Reforming Ordnance Survey

People have been talking about the Ordnance Survey rather a lot recently.

It’s a very strange beast. It has mountains of really useful data: electoral boundaries, postcode databases and the locations of all sorts of buildings, not to mention roads, railways and green spaces. Unfortunately, it’s a quasi-independent body, which has to pay its own way. It charges heaps of money for access to all these datasets, and makes them available under strange and deeply inappropriate licences.

This has long been a problem for civic hackers, and by implication, everyone else: if you want to make a service that needs to turn a postcode into a geographic location, you need to use their data, and most of the time, you can’t. It costs an arm and a leg.

Fortunately, it’s clear to everyone that the OS needs to be significantly reformed. They own datasets like electoral boundaries, that are fundamental to the operation of a democracy. They’ll charge you if you want to use them, even if you’re the government and you’re organising an election. That’s just nuts, and we all know it. So what to do?

Thankfully, people have been working on it. The recent Power of Information Taskforce report had lots of good things to say: Recommendation 7 was almost perfect. Liberate postcodes and boundary data. Make basic mapping data available for free to all, for modest use. Simplify licensing. Its only real downside was the absence of any mention of derived data.

Put simply, if you use an OS map to create a database of something — like postboxes, hospitals or parks — then OS share copyright in that data. In this modern age of user generated content, that position is completely unsupportable. It’s a shameless grab for intellectual property, motivated by their desire to receive extra royalties. For a private company, that might be fair enough, but it’s reprehensible for a body that exists to provide a public service.

For anyone who thinks this is merely a theoretical problem, it may be of interest that all those lovely crime maps launched by police forces across the UK are probably in violation of the OS’s licensing terms.

This, then, was the background to the latest budget. Having heard that it would contain some new announcements, we were waiting for its publication with baited breath, and indeed: it promised reform. The following day, Ordance Survey published their new commercial strategy. It is underwhelming.

There are some good changes: more data, including boundary information, will be made available through OpenSpace, their API. They’re going to revise their definition of “commercial” so that sites that use their data can carry advertising without being required to pay for licences. But that, more or less, is it. The rest of the strategy revolves around converting people using free licences into ones that become financially sustainable so they can pay. Fair enough, but hardly groundbreaking.

The real problems remain. OS still own electoral boundary data and postcode boundaries & locations. They still decide if you’re commercial, and you still have to accept their onerous licences to use their data. OS still maintain a stranglehold on any data that they consider to be derived from theirs. They’ll still charge you royalties to use that data, of at least £1000 a year. You still won’t be able to do anything with that data that’s not acceptable under their licence, like adding it to OpenStreetMap.

This new strategy is progress, but only just. It is at best a fractional improvement upon what we had before.

A lot more needs to be done.