When moving the machinery of government, watch out for the technology spaghetti
The departments of the UK government are chopping and changing a lot. Back in 2023, there were five major changes to departmental structures, including the creation of Department for Energy Security and Net Zero and Department for Science, Innovation and Technology.
This week, rumours are circling about the latest MoG (Machinery of Government change, as these things are called) - maybe DSIT splits and merges into DCMS and DBT (the Department of Science, Innovation and Technology splits into two parts and each part is merged into Department for Business and Trade and Department of Culture, Media and Sport respectively).
According to Institute for Government: creating a department can be
“a useful way for a PM to signal their commitment to a policy area “ or
“because of political imperatives or party management – for instance, a PM’s desire to distribute patronage” or
“with the intention of improving the quality of services delivered by government.”
The first two reasons seem a bit… casual. Signalling and patronage come at the hefty, and rising cost of entangling and detangling departments.
“Our research shows that the direct cost of creating a new department can be around £15m, with up to £34m on top resulting from loss of productivity as staff adjust to the new organisation. In one case – the creation of the Department for Work and Pensions (DWP) in 2001 – the estimated cost exceeded £170m.”
The challenge in 2026 is that the machinery of government changes are increasingly becoming a right old faff. You might imagine technology can come to rescue - but it’s far more likely to be the cause. This isn’t because government technology is bad (some is, some isn’t), it’s because there’s really a lot of it, and it’s increasingly interconnected and layered.
To understand this, it’s useful to describe what technology in a government department is actually like. No two departments are alike, but there’s a lot of similarity and patterns.
All departmental tech starts with the basics. Staff need laptops, buildings need WiFi, Microsoft need a big contract because there’s nearly always a lot of Microsoft around. This stuff is a short term faff. But because there’s lots of standard patterns to split or merge, it gets done.
But there’s a sting in the tail. If you want to change people’s email addresses (like the time the government flirted with cyril.servant@levellingup.gov.uk) and you start to pull at the ends of the great ball of technology spaghetti at the heart of every department.
Lurking in databases everywhere, the innocent email address is the key to a user’s identity - it’s part of how the computer knows who is trying to access data or make a change. Change everyone’s email address, and a simple IT configuration becomes a migration mega programme costing £millions, particularly for larger, more operational departments. And this is just the surface of the spaghetti ball.
The great balls of technology spaghetti
The part of MoGs that often gets missed, but drags on for years following splits or mergers, is the ball of technology1 spaghetti that holds all the flows of information together in a department. Complex, entangled technology, with a rich data sauce.
Within the spaghetti, there are layers. Systems sit on top of each other, like a stack of lasagna - the top is what users see (services), and the layers are platforms that provide a toolkit to the services. What turns lasagna into spaghetti is that systems connect sideways and diagonally in intricate ways. Organisations always aspire for neat lasagna, but end up with a spaghetti.
Some technology spaghetti balls are truly chaotic, fraught with decades of unmodernised, unloved legacy technology. Some spaghetti balls are more modern, more well organised, but step back, look at the information flows, and the lines of interconnectivity are complex, regardless. Entanglement is just how information works at scale. No matter how you look at it, it’s quite spaghetti-like.
In the justice system, where I used to work, you could see this come together in really visible moments. The moment when a prisoner first arrives at a prison for example, where data from right across government is needed to do the job well of settling them into prison - sentence details, health conditions, relationships with other prisoners, family contacts, finances and so on. This data comes together with a complex mix of paper, telephone, email, APIs and messaging systems.
Across the justice system as a whole, I recall the Ministry of Justice having around 400 large complex information systems. One such large system supports the everyday lives of every prisoner in an English prison. Another manages every online application for legal aid. A longer ‘long tail’, in the tens of thousands, of smaller systems exist. These range from low code configurations and Access databases through to Excel spreadsheets. The department couldn’t function without many of them.
The modern MoG has to grapple with combining or tearing apart this technology spaghetti. A department splitting apart pulls at countless technology-powered information flows. Merging departments find it deeply challenging to reconcile two balls of technology spaghetti which have formed independently. The separate balls of spaghetti on the plate of a merged department can remain separate for many years after a MoG mirroring the fractures in human relationships created by MoGs, and exacerbated by the cost of combining.
A MoG today, done properly, following through with the full intention of the split or merger, will cost £millions more than historic MoGs - much of this additional cost will be delivering changes to technology and data which otherwise wouldn’t have been necessary.
MoGs are a Very Big Deal
In 2007 the Ministry of Justice was formed when some functions of the Home Secretary were combined with the Department for Constitutional Affairs. At the time, it was felt to be an expensive endeavour, and Public Administration Select Committee commented at the time that:
“Parliament should be given the chance to vote on Machinery of Government Changes. “
With a specific reference to the cost.
“The Government needs to provide a business case for the changes it is proposing. ... Proper scrutiny relies on cost information being available.”
In 2007, technology was already used extensively across all affected areas of government - but the scale of increase to 2026 is exponential. The cost of an equivalent MoG in 2026 would be far higher, reinforcing the argument made for Parliamentary scrutiny. It seems to me that, at the very least, a MoG should be thought of as a Very Big Deal - more like a (de)merger of a conglomerate than redesigning the Cabinet table.
What can Departments do?
Because of the challenges to MoGs that technology brings, departments simply have to ignore the threat of MoGs until they become real. They must act as a confident whole - and with regards to technology, get to grips with their own peculiar ball of technology spaghetti, and find the best ways to stay connected to the information they need inside other department’s spaghetti balls.
What can government do?
We need a bit more creativity. For example, why not open the door to continuous micro-MoGing (this phrase won’t catch on). Why should departmental boundary changes be ‘big bang’ alongside big Cabinet reshuffles. Why not iterate boundaries all the time, make the exchange of small parts of the ‘machine’ of government the norm?
That’s not going to be easy, but it would be cheaper, quicker, less risky, less disruptive, and likely more effective for policy alignment intentions.
My spaghetti warning is all perhaps a bit too late for DSIT. There appears to be much faffing ahead.
For technologist readers - I put the following tech inside the ball of spaghetti: IaaS, PaaS, CRMs, ERPs, case management systems, forms, dashboards, databases, data lakes, integration layers, middleware, messaging systems - the whole techno-informational shebang.