At a bus stop on Waterloo Road, the dot-matrix sign says the 59 is four minutes away. So does the phone of the woman standing under it, and the phone of the man twenty metres on who never looks at the sign at all. The same prediction, the same four minutes, is travelling down at least three different paths at once: into the orange display overhead, into a journey-planning app built by a company the size of a flat-share, and into a map made by one of the largest firms on earth. None of them generated the number. They are all reading it from the same place — a free public feed published by Transport for London — and that quiet fact is one of the most consequential things a European transport authority has done this century.
The story usually told about London’s mobility data is a story of a sudden opening — a long refusal, then a release. The truth is more interesting and rather less dramatic. TfL did not finally relent. It was, by the standards of European transport bureaucracies, an early and stubborn believer.
The myth of the locked door
It is tempting to imagine TfL as a guardian who spent a decade saying no to developers before throwing the gates open. The documented timeline does not support it. As far back as 2007 the authority was experimenting with releasing data; around 2009 it opened a dedicated developers’ area on its website; and by 2010 the number of people registered to consume its data was already in the hundreds. The London Datastore, the Greater London Authority’s open-data platform, launched in 2010 and began carrying TfL feeds alongside everything else the city published.
The genuinely hard part was never the decision to publish. It was making the data usable. Between roughly 2007 and 2011 TfL was, by its own later account, feeling its way — a scatter of separate feeds in different formats, each describing one slice of the network. The live bus-arrivals API arrived in 2012, in time for the Olympic and Paralympic Games, which forced a full transport-data portal into being on a deadline. Underground train-location and Journey Planner APIs followed. The registered-developer count climbed past a thousand in 2011 and past four thousand the following year. The door had been open for years; what changed was that someone finally built a corridor behind it.
What “real-time” actually means here
The live numbers were never magic. Under London’s buses sits a system called iBus — an automatic vehicle-location platform rolled out across the fleet between 2007 and 2009, the same system that triggers the announcement “the next stop is” in a now-familiar recorded voice. iBus knows where every bus is. That knowledge fed the Countdown signs at bus stops, complete across some 2,500 key stops by mid-2012. The decisive move, around 2011, was to take the feed that drove those signs and make it available to anyone — not just to TfL’s own hardware on the pavement, but to a text-message service, a website, an app written by a stranger.
The same logic now runs across the whole network through what TfL calls its Unified API — a single, consolidated interface that lets a query ask one consistent question of the entire system. Live arrival predictions for a stop: which line, heading where, in how many minutes. Line status for the Tube, the DLR, the Overground, the Elizabeth line, the trams, with the severity of any disruption and the reason for it. Journey planning, leg by leg. And the docking data for the cycle-hire scheme — the BikePoint feed, which reports for every station how many bikes are free and how many empty docks remain. By the middle of the last decade TfL was publishing in the order of eighty data feeds, around three-quarters of them reachable through the API.
The small teams and the giants
The clearest proof of what an open feed is worth is the company that was effectively born on it. Citymapper began in London in 2011 — first, briefly, as a bus-focused tool, before broadening into the multi-modal journey planner that became its signature. Its founder, Azmat Yusuf, a former Google engineer, built the early product around live TfL data at a moment when most navigation apps still treated public transport as a static timetable. The app that resulted made the city legible in a way that the timetable never had, and it did so without TfL writing a line of its code.
Citymapper is the famous case; it is far from alone. The same feeds flow into Google Maps and Apple Maps, into Waze, and into a long tail of smaller bus-times and journey apps maintained by tiny teams and lone developers. TfL’s own count put the registered-developer community in the hundreds in 2010; it has since passed eleven thousand. The point of an open feed is precisely this asymmetry: a one-person app and a trillion-dollar firm draw on the identical public source, and the city does not have to choose between them or build for either.
Our guiding principle ever since has been to make non-personal data openly available unless there is a commercial, technical or legal reason why we should not do so.
Foreword, “Assessing the value of TfL’s open data and digital partnerships”, Deloitte for TfL, 2017
What it is worth, and what it costs
The temptation with open data is to talk about it as a virtue rather than an investment. In 2017 TfL commissioned Deloitte to put numbers to it, and the numbers are unusually candid for a report a public body paid for. The headline estimate was that releasing the data generated annual benefits and savings of up to around £130 million — the great majority of it not revenue at all, but saved time for passengers who could now plan around delays, valued in the range of £70–90 million a year.
The directly commercial figures were more modest and, to TfL’s credit, presented as such. The gross value added by London companies building on the data was put at between £12 million and £15 million a year. The data was estimated to support around 500 jobs directly and a further 230 across the supply chain — roughly 730 in total. Against this, the cost to TfL of publishing the data was estimated at about £1 million a year. Whatever one makes of the precise figures — and they are estimates, with ranges, not measurements — the ratio is the argument. By the report’s own framing, the authority was reaching forty-two per cent of Londoners through apps it never had to build, for roughly the cost of one marketing campaign it never had to run.
The licence is the policy
The part that travels least well to other cities is also the most important, and it is not technical. It is a licence. TfL releases its open data under a version of the Open Government Licence — the same permissive instrument that governs much of the British public sector’s data — which means the feeds can be reused for commercial and non-commercial purposes alike, with attribution, and without a separate negotiation each time. A developer does not have to ask. That single decision is what lets a weekend project and a multinational draw on the same source without a contract between them.
It is also where the limits are honestly drawn. TfL does not release personal or commercially sensitive data; the open feeds describe vehicles, stops, lines and docks, not the people moving through them. Historical cycle-hire journeys have been published in bulk since 2015, stripped of anything that could identify a rider. The principle that the foreword to the Deloitte report states plainly — open unless there is a commercial, technical or legal reason not to be — is doing the real work. The API is the plumbing; the licence is the policy.
What the rest of Europe is actually copying
For a continent now legally committed to opening its transport data — under the EU’s rules on national access points and standardised travel-information feeds — the London case is less a model to admire than a worked example to interrogate. The instructive thing is the sequence. London did not begin with a grand data strategy; it began with a system that already knew where its buses were, and then made the unglamorous decision to let other people read it. The Unified API came later, as consolidation, not as the founding act.
That is the part hardest to replicate, because it is not a product. It is a posture — the willingness to treat publicly owned data as something the public can build on, and to keep doing it for fifteen unbroken years while the developer count climbed from the hundreds into five figures. Back on Waterloo Road, the 59 arrives at the minute the sign and the phones all promised. Nobody notices, which is the whole achievement. The infrastructure that makes a city legible works best when it is invisible — and London’s is invisible precisely because it was, for once, given away.
Sources: Transport for London (Unified API and “Our open data” pages; api.tfl.gov.uk); “Assessing the value of TfL’s open data and digital partnerships”, Deloitte LLP for TfL, July 2017; London Datastore (Greater London Authority); Open Government Licence (The National Archives); iBus and Countdown programme records; the Open Data’s Impact case study on TfL (GovLab); company background on Citymapper. Economic figures are Deloitte’s 2017 estimates, given as ranges.
