The Empty Architect

The building that houses a government department often tells you everything you need to know about its current health. In the mid-century, these were cathedrals of bureaucracy. They were imposing and perhaps a bit drab, but they hummed with the energy of thousands of career civil servants who expected to retire in the same building where they started. Today, walk through the halls of a major ministry in London, Canberra, or Washington, and you will notice a different kind of quiet.

The people are still there, but look at the lanyards. A significant percentage of the people making the most consequential decisions for the public do not actually work for the public. They are "partners," "associates," or "leads" from global consulting giants. They are the temporary tenants of the state.

This is the era of the hollowed-out government. In most liberal democracies, we have spent the last forty years engaged in a massive, quiet experiment: the outsourcing of the state’s core cognitive functions. We did not just outsource the janitorial services or the cafeteria. Instead, we outsourced the thinking. We outsourced the design of our welfare systems to software firms, our emergency responses to management consultants, and our infrastructure to private equity.

The result is not a "leaner" government. It is a government that has forgotten how to learn.

The emergency room of the state

To understand the stakes, we have to look at what happens when the lights go out. During the early, panicked months of the COVID-19 pandemic, governments worldwide realized they had no "muscle memory" for a crisis of this scale. In the UK, the response was a frantic call to Deloitte, PwC, Ernst & Young, and KPMG.

On the surface, it worked. Testing sites were stood up, and logistics were managed. But accounts from within the civil service point to something harder to measure: a sense that core state capacity had been temporarily displaced rather than reinforced. The Institute for Government has noted that the speed of the pandemic response led departments to buy in external capacity at scale, often leaving little time to integrate that work into existing systems or lines of accountability. In practice, this meant that key functions were sometimes carried out alongside, rather than within, the structures of the state.

It was not just that the consultants were expensive, though they certainly were. Reporting by the Financial Times found that some consultants working on Test and Trace were billing up to £6,000 per day at the height of the crisis, far exceeding the prorated daily cost of even senior civil servants. A review by the National Audit Office similarly found that the scale and pace of outsourcing created parallel systems that were not always well integrated into public health infrastructure. The issue was not only cost, but structure: consultants were often operating as a semi-autonomous layer, with their own workflows, incentives, and chains of command.

When the consultants packed their bags and left, they took the "learning" with them. They did not just deliver a service: they took the data, the process maps, and the hard-won experience of what worked and what did not. While the National Audit Office's review of the UK's pandemic procurement focuses on the procedural breakdown and the "diminished transparency" of contract documentation, the systemic implication is what I call the "Knowledge Vacuum." By failing to properly document the "how" and "why" of these emergency awards, a key finding of the NAO, the state was left with a pile of reports but was no wiser than it was in February 2020. A government that does not own its systems is a government that cannot diagnose its own failures.

The algorithmic shadow

Perhaps the most dangerous frontier of this shift is the "Contract State" in the digital age. We are no longer just outsourcing labor. We are outsourcing the logic of the law itself.

Take the 2021 collapse of the Dutch government. It was triggered by a scandal involving an automated fraud detection system. The system, designed to catch "cheats" in the childcare benefit program, ended up destroying the lives of thousands of innocent families. It stripped them of their income, forced them into debt, and in some cases, led to children being taken into foster care.

The chilling part of the Dutch childcare benefits scandal was not just the error. It was the opacity. While the software was developed by the authorities themselves, its logic functioned as a “black box” to those it governed. The primary failure was not only technical, but structural: essential information about how the system operated, including the risk indicators driving its assessments, was known within parts of the bureaucracy but not accessible to line officials processing claims or to the citizens being investigated. The result was a system in which life-altering decisions were made through processes that were difficult to explain, challenge, or meaningfully scrutinize. The state had delegated its most sensitive power, the power to judge its citizens, to a piece of code it did not fully understand, even within the institutions responsible for enforcing it.

When we talk about "proprietary architecture" in government, we are not just talking about tech specs. We are talking about a loss of democratic accountability. If a citizen cannot challenge a government decision because the government itself does not know how the decision was reached, we have exited the realm of liberal democracy and entered something much more Kafkaesque.

The consultant’s grammar

We must also talk about the "Consulting State" as an intellectual problem. Consultants bring with them a very specific way of seeing the world. They speak in a grammar of "deliverables," "KPIs," and "optimization."

This language is seductive to politicians because it promises clarity in a messy world. But the things that make a society function, such as equity, long term resilience, and the quiet trust between a citizen and a clerk, do not fit neatly onto a PowerPoint slide.

When you hire a consultant to reform a social service, they look for efficiencies. They look for what can be measured. You can measure how fast a phone is answered in a welfare office, but you cannot easily measure whether the person on the other end felt heard or whether the advice they received prevented a crisis. While bureaucrats are certainly not immune to the lure of metrics, the consultant’s mandate is uniquely tied to "deliverables" that justify a contract’s cost. Because a consultant lacks the long-term "skin in the game" of a career civil servant, they are incentivized to prioritize visible, short-term wins that can be captured in a final slide deck. Over time, this pressure forces the state to prioritize the metrics over the mission. We are optimizing our institutions into a state of clinical soulnessness.

