What does government do anyway?
It’s fair to say that government does a lot of stuff. Usually, when I’m talking about “delivery of services”, I’m referring to things that impact a large number of people (driving licenses, tax returns) or citizens with a dependency on the state or with complex needs (universal credit, personal independence payments).
The reality, of course, is that the remit of the UK government is vastly wider. The best data visualisation for government spending I’ve found is this one from the Guardian with data from 2011–2012 (if there’s an equivalent one from the ONS with more recent data, I couldn’t find it). A number of huge things jump out from that, which people may not immediately think of as “government”.
That might include some of the following:
- The whole of the NHS
- Education
- Support into work
- The military
- Courts and prisons
- Road and rail
On top of that, there’s a significant number of more obscure large areas of spend.
A very small number of examples are:
- Physical infrastructure such as buildings (more often in the news these days)
- Public health
- Social housing
- The Nuclear Decommissioning Authority (£4bn in 2024)
On top of all that, there are still more important organisations with relatively small spend, such the Home Office (immigration), Foreign Office and International Development — to name just three.
Finally, there are the properly obscure groups in government who are doing vital work that only impacts a relatively small number of people, like the The Rural Payments Agency, The Veterinary Medicines Directorate or the Medicines & Healthcare Products Regulatory Agency.
Let’s not even get into Research Councils.
The reason for these lists is not just to highlight the breadth of what government does — but also that it’s extremely difficult for anyone, or even any organisation, to record, never mind understand, this often changing environment. You only have to chat to the poor souls who try to maintain up-to-date organograms for ministers or others to use and you start to get an idea of why cross-government collaboration is hard, even before limited budgets and departmental territoriality kick in.
Now, imagine you’re setting up a whole new government organisation from scratch. From the get-go, you want to track in some detail what your organisation is for, as well as how and why it does what it does. How would you even start to do that?
Let me introduce Need-O-Tron.
All the way back in 2011, the shiny new Government Digital Service started out trying to record the what, how and why in user stories. For folks who’ve not come across user stories before, they’re often used in the software development community to facilitate conversations between technical and non-technical folksm to ensure everyone is on the same page.
A user story looks like this:
Alas, after a relatively short period of use, GDS retired Need-O-Tron, probably because it was too manually intensive. To my knowledge, it was never even trialled by any other organisation.
So, why bring this up now? I do have some sympathy for the new incoming government desperately trying to wrap their heads around the detailed machinery of government, while their political capital drains away. However, that’s neither of the two main reasons.
The first is that I’ve recently become quite interested in people surveying and one of the things I came across in my research was outset.ai, a new start-up that promises “the scale of a survey, the depth of an interview.” My wife used to be a market research coordinator, so I know how expensive and complex it can be to perform in-depth surveys with large numbers of people — never mind trying to make sense of the results. I find myself wondering if it would be possible to use automated AI-based surveying to build a bottom-up view of what government does, as it’s understood by all the people actually involved in the doing.
There’s a substantial hype-cycle around AI in government at the moment. I’ve seen some excellent examples of user-focused limited-scope AI products that have made impressive improvements in real people’s lives. I’ve also read some outrageous smoke-and-mirror papers on how government should go “all in” on AI. This idea feels like something small that could be prototyped in a medium-sized government organisation and then scaled up if successful.
Imagine the following scenario. Every civil servant in the organisation is informed that they will receive an email with a link to an “AI interviewer”, which they have to complete some time in the next month. The interviews are capped at 30 minutes. Managers are responsible for ensuring everyone in their team takes part. Employees are assured that the information received will be anonymised and aggregated before being shared with anyone (to allay fears that this is a mass “interviewing for your own job”).
The data collected can be used in various ways but, personally, I’d love to see it become the basis from which a government organisation generates and maintains a public set of user needs (in the form of user stories) that not only help everyone understand what the group does but also allow civil society to monitor which needs are being met.
The second reason is that I’ve just finished re-reading the truly excellent Radical Help by Hilary Cottam. In it, she describes prototyping a number of cross-cutting services focused on prototyping life-event based services, such as ones for young people, older folks or people looking for work. In every case, there were huge benefits gained from just cutting down the number of people from government that each citizen had to work with.
My hope for this kind of research is that, among many other things, it will start to show where government organisations and services overlap and hence help us move much more toward a culture of joined-up thinking and delivery.
Of course, to do this, government organisation managers have to trust their employees and promise to release the results, no matter what is found (I’m willing to bet more than one Permanent Secretary will find out that their organisation does things they didn’t know they did — or vice versa).
Even just generating a list of citizen-facing services the organisation is responsible for would be an amazing start.
If nothing else, can we at least finally get the up-to-date organograms?
