The architects of digital exclusion

Jacob Huibers — Statecraft, April 2026

A woman from Amersfoort becomes the victim of car theft. She files a police report. She notifies the National Vehicle Registration Authority (RDW). She expects the system to register her as victim and close the case. Instead, she is pursued for over ten years by road-tax assessments, motor-vehicle tax, parking fines and ultimately enforcement orders for a vehicle she does not own. No civil servant acts in bad faith. No organisation breaches its competence. And yet the whole of organisations together produces an outcome no one designed and no one can reverse without the citizen herself proving that the system has wrongly treated her all those years.

This case, documented by Arjan Widlak and Rik Peeters of the Kafkabrigade, is no anecdote. It is exemplary of a pattern that has made them, over the past fifteen years, the most thoughtful critics of Dutch digital government. Not from an ideological rejection of automation, not from a romantic longing for the counter clerk of the past, but from a precise analysis of how information architecture fundamentally reorders the relationship between government and citizen. Their work deserves a place within Statecraft because it diagnoses what is usually obscured in professional journals and almost never named in policy documents: that execution has long ceased to be shaped by who works there, but by how registers are interlinked.

The diagnosis the government refuses to make

Widlak began his career not as an academic but as an entrepreneur. He spent fifteen years as director of an IT company before joining the Kafkabrigade Foundation, where he grew into the most-cited Dutch voice at the intersection of administrative law and digital execution. That background explains the exceptional quality of his work. It is not the academically trained public-administration scholar surprised by how organisations work, and it is not the techbro who thinks systems will come right if the architecture is correct. It is someone who has built what he now dissects, and who precisely for that reason sees what the builders of the Dutch system did not want to see.

His first book, De Digitale Kooi (The Digital Cage, 2018), written with Rik Peeters and published by Boom, opens with a simple thesis: Dutch government has placed its citizens in a cage no one explicitly built, and from which no one has explicitly cut off the exit. The casework is staggering. A woman who regularly stays abroad for work is automatically removed from the basic personal records database (BRP) and thereby loses her status in dozens of other systems at once: parking permit, voting rights, the ability to invoice as a self-employed person, everything. The error does not sit in one register. The error sits in the assumption that registers represent one indisputable reality, and that an outcome in that one register can lawfully propagate to all others.

In 2018, the Council of State issued an unsolicited advice on precisely this problem — an instrument deployed only three times in the preceding forty years. The Council concluded that citizens can no longer establish which rules have been applied to them, that it is no longer possible to verify whether rules do what they were meant to do, that no account is taken of specific circumstances, and that the burden of proof on errors is reversed. The advice met with broad agreement and little structural action. Two years later the childcare benefits scandal broke in full, and the digital cage proved no rhetorical image but a factual description of how a constitutional state can grind down its own citizens.

From organisation to infrastructure

What is genuinely innovative in the work of Widlak and Peeters is their conceptual shift from organisation to infrastructure. Classical public-administration scholarship handles execution problems at the level of the individual organisation: the Tax Authority that fails, the employee-insurance agency that runs behind, the driver-licensing authority that is too slow. Classical IT analysis handles problems at the level of the individual application: this system does not work, this connection is error-prone, this database is outdated. What sits between those two levels — and where the actual power of contemporary administration resides — is what Widlak and Peeters in their June 2025 article in Government Information Quarterly have named infrastructure-level bureaucracy: the totality of organisations linked through data exchange, which as a network exhibits properties not visible at the level of the individual links.

The analytical gain of this concept is large. It explains why the childcare benefits scandal is not primarily a Tax Authority affair, even though it is framed that way. It explains why the fraud signals produced by the Fraud Signalling Facility were taken over by dozens of other government organisations, with cumulative effects no one intended and no one designed. It explains why solutions aimed at individual organisations are so systematically inadequate: the architecture that produces the problem is touched by no repair programme, because that architecture sits in no organisation chart.

For anyone within Statecraft looking at Mark Moore’s Strategic Triangle, this analysis delivers a precise diagnosis. The heaviest shortage does not sit on the corner of political legitimacy, because almost everyone agrees that the childcare benefits scandal should not have happened. The heaviest shortage also does not sit on the corner of public value, because almost everyone agrees that the citizen should be more central. The heaviest shortage sits on the corner of operational capacity, and specifically on a layer of that operational capacity at which public-administration training did not look until recently. Information architecture is the operational reality of modern execution, and government has outsourced, fragmented and placed that layer outside its own steering options.

