Mar 17, 2016
I’ve got out of the habit of blogging. The problem with that is that posts grow, anyway, inside my head. If they aren’t tended, they tangle furiously, and before you know it you’ve got Sleeping Beauty’s garden to hack through before you can get anywhere.
Then a wise man said: “start small. iterate.” So I will.
A relatively new part of the UK government’s strategy for making public services better is the concept of ‘registers’. They’re explained very well here – in an line, they are “authoritative lists of information you can trust”. Nothing new in that idea, really. Anyone who’s ever built any kind of database has worked with that concept.
Where things get fruity, as always, is when you introduce complicating factors like scale, stability, organisation and good old human nature. (This isn’t an exhaustive list.)
Scale? A register with 100 records behaves differently to one with 64.1 million. You won’t be able to maintain them both using the same tools and techniques.
Stability? The register of countries recognised by the UK government (the launch example shown at Sprint 16) changes quite infrequently. A register of every business created in the UK changes almost by the minute. Different dynamics, different issues. (Something like a register of holders of a particular licence that it’s quite hard to qualify for will lie somewhere between those two examples.)
Organisation? As soon as information supports the business of several organisations, other things start happening. Trust, for one thing. If you’re accountable for sending out information to a particular group of people, then at least if it’s your own database of recipients you’re using you have some sense of end-to-end control over what happens. Get that data from somewhere else, and accountability is separated, possibly diluted. Whose cock, as they said in my early days in Whitehall – when things were somewhat less enlightened and a lot less diverse than now – is on the block?
And that brings us to the human nature bit – if those other things get complicated, then the human instinct is to self-preserve. To silo (please, please forgive me). To build walled gardens. To duplicate, fudge, kludge and, of course, waste… From homespun spreadsheets here and there to mighty Oracle instances, the idea of canonical information is readily sacrificed on the altar of expediency, or just plain old survival. (Obligatory Upton Sinclair reference here.)
So – registers. Top idea. They’ll be definitive. They’ll be owned. They’ll feed and support other systems.
But will they, y’know, work?
I’m not sure. I used to be a young turk information systems type. Now I’m a greybeard, retired from the fray to the far less capricious world of pixels. I’ve seen centralisation, federation and linking of data. The rise, and the fall, of many programmes. The Citizen Account; the Single Business Register; the Government General Practitioner; health resources; gazetteers; land and property; military warehousing systems… you name it, I’ve feebly tinkered with bits of it.
I can’t not like the starting point set out in Paul’s post: begin with the simple principles – and with manageable scale, clarity of ownership, low volatility and all that. Establish how they’ll work, then do more. Start small. Iterate.
So what’s nagging me? I guess having seen so many “the one true…” projects founder, I’ve got a few scars and prejudices. Look hard at almost any data set, and it’s less canonical, less binary, than you might think. That list of countries? So should Palestine be on there then? Perhaps it should, for some purposes – but not for others? I don’t know – I’ll leave that to the foreign policy people to thrash out. But you get my point. Another of my inspirations in this area wrote that “Digital is political“. True indeed, and data is also very often political.
Then there’s this question of volatility. There are inevitable limits on how well you can keep up with fast-changing data. For some purposes it may be sufficient to know which companies held a particular licence as of the first of every month. Other requirements may need that status to be verifiable on a minute-by-minute basis. There’s some serious analysis to be done to make sure that an authoritative register can meet all those needs. Because if it can’t, the hydra’s heads will start to sprout…
And on that “feeding other systems” point – the risk of one hell of a dependency culture springing up. Yeah, er, we’re down today because one of the tables in our system is fed from a, yeah, and that’s down, and, er, well the API’s been flaky for a while and we’re not exactly sure whose problem it is to fix coz it’s on the boundary… And so it goes. To say nothing of the points of failure and vulnerability should someone want to pop a nasty bit of grit into the gears of the government machine… So there’s that.
The trust thing will take some real thrashing out too: you’d think that concepts such as company registration, or the existence or non-existence of a school or hospital would be nice, uncontroversial matters, with clear alignment to one or other government structure. You might think that; I couldn’t possibly fail to quote Francis Urquhart.
Lastly – the old curse of change programmes – it’s really easy to do new, good stuff. It’s very hard to stop doing old, bad stuff. Will teeth be required?
To conclude: I guess I’m cautiously supportive of the concept. I’d really like to see some more development of the vision for what future registers may be created.
I know – it’s a fine line between writing cheques you can’t cash, and setting out some aspirations. But would it be going too far to draw up a wish-list of just some of the more appealing candidates? An authoritative gazetteer, perhaps? The registry of land and property ownership? Companies? Charities? Patents? The Electoral Roll?
I don’t really mind which – juicy ones like those, offering enormous scope and value – or just more niche stuff: good parts to build better engines. But more, please. A little boldness goes a long way.
Not people though. Not citizens/subjects/taxpayers/voters. You and me. Uh-uh. That one’s definitely out of scope.
Or is it? (I’ll leave further exploration of that one for the next post…)
They did actually say that thing about cocks in my first meeting in my first central government job. I had to include it as a piece of social history. Sorry.
UPDATE: 18 March
It didn’t escape me yesterday that the Budget contained something very relevant to all this: £5m to build a new, open address register. It was one of the triggers for writing this post.
I won’t say much, as it isn’t yet clear who’s going to do this and what they’re going to do. Though I think I’m safe in saying that this time it won’t just get shoved through the door of 123 Buckingham Palace Rd in a brown envelope, to be blown on “stakeholder engagement meetings” before any actual work gets done.
I shall merely add:
– there is some devil in the detail of the differences between a gazetteer, a database of delivery points/postcodes, and underpinning geospatial information; their contributors and users are not homogenous, so it’s always good to be clear exactly what’s on the table here (which I’m not yet, because no detail has been published)
– IF this is to be the new, authoritative register, then what’s its relationship to the other one – the one that got flogged off? Both could be maintained in parallel, but that would clearly be absurd. One would have to be the master to the other. Which? How? Who makes that happen? Who pays? And so on.
That’s the big plus, and the big minus, of “authority”.
And even in 1746 Goldoni knew that this was a splendid way to create a farce.