The coming storm

In which I present an entirely un-asked for, unscientific and unsourced take on the last 25 years of digital government with some reference to the Cummings tornado that’s sweeping in. It breaks my rules on blogging word counts by some distance. But as I don’t really blog now on this sort of thing, I don’t care.

I’ve seen enough of this now to know the cycle. You all know the cycle. Government business is being done badly. Everyone’s fed up. Influential voices outside the incumbent delivery team grow and grow. Eventually they get a go at it. Rinse. And repeat.

[I talk about “government business” here in what I recognise is a limited sense: some see it as embracing the machinery of democracy, of public engagement and policy shaping, in addition to the management of information and delivery of services. The big picture isn’t wrong, but I’m focusing on the latter aspects for the purposes of this post: information and services.]

Old-style IT teams within government were given the job of trying to make sense of this new internet thingy 25 years ago. They had a pop at it, but were largely hamstrung by two things: leviathan contracts that meant they couldn’t change a text string without a change request passed through the supplier’s change process. And the prioritisation board. And the overarching TDA. And then another board. And a steering group.

The second anchor on the ship was departmental separation. The quite-good-in-theory requirement for all government activity to trace a neat, linear chain of accountability up to a particular minister tends to create all sorts of horrible friction when we’re talking about the sort of change that the internet enables. (And this wasn’t even a particularly new problem: as soon as information processing scaled beyond the simplest manual record-keeping, we saw benefits being haggled against by those who had something to lose by changing.) As I say, not new, not even confined to public services.

So things didn’t go very well.

Then the cross-cutters appeared around the end of the 90s and into the start of the new millennium. Usually emanating in some way from the Cabinet Office, we had our digital envoys and czars and centralisers… and our digital strategies. These said a lot about “common good” projects – about doing something well once so that everyone else could use it. We didn’t yet have the language of government-as-a-platform (and the ‘platform’ word of course has much more to it than just enabler or infrastructure initiatives) but you can see some similarities.

They were also about stuffing services onto the internet, no matter how badly – and even if ‘web-enabling’ was just a cipher for hosting a pdf form you had to print out, complete, and post. (I’m not joking, younger readers.)

Still the monster contracts, of course. The mechanics of financing meant that it seemed easier to let a contract for a hundred million than for a grand. To spread umbrella contracts over 30 years, hiding the details, letting someone else take on all the difficult stuff – all under one roof, all somebody else’s problem. The contracts were so big that only a tiny number of organisations could even contemplate bidding for them.

So that all went really well too.

In terms of the explicitly digital part, the tension between departments and centre led to strange chimeras. Yes, you can decide these parts of the policy and service stack – they’re your core task! – but you have to use this common hosting, or communication, or website structure. Even if the overall soup still tastes a bit funny (and doesn’t really reflect user needs) at least we got you all in one pot. I’m talking at this point about Directgov, which I had a finger or two in.

Let me rewind a bit. You’re probably thinking why is this photographer rabbiting on about digital government things? Truth is, I wasn’t always a photographer.

I started my career in public services technology in 1975, fascinated by the surveillance state. Ok, by technology, I mean the back of a school exercise book and a pencil, and by surveillance I mean my brief attempt to log everything I could find out about my classmates in a nice ordered structure. I shall draw a veil over the precise nature of my authoritarian ambitions, but I’d begun a lifelong fascination with what we can reliably (or sensibly, or ethically) know about people; how to classify their needs, understand and serve them.

Shifting forwards 20 years after a stint writing code in the bowels of a nuclear fusion reactor I landed up doing a temp job in a council housing department. Databases, system migrations, bit of admin. The way temp jobs used to in those days, being vaguely competent and curious led to promotions. And after a couple of years I was heading up the IT client function. Old-timers will remember the era of compulsory market-testing of services, and the creation of artificial client/contractor splits in some councils (everybody worked for the council, but they pretended they were delivering services under contract, a mockery designed either as a shield against, or a preparation for, a formal tendering process depending on your viewpoint.)

So: some of my baggage, before I go any further. I’m a creature of the database era, not the web. For some really weird reasons I refused to have anything to do with the web until horribly late in the 2000s, so I missed out on much. I’m also an unashamed transactionalist. For me, making processes work better is a huge buzz. The design and presentation of information is also important of course, but cutting out the systemic failures and the cargo culting inherent in so many transactions was my particular area of interest.

And having gone from the council to a big firm consultancy by the simple act of a) knowing a bit about public services as commissioner and supplier b) being rather too handy for my own good at talking on the fly about all this technology stuff to people who needed help understanding and c) fancying a tripling of my pay, I found myself around 1998 in a KPMG office having what became a very familiar conversation.

