Okay, thank you.
There are a couple of lessons or things that I want to talk about here. One is examples from my own experience of how not to do it, which is to be absolutely dictatorial to the departments; or indeed to accept the standard once and expect everybody to be able to follow it, even with what you might at the time assume to be simple guidelines.
What we've been trying to do at the Open Data Institute is to create tools that enable people to know whether or not they're meeting the agreed-upon standards. One of the data formats that I think a number of your witnesses in this and other sessions will talk about is something called CSV. It's publishing simple tables with certain columns in a standard format every month.
In the U.K., every local council—so 454 different authorities—publishes the same data set to the same standards every month, in theory. In practice, you have around 400 elegant variations on the theme. That's not because people don't know what they're meant to be doing. It's because they're following a process and they don't understand it, and they can't see what good looks like. So they get the file at the end that looks like a CSV, and they're happy.
We've built a simple validation service, which enables you to check against a schema. That poor desk officer in his local authority, who thinks he's doing a good thing, can check. He can upload the file and say, okay, that hits the standard.
That's how we're supporting this sort of federated approach towards data setting standards. The mechanics of that, you can do elsewhere.