Accurate Maps, Wrong Country

Notes on trying to get things done inside the British State

Tower of Babel, Peter Bruegel the Elder, 1563.

The British state is not short of good ideas. Walk the corridors of Whitehall long enough and you will trip over fantastic proposals that never saw daylight. Pilots quietly shelved, two-pagers that circulated for months before vanishing without trace, funding that expired unspent while the business case was still being reviewed. The graveyard is full. The problem was never imagination.

The problem is a system that has become exquisitely capable of consuming its own ambitions. Civil servants find themselves within an environment that is locally ‘hackable’ but globally broken. A place where the tactics for getting things done are an indictment of the fact that tactics are necessary at all. Where the map is accurate and the country is wrong.

I had the privilege of spending the last year in the UK government, working on the implementation of the AI Opportunities Action Plan. In that time, I was involved in the early days of the Sovereign AI Unit, working with a great team on funding the OpenBind Consortium and Encode AI for Science Fellowship, as well as designing our first compute allocation programme. I also supported the development of DSIT’s AI for science strategy.

This is some of the proudest work of my career. I am delighted I made the decision to go into government. But it was also a very radicalising time for me. I watched, repeatedly, as brilliant proposals and brilliant people were ground to dust by a system that at times felt destined for paralysis.

I hope that some of these lessons can be useful for people trying to understand the mechanics of the machine, but also the superb, unflinching officials (and aspiring ones) that still believe they have a role to play in public service. It is very hard to be in the civil service for years on end, while still maintaining the optimism and ambition to strive to deliver much needed improvement to public services and our economy.

I start by mapping out the various challenges I observed during my time in DSIT. Following this, I conclude with a set of seven principles I think can be effective for getting stuff done.

Although I think many of these forces are universal across Whitehall, I don’t wish to claim that this is a regime-complete account of government. My time was spent mostly working with the Department for Science, Innovation and Technology (and to a lesser extent parts of No10), in particular focused on AI. Parts of my experience may well differ if I was working in an alternate department on an alternate domain. My experience is also dwarfed by those that have been inside the system much longer than me. But sometimes an outsider coming in and doing a quick stint will offer a novel perspective compared with those whose career is tightly bound with SW1.

The Challenges

Before I moved into government, I held a view that was typical of a think tanker believing their profession is the critical node. I thought that what was missing most from government was good ideas.

But this was a mistaken belief. I was surprised by just how many good ideas there were in government. We are not short of an exciting, crisp two pager. In fact, if half of the good ideas that were kicking around Whitehall were actually implemented, we would certainly not be in the state of malaise that we currently find ourselves in.

The problem is that most of these ideas never even make it to press release, let alone delivery.

Why does this happen? There is not one singular theme to point the finger at here. Rather, a combination of factors often mean that a valuable proposal rarely successfully runs the gauntlet of the bureaucracy.

The Multitude of Veto Points

In order to actually fund something significant (>£5m although varying by department) within the British State, you likely will need to seek approval from the following people/institutions:

• Your Director, and Directors of other affected teams within your department

• Your Director General

• Your special advisers

• The Minister

• The Permanent Secretary of your department

• (sometimes) An Investment Advisory Committee, which is made up of legal teams, analysts, and commercial/grant making teams

• Other affected government departments (potentially the Cabinet Office)

• The Treasury

• Number 10

Picture a proposal threading a needle through nine separate gatekeepers — each one a potential landmine, each with their own priorities, timelines, and instinct for self-preservation. A single raised eyebrow at the wrong moment, and months of work quietly risk disappearing into a filing cabinet.

The ballooning size of the civil service has made this challenge more pernicious. Take HM Treasury, where the department doubled (!) in size since 2016. What this means is that you now are sparring with a lot of very smart people who see it as their job to comment on your proposals, and show their seniors and ministers they are doing their job.

It is worth reflecting on why these vetoes may be exercised. A very valid defence of these chokepoints is that the British state genuinely has finite resources and bandwidth. Your Thing may well be valuable, but so are Other Things. So the challenge for officials should be, in part, to prove why Your Thing is genuinely important enough to take up such limited resources.

The Proceduralist Fetish

People often comment on the level of process that government requires. Paul Ovenden ties this back to a theory of the ‘Stakeholder State’: a thicket of external quangos and interest groups that soak up the time of the Centre. Although I think these dynamics exist, the problems run much deeper. That particular aspect of is only one contributor to the decline in state capacity.

Many officials will have more ambitious forms of their work collide with what is a very strong rules-based culture within government. Whenever you try to deviate from the norm, it is common to encounter a hesitance to deviate from particular norm from some core civil service teams, not just the quangos. Collectively, this would mean that any time there was a rule or risk that officials come up against, there is insufficient energy to try and work through this.

