Recent calls for dramatic civil service reform – a new Northcote-Trevelyan, an overhaul of the civil service code – may offer a promising path to strong, stable, scalable transformation eventually. But there may be quicker wins available.
Our current civil service powerhouse can do better solution design, and they can also fix things sustainably. What they need are the right tools, the right mindset and – hardest and yet most important – permission from the top to move fast and focus on sustainable impact.
Darren Jones makes an appealing case for acceleration in his “move fast and fix things” speech. His remedies – digital transformation, innovation fellows, cross-departmental taskforces – are directionally sound, but they rest on an important assumption: that you can move faster using the same architecture that made government slow.
Enter that familiar paradigm of New Public Management – so nice, they named it thrice: the triple-baked discipline of Strategy, Policy and Delivery.
Whither SPD?
Strategy, Policy and Delivery solved the problems of its day. When Michael Barber and chums set out to tackle rich intelligence and poor delivery, they developed a suite of practices on the basis that linear targets can be used to convene stable systems around well-understood problems. And it worked: hospital waiting times fell, school standards rose. SPD continues to work where stability, durability, accountability and coordination are required – the traditional firmament on which parliamentary government rests.
None of that is trivial.
But SPD has a tougher time with the problems Jones wants to solve now. The digital state demands adaptability and enthusiasm for edge knowledge that lies outside SPD's conceptual grasp. Net zero, health inequalities, and economic inactivity are wicked problems: non-linear, contested, and impervious to solution by decree.
Whatever the original intent, today, in SPD, information flows downward and accountability flows upward. The wider innovation systems have a hard time being heard. Tacit signals are imperceptible to the quantitative umwelt of data-based management. Technology is an afterthought to policy intent and learning is deferred to evaluation, so the system cannot correct itself mid-flight. Pushing harder on that system does not make it more interesting, imaginative or adaptable: you can end up doing less with less.
So what, exactly, do we need to rewire here?
Rewire at source
Civil service reform is often targeted and actioned through senior civil service interventions. The people doing analytical work – the grade 6s, grade 7s and higher executive officers – appear in reform narratives as recipients rather than agents. Joe Hill of Re:State found that 70% of civil servants feel processes get in the way of their work: positive professional ambition tethered by risk-aversion and compliance. If we can be frank, there is a dearth of opportunity for people at these grades to learn the exciting things that can make their work, work. I hear this when they come to my training courses and say exactly these things.
But these are the powerhouse grades: where problems are understood, where evidence is synthesised, where the gap between ministerial intent and operational reality is felt most sharply. Any reform that does not reach these grades is not real reform – it is re-organisation at the top.
Switch circuits
One potential reform starts with the distinction between policy process and solution design. This is not just semantics – their purposes and worldviews are fundamentally different. A policy process generates, through hierarchy, a statement of intent, which is expressed in legislation, regulation and guidance. Technology arrives at the end – a tool to implement what's already decided.
In solution design, technology is considered the point of action from the outset. Not "what does the policy say?" but "what do we need to build?" This technology combines many enabling things: digital infrastructure, contractual agreements, regulatory requirements, hardware, code and user experience; all held in productive tension together, to be tested in real time.
We can go further. Technology is not designed in isolation. To build a digital state or solve wicked problems, three disciplines need to operate together: Knowledge, Technology and Data (KTD).
Knowledge: what do we know (believe and understand)?
Technology: what is the mechanism of change?
Data: what happens when it meets the real world?
This is different from SPD, but not completely original – it is, of course, anchored in implementation science, in realist evaluation, in impact literature and more.
Plug in the Network
KTD is intended to operate through an impact network: trained practitioners anchoring connections outside Whitehall. Knowledge draws in academic researchers, frontline workers and communities whose insights never reach policy papers. Technology is the site of action, covering the full suite of intervention mechanics: digital tools, infrastructure, contracts legislation and more. And data creates real-time feedback from the moments where mission meets citizen: learning becomes continuous, not retrospective.
Critically, sustainability assessment runs throughout: distributional consequences (for example) are primary expectations to be tested against the data in real time. Procedural justice tests are better met by a system that instinctively reaches for edge knowledge, and from a governance perspective, the network is permitted to be self-correcting in ways hierarchies are not.
This is, of course, more demanding than working sequentially. And it is considerably more likely to succeed in convening the networks and systems essential to building the digital state and solving wicked problems.
Show me!
The Vaccine Taskforce is a powerful precedent. Domain knowledge about vaccine science and regulatory pathways; technology of delivery covering manufacturing, cold chain and procurement; real-time data on portfolio risk and trial results – all happening simultaneously, not in sequence. Direct reporting, delegated authority, permission to think differently.
It would be an error to assume that the VTF succeeded only because of abstract emergency privilege. Sure, the VTF team had permission from the top, but the interesting bit is the essential design principles they imported from the science and technology of solution engineering. These principles can be built into non-emergency government structures, if the state chooses to rewire itself that way.
What does this look like?
An impact practitioner’s work looks very different from that of a policy team. For example, they focus on technology as the site of action and convene a network of the right experts to do this. They also spend a lot of time outside of government departments, iterating designs with system designers, testing ideas with front line workers and subject experts and coordinating with other practitioners who are working on cross-cutting projects.
What does not change is the underlying craft that makes the civil service an outstanding practice: drafting, analysis, coordination, accountability, the standards of the policy profession. KTD does not displace these. It deploys them through the network, rather than the department, in order to produce solutions rather than policies.
The deeper shift is cultural, not curricular: to understand that treating external knowledge as a primary input rather than a consultation courtesy is a matter of rigour, rather than weakness. The impact practitioner must operate with deliberate epistemic humility, recognising that the people closest to the problem almost always know something the policy paper fails to capture.
Start small
But epistemic humility without practical tools is just a conference theme. The civil service has the skills and talent it needs to commit to change. What is missing is a structured set of practices to put those qualities to work at speed. The impact network could be one such structure: start with a few small teams convened around the right problems, then test, learn and grow from there.
The moment of proof is not a pilot evaluation; it is the first time a KTD team finds a distributional consequence the standard framework has missed or makes a course correction well before the problem calcifies. That moment is achievable. The tools, the training and the precedent all exist.
What is required now is the institutional will to use them.
Jake Love Soper is a former civil servant who now works in international governance advisory. He is also a long-term associate of Total Training (formerly Dods Training) and designer of Policy Design and Sustainable Impact, a training programme for government departments