“These services are really awful!” “We can do them so much better!” etc. etc.

The same song that drove the cycle I started this piece with. Yes, I was outside the civil service. But I had access within it to people and projects with 9 figure budgets. (I know, all very unjustified.) I was a small cog in the ‘advisory’ machinery, but could spot a couple of things that were simultaneously true: the public payroll was stuffed with incredibly smart people who cared passionately about doing a good job. And that for almost any important decision with a ‘technical’ or ‘specialist’ angle, they were turning to us: the wraiths outside.

Yes, reader, I was one of the problems Cummings talks about. Oxbridge, overpaid, very likely under-invested (if my project went belly up, as so many did, I’d just be allocated to another one). But I can tell you all sorts of nice stories (most of which I do almost believe) about how we made better use of resources by being temporarily in the right place at the right time, rather than weighing down the public purse in the form of inflexible headcount, and about all the terrible projects that were made ‘better’, if not exactly ‘good’.

And then I pitched up at Directgov, first as a consultant on the payroll of a big outsourcing firm, then when even I couldn’t swallow that level of wrongness any more, an independent contractor. The Directgov strategy was fairly straightforward. To manage the tension between departments who wanted to do whatever they wanted on the web with the need for some form of coherence across services. I’m being slightly unfair to the departments; some of their work was related to user needs. But some wasn’t. The awkward ‘franchise’ model for content meant endless editorial battles about how things should exist.

One of my roles was to try and come up with a framework for what should exist at all. Dark days, deploying razors of rationality as fast as I could, but with little to show for it. On the transactional services side, the Directgov aspiration ran as far as something about consistency of branding and presentation, but recognising that departments were never going to be within scope for a full process and tech redesign. How could they be, when budgets ran into the billions, and vast contractual edifices powered things like tax and benefits? So, bring your transactions, slap an orange header on them, and access them via Directgov. Let’s all pretend. Transformational it wasn’t.

Having already confessed to being an unashamed transactionalist it’s no surprise that I found this whole “white-labelled transactions” approach underwhelming. There’d been one genuine transformation of a service – the tax disc – but even by the time I arrived we’d been dining out on its stale corpse for several years. (And tl;dr, this is an unusual service – it’s about cars, not people. Who you are and who pays the fee aren’t important, rendering it much more malleable to simplification as a web transaction.)

I was actually quite happy to embrace pretence, to a certain extent. I remember at one of the first govcamps (annual gatherings of digital government people) in 2009 I think, suggesting that MySociety, then a focal point for smart people who were into all this stuff, could do a radical redesign of at least the user interfaces of some of the big transactions. Present DWP (or whoever) with a fait accompli: a beautifully content- and interaction-designed front end that spat out at the back (via an API) precisely what the grunting old legacy mainframes could deal with. “Here, have this, test this, use this – for free!” What could possibly go wrong? How could that not represent something better? MySociety had form, in the shape of services like FixMyStreet, for ripping up the model of how the public sector thought things should work in favour of something that was vastly more user-centred.

Idealistic arm-waving, sure. But any time this type of ‘improvement’ was mooted, the cry was always “lipstick on the pig”! There were other cries – familiar to anyone who’s ever tried to get a change programme going – “but we’re doing that already”/“we’ll have it sorted in x years when the underpinning y bits are ready”/“ah well the law doesn’t allow us to do that” and so on.

Anyway, I still maintain that some pigs do look better in lipstick. Don’t judge me.

As you might expect, this all ended up going…rather badly. It was clear that progress in digital services, in an era of Amazon etc., was falling waaaay behind expectation. And in came Martha Lane Fox and her report, which, with some political top cover, led to everything changing: GDS to the rescue!

In the run up to the report, those voices around the edge – the MySociety types and so on, along with notable clusters inside the BBC, Channel 4, the Guardian etc, were thinking about regime change, and what should come next. I was in some of these rooms. After numerous conclaves and Cabinet decisions, in came the new GDS leadership.

For me, this really signalled the end of my formal involvement. I was the old world, and definitely didn’t have the skills for what GDS was all about – deliverers, not describers. Mike Bracken led a re-energised organisation with genuinely skilled practitioners – some of the world’s best designers, developers and civic society experts. And they hit the ground running, creating noise and friction in equal measure. I went off to a completely different career.

I can’t attempt more than the briefest sketch of what GDS was all about. Disclosure – I went back in there briefly on a short part-time contract to do some communication work in 2013 (and have had many friends there, past and present), but I wouldn’t ever claim to be an insider.