The case of OpenBind was the clearest illustration I saw of the grip that legal proceduralism has over Whitehall. The consortium involved the UK’s national synchrotron (the only place this data could be generated), Isomorphic Labs (the UK’s jewel in the crown of AI for drug discovery), top academic labs in Oxford, and Nobel Laureate for protein design Professor David Baker. This was basically The Avengers of protein-ligand interactions.

What followed, rather than getting money out of the door quickly to kickstart the project, were months of back and forth across a range of different policy, legal and commercial teams through the business case process that consistently asked questions such as “why have you given this as a direct award rather than through a ‘fair and open’ competition?” and lots of pressure to prove that this project doesn’t violate subsidy control.

Some of the scrutiny through this process definitely improved governance of the project. But the overall posture shown towards what was a bit of a ‘no-brainer’ made getting it over the line extremely difficult. This was meant to be a slam dunk for government. Imagine what it is like for great projects that actually have some actual element of risk attached!

Political Steers are Everything, and They are Nothing

The Prime Minister recently garnered attention for miming pulling a lever, complaining: “every time I go to pull a lever, there are a whole bunch of regulations, consultations, arms-length bodies that mean the action from pulling the lever to delivery is longer than I think it ought to be”.

There is, of course, some truth to what the PM is saying. But it is quite remarkable just how much variance there is within Whitehall when it comes to responding to direct Ministerial requests. This draws some subtle lessons for what it takes to get things done in government, and what likely needs to change. Consider two separate examples of ministerial directions that I saw in DSIT.

One concerned the provision of support to a promising UK startup: one that is genuinely best in class globally in its particular application area. There were concerns that DSIT could not provide this support because it would violate subsidy control.

But ministers possess the legal power to make a final decision on a subsidy even if it goes against the legal advice they have received.

A ministerial direction, where they claimed accountability for this decision, was enough to break through such legal deference.

But there were also examples of the opposite set of forces.

When the AI Opportunities Action Plan came out, requests were sent from ministers to other teams and agencies to provide detail on how they would help to support the delivery of the Plan.

I recall seeing a response from a particular agency, in slide deck form. For each of the fifty Action Plan recommendations, their response basically amounted to ‘continue funding our pre-existing programme’.

Why is there sometimes no attempt at genuinely thinking about shifting work programmes to respond to something the PM or minister has thrown their full weight behind?

Again, in my view, this does not have a neat explanation, but as with most important things, this behaviour stems from a melange of incentives that officials are faced with

A charitable view might consider the cyclone-like policy churn in recent years. Ministerial priorities (and the Ministers themselves) frequently change. As a result, is there much point pivoting on your programme of work when the priority and principal could change the following month?

People can also be particularly territorial over the sources of their subject matter area. This is a natural tendency we can all succumb to at times, in organisations of all sizes. But it is a sharper problem in large organisations that reward building fiefdoms rather than delivering actual outcomes (The Enterprise and Growth Unit in HMT, for example, has shot up from around 70 staff in 2019 to 230 in 2024).

Accountability Sinks

In his book, “The Unaccountability Machine”, Dan Davies describes ‘accountability sinks;’ where a role, committee, process, or institution that is formally responsible for oversight, risk, or decision-making, but is incapable of actually being held to account for outcomes. Davies lays this out in much more vivid detail and with more useful examples than any I encountered, but I want to consider one particular example I observed.

The best illustration of an accountability sink festering that I saw referred to a particular failing organisation that government was continuing to support.

When I asked officials if they would consider calling for senior changes to this organisation, I was told “this is a political decision - we cannot call for that”. When I posed the exact same question to the political team, I was informed ”‘we would wait for officials to recommend this”. I am not trying to assert which of these groups was wrong or right, but there was clearly an inconsistent understanding of who was responsible for what.

The gravitational pull of the accountability sink is worsened by an ‘over-functionalisation’ of Whitehall. Over many years, we have cultivated a system where functions guard their territory fiercely, and in order to deliver projects, officials must seek sequential approval from finance, HR, legal, digital, commercial, project delivery, legislation etc. etc. This is a way of outsourcing and spreading the risk profile across multiple micro-decision points. In practice, it grinds any momentum you may have had to a pulp.

The Context Switching Tax

When you spend your time working in think tank land, as I did, your internet tabs will be cluttered with papers from other think tanks and researchers. It would therefore be natural to assume that once you are in government, this would continue. After all, you are looking for great ideas and suggestions on how to deliver on announced commitments.

