12,585
Points measured
v0 · first drive results. On 1 Nov 2026 we drove the model's detours: 5 of 5 comparable directions matched. Area figures below are still model values. See every pass
3 places driven · 1 dayDivided roads move traffic along a corridor and cut the people beside it off from each other. Nobody measures that cost. This does — from open data, with the method published so any claim can be checked, including the parts still wrong.
12,585
Points measured
2.2–3.8×
Beyond the ring road vs the core city
94
Areas ranked
5 / 5
Driven directions that matched the model (1 Nov 2026)
The detour index is how far you must drive to reach a point directly across the road. An index of 40 means forty times the straight-line distance. Measured in concentric rings from the centre, the cost does not simply rise with distance. It climbs steeply to a peak between 26 and 40 kilometres out, then falls away.
| Ring | Points | Median index | Above 50 |
|---|---|---|---|
| Core city, 0–14 km | 5,521 | 18.05 | 20.5% |
| Ring road band, 14–26 km | 3,955 | 33.76 | 39.9% |
| Peri-urban fringe, 26–40 km | 1,045 | 57.52 | 53.8% |
| Rural, 40–60 km | 1,589 | 48.10 | 49.0% |
| Beyond 60 km | 475 | 47.26 | 49.1% |
A reading, not a conclusion. The fringe is where highway-grade divided roads already exist but the local street network behind them has not filled in, so the severance has nothing to absorb it. Further out there are fewer divided roads and less severance. The shape is consistent with that explanation and has not been tested against it. Patancheru and Isnapur, where this project began, sit at roughly 36 km — inside the peak ring.
The detour index is how far you must drive to reach a point directly across the road. An index of 40 means forty times the straight-line distance. Crossing a road beyond the Outer Ring Road costs between two and four times what it costs in the core city, depending on how severely the data is filtered. It narrows under correction. It never disappears.
| Treatment | Core | Beyond ORR | Ratio |
|---|---|---|---|
| No correction | 18.02 | 51.38 | 2.85× |
| Drop routes over 10 km | 17.91 | 46.61 | 2.60× |
| Drop routes over 5 km | 17.41 | 38.00 | 2.18× |
| Literal crossings only, under 25 m apart | 24.05 | 92.42 | 3.84× |
| Share of points above index 50 | 20.4% | 50.6% | 2.48× |
"Beyond ORR" here means everything past 26 km, including points up to 85 km from the centre. On 26 October 2026 we found that 12.2% of the measured set lay outside the study area this project had declared. Those points score lower than the true fringe, so every ratio in this table understates the fringe rather than overstating it. The ring table above is the corrected view. The next run will be clipped to the declared area and these figures replaced, not revised.
| Area | Severity | Index | To cross | Points |
|---|
Areas with fewer than thirty measured points are computed but excluded, because a median over a handful of points is noise rather than a finding.
Read the ranking, not the numbers. The open map under-records small connections — median openings and service roads drivers use daily are often missing. That inflates every absolute figure here by an unknown amount. It affects all areas alike, so comparing one to another holds, while individual values do not yet.
Where a typical route runs past ten kilometres to cross twenty metres, a hole in the map is the likelier explanation than a real detour, and the row says so. Those are the first places we go back and correct the map.
A descent from the globe down to a single junction, with severance carried in height and colour, and a click through to any area's own measurements. It is specified and it is not built, and the reason is worth stating rather than hiding behind a spinner.
Real building heights from open satellite-derived data, so a severed block reads as somewhere people live rather than a shape on a plan. Severance encoded in height and colour together, so the worst places are visible in the opening frame without reading a legend.
Terrain. Hyderabad spans 32 m of elevation across the whole measured area, so relief here would be drama invented in a shader. That was tested before it was designed, and the threshold was written before the test ran.
Street View is Google's and is never used here, at any point, in any form. The open alternatives were checked on 25 October 2026: Panoramax returned one image across the entire 85 by 77 km area. Mapillary is still being probed.
If neither has real coverage, there will be no imagery layer and this page will say so, rather than shipping a button that works one time in three.
Why it is not here yet. Polish reads as authority. A beautiful model of a number nobody has checked would persuade people of something we do not know, and when it broke it would take the method down with it rather than just the picture. The model comes after a route has been driven.
Distance and geometry only. Congestion and roadworks are deliberately excluded — a median still stands when the construction ends, and that is the claim worth making.
The open map holds 407 pedestrian crossings and 212 footbridges for a metropolitan area of roughly ten million, and none within 400 m of Abids or Secunderabad station. Absence in the map is not absence on the ground, so every claim here is about driving.
We publish what an authority recorded and what we measured. Never what we think that means about anyone's motives.
Kept current, in public, because a limitation found by a reader is worse than one we published ourselves.
Every answer below cites where it comes from, so you can check it rather than take it. If an answer here disagrees with the data, the data is right and this page is out of date — tell us.
That is the most useful thing you can send. Say which area or junction, and what the road actually does. A correction to the underlying map improves every navigation app in the city, not only this one.
A monitored email address is being set up. Until then, write to us through the project's public issue tracker on GitHub; every message is read.