GDS did some things brilliantly, especially at first. Its strategy was focused and clear. I can still recite the basic objectives, even though I wasn’t working there at the time. 1. Fix government on the web. 2. Fix transactions. 3. Go wholesale – an ambition that I initially thought of as mechanising the way government worked together on the internet – using APIs and so on – but which quickly expanded into a wider government-as-a-platform ambition; the old concept of common good services such as notifications and payments, just done really, really well.

Objective 1 was a roaring triumph: not just redesigning services, but completing the mammoth change task of closing down all the individual departmental services and migrating the content – or the content that demonstrably met user needs (not the same thing at all) – under one new roof,

The achievements that surrounded’s success were themselves staggering: the distillation of “content design” as a discipline; the dissemination and coaching of Agile across the civil service; easy-to-grasp concepts like “the strategy is delivery” and “publish, don’t send”; a previously unthinkable commitment to “working in the open”; graphic and stylistic design that was recognised as genuinely world class; the use of DevOps; building in-house technical capability on a level that seemed previously impossible (and free of the death grip of the big suppliers); and resonant changes in culture and diversity.

Phew. Pause for breath. That’s a hell of a list. It’s not comprehensive, either, just the ones I remember most clearly. I could probably continue for a long time.

But it wasn’t all great. Ruffled feathers, perhaps over-zealous “Agiling of Everything”, lack of mission sustainability once the initial leadership rolled into new challenges elsewhere. Genuine diversity at the top table? Hmm. Retention of talented leaders at all levels? Ditto. Just managing the rapid growth in headcount would stress even the most disciplined team.

And though I loved how objective 1 was delivered, objective 2 disappointed. Fixing transactions turned out to be a time-bounded improvement exercise across around 25 services varying in scale, ambition and complexity. I can only imagine the political jousting that was required to get even that list together; that said, there was no “car tax II” in there. In a world where somewhere north of 3 million driving licences are still sent by post every year to a processing centre (as part of the speeding penalty process) where they sit on a desk while a piece of paper is manually marked up before being sent back to their owners, unchanged (risking loss, theft and who knows what else en route) this wouldn’t have been where I would have chosen to start. I was heavily influenced by Alan Mather’s post on this: transactions are the real prize; it’s well worth a read. But perhaps mercifully, that service selection wasn’t in my gift to make.

Objective 3 seemed to stutter, then regather momentum with Notify and Pay getting proper traction. Verify (the one I was most interested in and will say least about) didn’t, and ended up getting used as a stick to hit all of GDS with. “We gave you all that money, and what did you do with it?” Whatever the truth of why it failed (I have my theories, but will save them for another day), the brakes seemed to start going on. Bumpy leadership handovers, what looked like a reclamation of control by the departments, and an apparent dilution of ambition all took us to where we are today.

Oh yes, and the communication dried up. All that “working in the open”. One reason why it’s so easy to rattle off what GDS was all about in the early days was that openness. Blog (and sub-blog) overload at times, maybe, but that instinct to share what was happening, for better and worse, had been very embedded. Ending it was a great loss to us all. Again, I don’t know exactly what change in leadership, culture or habit caused it, but I hope we find out one day.

The reason I’ve just done a 2,000-word rattle through digital government cycles is this:

There’s a new one rolling in, fast.

Dominic Cummings only wrote his now-infamous “assorted weirdos” post two days ago. It already seems like an age. I’m not going to dissect the whole thing – others have done rather better at that – but I have an ear for the tone.

When something like that comes out and makes such a noise, I listen for the echoes. It didn’t take long. Tweet threads like this have an unmistakable resonance.

I’d never want to see cultural hegemony at the heart of government’s digital capability. Not healthy. But to see a “digital leader” so quickly invoke Brendan O’Neill and tie the decline of GDS to its “woke tics” is something else indeed. This last take gave me great pause indeed. The work done on diversity and acceptance – one of the most important achievements that resounded far, far beyond GDS – dismissed like this, is hard to believe. Firstly, that work was essential, humanising, empowering and unambiguously for the better. Secondly, to argue that writing about mental health, neurodiversity, tampon clubs or freedom to fail is somehow destructive to the delivery of public service outcomes is bizarre. Thirdly, to address an audience of the people who have cared most about digital government, using terms like that, is pure cultural warfare.

But is that outlook, and the Cummings ad, and all the emerging voices like them, just who we were ten years ago? The voices around the edge, saying they can all do it better. Readying themselves to have a go. I’m not in that number now; it’s who is that really worries me.

Whatever’s coming next is going to look very, very different. Brace, brace.


Footnote: Paul Shetler’s untimely death was announced in late January 2020. Shortly prior to his death he updated his Twitter bio with the last line of this blog post. RIP.