But this couldn’t be further from the reality.

A typical day could scatter your attention across a dozen disconnected conversations. A funding review here, a stakeholder briefing there — until the actual work you came to do recedes into a thin sliver of your week, visible only in the gaps between meetings.

If you do not work at it, half of your day could be dedicated to reviewing documents and speaking to teams that will have no material impact on you delivering your work. Joe Hill has written a thundering account of ‘Plus-Oneism’, where an accumulation of small requests from a large number of teams results in metastasised projects and inboxes, while imploding your focus.

It should be no wonder that officials of all stripes bemoan that they do not have the space to do ‘blue sky thinking’ or focus on delivery.

Senior officials often end up being the causalities of this, and they become extremely time bottlenecked. When you have a major hierarchy coupled with Plus-Oneism, it means that senior civil servants (Deputy Director and above) have very limited bandwidth for actually making decisions, despite the fact that the rest of the department are spending a lot of time awaiting such calls.

The Talent Theory of Everything

It is a well-trodden line now to say ‘people, ideas, machines, in that order’. But it is well trodden because it is true.

Most of the exceptional work I saw done during my short stint inside Whitehall was done in small teams by extremely capable, hard working officials.

The veto points that I referred to previously speaks to the extent to which the civil service believes in people’s judgement and trusts them to get shit done. The simple answer is that they don’t.

Part of this eroded faith can be explained through a rules-based culture. Nevertheless, the calibre of people simply isn’t always high enough to endow such trust. The result is what Jen Pahlka denotes as a ‘cascade of rigidity’, where rules, structures, and processes designed to ensure fairness, accountability, and risk reduction pile up over time, largely as a way of protecting government from itself.

This is not a counsel of contempt. Many of the most talented people I have encountered anywhere work in the British state. But talent alone cannot overcome a system architecturally designed to frustrate it.

Many of these challenges accrue to breaking point, in what has proved to be an existential problem for Whitehall — a major talent drain. This flight of exceptional people from public service is something increasingly discussed externally.

But internally, this churn is seen as a feature, not a bug. The best people leave quietly — not dramatically, but with a LinkedIn update and a faint sense of relief.

A regret I have from my time in government is not trying to actively recruit more people from the outside into the system. There is a real power in network effects where if you can bring in a good person, they can bring in further good people. Leaning into that dynamic is something largely ignored by the civil service.

The toxic recipe of learned helplessness, an obsession with rules, and a jungle of veto points makes it extremely difficult to make those dreams a reality.

What Should You Do About it?

Paralysis is not inevitable. We shouldn’t languish, as John Bew says, as a ‘council of despair’.

The future of the nation depends upon people developing a plan to rebuild the broken foundations of Whitehall. But despite all of its flaws, you can still get an incredible amount done within the system. Public service done well is an unparalleled experience and a phenomenal contribution made every day by people across the country.

And it is worth saying plainly: the officials who stay, who keep pushing, who maintain their ambition despite years of navigating this — they deserve much more credit than they receive.

Here is what I found to be helpful in trying to get stuff done.

Finding The Forcing Function

This is something that Henry de Zoete also lays out. For us, London Tech Week served as one of the calendar highlights of the year for DSIT. This meant Whitehall was desperate to find announceables for the conference, especially when the likes of Jensen Huang was in town to see the PM. This allowed us to push for the Encode AI for Science Fellowship to be one of these news items. Often these public moments where a Minister is on the hook for something in front of the entire ecosystem is what incentivises Whitehall to move. Ultimately, this is a sign of failure to an extent. Why is it that we have to wait for a summer conference before the system actually opens up and works as it should?

Short Circuiting Directly to Spads and Private Office (but at your peril)

There are times, such as when a department is inundated with proposals for how to use up their annual underspend, where being able to short circuit the normal process which runs through a particular team can make a world of difference. A single WhatsApp from the right Spad to the right minister can do what six months of formal process could not. One message, typed in thirty seconds, and a project that had been quietly suffocating in a business case review suddenly has oxygen again. Use it sparingly. Use it deliberately. But use it.

However, do this only when necessary! It should be noted that this is a risky maneuver. Because I was a short term secondee, I was likely willing to take more of a chance as I knew I was here for a shorter period of time. It is plausible that this may be regarded by some people as ‘not being a team player’, and others that spend longer within the civil service may not be willing to ruffle feathers in this way. There were probably times where I over-stepped the mark here, and it is important to learn from those moments. But at the same time, if you aren’t ruffling some feathers in Whitehall, you aren’t pushing hard enough.