The carillion ghost

The argument for outsourcing is almost always efficiency. We are told that the private sector is more nimble, more innovative, and crucially, cheaper.

But this is often an accounting trick. The short term savings on a balance sheet are frequently wiped out by the long term cost of "Vendor Lock-in." Once a government has embedded a private firm’s technology into its core nervous system, the cost of switching is so high that the government becomes a hostage.

Look at the 2018 collapse of Carillion in the UK. This was a company that was "too big to fail" because the state had outsourced nearly everything to it, from the construction of state of the art hospitals to the cleaning of primary schools. When Carillion imploded under a mountain of debt, it exposed the fiction of the "risk-free" contract. The private sector took the profits during the good years, but when the collapse came, the public was left to pick up the bill and keep the schools open.

This is the central paradox of the outsourced state: you can outsource the work, but you can never truly outsource the risk. If a private contractor fails to deliver water or electricity, the government still has to provide it. As analyzed by the Institute for Government, we have created a system of socialized losses and privatized gains. The state acts as the ultimate backstop for companies that are incentivized to cut corners.

The erosion of the public mind

There is a deeper, psychological consequence to this trend. When a government stops building things, it stops attracting people who like to build.

If you are a brilliant young engineer or a visionary policy designer, why would you work for a government department where your primary job is to manage a contract for a private firm that is doing the actual, interesting work? You would not. You would go work for the private firm.

This has created a devastating brain drain. Our public institutions have become "thin." They are staffed by people who are excellent at procurement and contract law, but who lack the deep, technical expertise to challenge the vendors they are supposed to be overseeing. When the state loses its "Public Mind," it loses the ability to be a sovereign actor. It becomes a middleman, a coordinator of other people’s expertise.

Rebuilding the internal brain

How do we stop the rot? It is not as simple as banning consultants. In a complex, technological world, governments will always need to buy in specialized skills. The goal is to move from substitution to supplementation.

We need to treat institutional knowledge as a strategic asset, as vital as a gold reserve or a military. This involves several critical steps. First, we must implement a knowledge transfer mandate. No major contract should be signed without a legally binding requirement for the private firm to train and upskill their public sector counterparts. If a consultant leaves a building and the civil servants are not smarter than they were when the consultant arrived, the contract has failed.

Second, we must invest in elite public tech units. We need to look at models like the UK’s Government Digital Service (GDS) in its early days, or the United States Digital Service. These are units that brought top tier engineering talent back into the heart of government by offering something Silicon Valley cannot: the chance to work on problems at a scale that actually matters. They showed that when you give smart people the power to actually build things, they will often choose the public mission over a corporate salary. However, this idealism is increasingly hitting a wall of economic reality. As private-sector compensation continues to outpace the civil service, the traditional "security-for-salary" trade-off has frayed. In an era where public institutions are being aggressively streamlined and job security is no longer a guarantee, the state must find a way to offer more than just a mission, it must offer a viable, protected career path for the talent it hopes to retain.

Third, we must end vendor lock-in. Governments must mandate open source standards for all public infrastructure. There should be no more "black boxes." If we pay for a system with public money, the public and the state should own the code. We can see the cost of getting this wrong in the F-35 program, where proprietary intellectual property rights have allowed contractors to lock in decades of follow-on work. By contrast, the Army’s procurement of the Joint Light Tactical Vehicle (JLTV) prioritized public ownership of technical specifications, a move that allowed the military to successfully recompete the contract and break the monopoly of the original manufacturer.

Finally, we must recognize this as a new liberal priority. We need to stop seeing state capacity as a partisan issue. It is a fundamental requirement for a functioning liberal society. Whether you want a large state or a small one, you should want a capable one. As Jennifer Pahlka argues in Recoding America, building this capacity requires more than just new tools; it requires a cultural shift. We must move away from a civil service obsessed with rigid procedures and seniority and toward one defined by merit and results. A state that cannot execute its own laws is a state that is vulnerable to capture by special interests and populist anger.

The sovereignty of knowledge

Liberalism, at its best, is about accountability. It is about a government that can justify its actions to its people in a language they understand. When that justification is buried in a private contract or an opaque algorithm, the democratic link is broken.

We are currently living in a vassal state of our own making, dependent on a handful of global firms to keep the lights on and the checks moving. Rebuilding state capacity is not just about saving money or improving efficiency. It is about reclaiming our collective sovereignty.

We need to remember that the state is not just a platform for the private sector to play on. It is the primary instrument of our shared life. It is the only entity we have that is tasked with looking out for the public good rather than the quarterly profit.

If we want the state to govern effectively in the 21st century, we have to let it think for itself again. We have to give it back its mind. The coffee is still brewing in those ministry halls, and the lights are still on. But the real work, the hard, messy, beautiful work of building a society, needs to happen inside those walls again.

The question for the next generation of leaders is not just what the government should do. It is whether the government still knows how to do it.

Featured image is "Empty Headed?" CC BY-SA 2.0 Steve Fareham 2021.