Turkey

What the record shows

Turkey gives this atlas something it almost never gets: a controlled comparison. Two metropolitan municipalities, one legal frame, the same domestic suppliers, largely the same architecture — and results that diverge on decisions having nothing to do with the technology. Istanbul and Ankara both run their water networks on SCADA with district metering areas and pressure and night-flow monitoring. Istanbul's utility has brought non-revenue water down to 18.6 per cent by the end of 2024, from 22.3 per cent in 2019 — among the steadiest multi-year efficiency gains in the atlas. Ankara's opens its own 2025–2029 strategic plan at a 42 per cent loss rate, sets 25 per cent as the 2029 target, lists "high loss rate" among its self-declared weaknesses, and records SCADA renewal as nought per cent complete in 2025. A twenty-three-point gap between two cities running the same playbook. A control room tells you where the water is going; it does not replace the pipe.

The same divergence shows up in what each city does with a feed it already has. Ankara's transit authority pulled off the harder integration: its live arrival app carries not only the municipal fleet but the privately operated services that run much of the network, so the tracking follows the passenger rather than the operator. It has not published a developer feed. Istanbul's bus operator publishes GTFS schedules and real-time vehicle services on the city's open-data portal for anyone to reuse. That is a policy difference, not a technical one, and it recurs one level up. Istanbul's portal carries around 550 datasets under the municipality's own licence, described as compatible with the standard international one. Ankara's is the more interesting document, and it earns a new rung on the licence ladder this atlas has been assembling: its text names CC BY 4.0 and permits commercial reuse with attribution, then reserves the right to publish new versions and states that continued use means accepting them — while the grant it is quoting is irrevocable and cannot be re-versioned against an existing user. The failure mode here is not a missing licence, as in Lagos or Bangkok, but a bespoke wrapper that quietly takes back what the standard grants. Read the wrapper, not the badge.

Ankara's fare table deserves the same close reading, because it answers a question most cities are never asked. The city began accepting ordinary bank cards, phones and QR when it replaced its validators from January 2024, which is normally reported as an upgrade. Accepting them is not the same as treating them equally. Under the tariff in force from January 2026 a bank-card ride costs more than a municipal-card ride, and in the transfer and concessionary columns the table records the bank card as none: no transfer within the card's 75-minute window, no student or teacher rate, no early-morning discount, and none of the free travel the municipal card carries for over-65s. The open loop buys exactly one full-price ride. It is a visitor's fallback priced as one, rather than a second way to be a regular passenger — and it is worth setting beside Mexico City, where an ordinary bank card has opened every station since 2024.

Where the Turkish record is strongest is where a measurement is wired directly to an action, and the clearest case is not run by either city. Istanbul's earthquake early-warning system has been built and operated since 2001 by Boğaziçi University's Kandilli Observatory, linking roughly 110 strong-motion accelerographs across the Marmara region. It can automatically halt the Marmaray rail tunnel and shut off the gas network in the seconds before strong shaking arrives, and it generates post-quake damage maps from the building inventory. It functioned as designed in the magnitude-6.2 offshore earthquake of April 2025. This is the top of the ladder — the observation is not an input to a decision, it is the action — and in a corpus full of dashboards that oblige nobody, it is the exception that shows what the rest are missing. The city's own instrumentation sits a rung lower and is honest about it: the air network dates to 1995 and now spans 27 fixed stations publishing hourly, alongside a dozen national ones, and the M5 metro is one of the few lines in this atlas where the driverless label is stated as unattended operation rather than as a signalling capability, running that way since December 2017 with further lines built to the same standard.

The citizen-facing entries are where Ankara is ahead, and for an unfashionable reason: it publishes its remainders. Başkent 153 has taken requests since 2003, and what distinguishes it from a switchboard is the audit layer bolted on when it was rebuilt — a team that checks whether an answer arrived on time and whether the problem was actually solved rather than closed, complaints that can require a photograph confirming the fix, and a map of which parts of the city raise which problems and how long they have gone unresolved. The number worth having is the ratio it publishes: in seven months of 2023 the line logged 2,027,238 calls and recorded 631,133 formal applications, which is where a request line decides what counts as a request at all. Mor Haritam asks a question incident data cannot answer — it invites women to mark the places they avoid, measuring the fear that changes behaviour before anything is reported — while inheriting the reporting bias every such system has: the map shows where people chose to report, not where risk is. Istanbul's super-app bundles about a hundred services including the digital fare card, and in late 2025 became the subject of a data-handling investigation amid the wider prosecution of the city's opposition-led administration. If the entries below agree on one lesson, it is that Turkey bought two of nearly everything and got two different answers — which is the most useful thing a national record can offer, because it rules the technology out as the explanation.