Getting External, High Value Endorsements (or better, co-investment)

For OpenBind, we had made a lot of progress getting parts of the system excited. We made a strong case that this could lead to future ‘AlphaFold moments’, which had re-emerged in the media following Demis and John Jumper getting the Nobel Prize a couple of months prior. But it was only when we secured written endorsements and commitments to joining the consortium from Nobel Laureate Professor David Baker and London-based AI for drug discovery leader Isomorphic Labs that we were able to build true momentum to get it over the line. In some ways, this is turning the ‘Stakeholder State’ against itself. Sometimes the most effective thing you can do inside the machine is hand it a signal it cannot ignore from outside.

Challenge Legal Teams More on Their Advice

It is the job of legal teams to be cautious. And there is often good reason to exercise caution with high ambition work. But we have somehow ended in a position where we have internalised that any ounce of legal risk equals a veto. It is my view that the British State has a much stronger deference to rules, norms, and non-binding legalism than comparator nations.

A deluge of legal advice can sometimes break down upon a second opinion (the Attorney General Office lawyers are useful allies here). But if the Minister (and often it will need to be a Secretary of State) doesn’t exercise authority and accept a moderate level of risk, you can end up in a labyrinth of meetings leading to nowhere.

Call out the Accountability Sinks

If you have the appetite to ruffle feathers, you can actually help to fix short term accountability sinks. Often a failure mode to unblocking a problem is simply getting all of the relevant senior decision-makers in the room to hash out an issue like this. The Vaccines Taskforce benefited from uniting a lot of the functions of finance, HR, and legal into the same forum so actual accountability was provided and decisions could be made more quickly. A mistake I made was not trying to create more of these forums to surface the right decision-making points.

Be Ruthless With Your Time Management

One of the ways you must show bravery in upsetting people is sometimes learning to ‘say no’ to requests that are not a priority of yours. This isn’t to say you should not try to help others. But clearly signalling discipline around how you spend your time is essential to managing projects effectively, and will send a clear signal to those around you about what should be expected.

Furthermore, be clear of ‘the ask’ with senior officials, otherwise you are wasting your time in communicating with them. It took me an embarrassingly long time to understand that the carefully argued email I had laboured over for thirty minutes would receive, at best, a thirty second skim. Senior officials are drowning. Put the ask in bold at the top! Assume they won’t read your email to the end. In fact, you can’t even assume they will read your email at all, because they get so many. It might feel like you are having to spoonfeed them, but this is likely the best approach to get to a decision.

Find Your Co-conspirators, and Bring in More of Them From the Outside

You can’t rely on the best people to be introduced to you by your line manager. Your job, if you are inside the state, is to find those people (and the people that they also rate), and build trust fast. There are a lot of ‘situationally aware’ people within Whitehall that can move quickly, channel authority, and get various things over the line. In fact, building trust can be quite easy because it is quite easy to build bonds when recognising the chaos.

In fact, part of your responsibility as a newcomer into government should be to bring an entirely new network in with you. Bring in experts and excellent operators, and it improves your optionality to deliver in pretty much every dimension.

This is not just a responsibility for officials. Secretaries of State and Ministers are normally totally passive in HR and organisational management. The ones that buck the trend are those that build their own team and push through this built up procedure, often succeeding in the process.

If you don’t have hiring power, introduce those with influence over hiring (often senior officials and Spads) to exceptional people that are interested in doing a stint inside government. There is a real network flywheel where good people bring in other good people who then themselves bring in other good people.

Conclusion

It is incredibly easy to be downbeat about Whitehall dysfunction. It has blackpilled me for sure. This is one of the central issues of our current crisis, and one that the main parties are nowhere near reaching to resolve with a proper plan. We are coming to the end of something, but we don’t quite know what specifically will wither, and what comes in its place.

But this doesn’t mean that people just need to wait for some major system-level change to come. Every official working in government today will see high performance teams to some degree. Get the right people in, tilt incentives in your control, speak to the right external counsel, fix the accountability sinks. Many of these things are within the reach of a hard working, small team of officials.

The fact that these tactics exist, that they work, that experienced people need to be told them — that is the diagnosis. Many of these lessons together, suggest that the system is locally navigable but globally broken. That the tactics work at all is not a vindication of the system. It is an indictment of it. Every workaround is a monument to something that should have been straightforward.

The courage is already there. I saw it repeatedly, in small teams doing extraordinary things against the grain of everything around them. The question is whether the system will ever be reformed enough to stop requiring it.

Thank you to Jack Cottam, Tone Langengen, Sam Currie, and Ben Johnson for kindly reviewing and providing excellent feedback. Their role reviewing doesn’t imply endorsement of everything in the piece!