSM contains, not an official cadastre — and district boundaries here are OSM’s, which do not always match administrative ones. Source: © OpenStreetMap contributors (ODbL), via OSMnx.