Ten principles no one can ignore

Widlak’s second book, Volwassen Digitale Overheid (Mature Digital Government, Boom Bestuurskunde, 2022), formulated ten General Principles of Proper ICT. The reasoning behind these principles is simple and therefore inescapable. Our constitutional state knows general principles of good administration that have crystallised in case law over a century of administrative practice. But those principles were formulated for a world in which a civil servant read a file, made a weighing, and took a decision. In a world in which systems decide automatically, data are exchanged automatically, and consequences propagate through dozens of registers, those principles need counterparts baked into the architecture itself.

Whoever registers must inform. Whoever decides must give reasons. No automatic decision without the possibility of automatic correction. No automatic execution without execution upon correction. These are principles that sound so obvious that their formulation is itself proof that they are not being observed. Their power lies in the combination of two properties that rarely come together in policy debates. They are legally testable, hence suitable for the courts. And they are designable, hence suitable for the architect. That double character makes them the most productive normative framework formulated in the Netherlands for digital execution, and it makes it remarkable that they have not yet been structurally incorporated in the assessment framework of the Advisory Council for ICT Assessment or in central-government procurement rules.

Nieuwland and the design of the digital rule of law

On 11 February 2026, the think tank Achterkant van de Overheid (Back of Government), an initiative of the Kafkabrigade, presented the publication Nieuwland: a design for the digital rule of law. The title refers to the polders the Netherlands has made over centuries: territories that did not exist, which arose by design, and which were subsequently institutionally embedded. The analogy is precise. The digital infrastructure on which the Dutch state has grafted its capacity to act has come about over more than thirty years without administrative master plan, without constitutional review, and without the checks and balances that for analogue administrative arrangements are taken for granted. It is a new piece of land on which a constitutional state must be set up, and the think tank delivers the design for it.

Two elements in this design merit elaboration within Statecraft. The first is the spreading of information power. The assumption that efficiency arises by collecting data centrally and linking them as much as possible has, in recent decades, been elevated to doctrine and has demonstrably contributed to the digital cage. The think tank counters that information power, like other forms of power, must be spread to prevent abuse and system failure, and that the architecture must facilitate rather than fight that spreading. The second is the restoration of human perception and judgement within the chain of automated decision-making. Not as keystone afterwards, not as objection possibility for those who know the way, but as structural part of the design.

The second theme is related to what in De Richting van de Beweging is called the primacy of anchoring over movement. An executive organisation is judged not on the fact that it takes decisions, but on the extent to which the chain of which it is part remains correctable when it fails. Whoever builds without that correctability, builds without anchoring. And without anchoring, every executive decision becomes a potentially irreversible decision.

What distinguishes the Kafkabrigade

There are more institutions publishing on digital government. The Dutch Scientific Council for Government Policy delivered important work in Voorbereiden op digitale ontwrichting (Preparing for digital disruption, 2019). The Rathenau Institute follows the theme critically. The Court of Audit publishes regularly on IT projects. What distinguishes the Kafkabrigade is the combination of three qualities that elsewhere rarely coincide.

The first is empirical precision. Unlike much academic work on digitalisation, the Kafkabrigade stays close to the case. Not as illustration, but as source. A woman with a stolen car. A homeless person who cannot register. An entrepreneur who through interference of fiscal rules no longer knows what he earns net. The analysis rises out of the case, and descends into it again to test the generalisation. That is a research methodology still too little recognised in the Netherlands, but which forms the basis for the credibility of the conclusions.

The second is constructiveness. Widlak’s tone is never cynical, and rarely polemical. He names sharply what is going wrong, but almost always delivers an action perspective: a design principle, a legal principle, an institutional reform. In the terms of the Statecraft framework: the work shifts between blue interventions (rational, analytical) and yellow interventions (power spreading, institutional design), and refuses the monochrome blue in which government advisers often get stuck. It cannot therefore be dismissed as mere criticism.

The third is international standing. Government Information Quarterly and Public Administration Review are the two most influential journals in their field worldwide. Publications there are not only academic credentials, they ensure that the Dutch debate connects with international theory-building. The concept of infrastructure-level bureaucracy is the best example: it was formed in the Netherlands on the basis of Dutch casework, and is now being picked up internationally as an explanatory model for problems in other systems. That the Kafkabrigade has a sister organisation in Mexico, and that the work is followed in countries such as Ireland, Belgium and Germany, indicates that the relevance reaches further than the Low Countries.

A blind spot in the Statecraft domain

For Statecraft readers the work of Widlak and the Kafkabrigade is indispensable for two reasons. The first is that it fills a blind spot which, in almost every interim assignment, at some point turns out to be the actual cause of the problem. When youth care gets stuck, the blockage often sits not with the team, not with the policy, but with the registration system that cannot distinguish what the professional sees from what the funding code allows. When social-care execution complains about administrative load, it is rarely about the application itself and almost always about the chain of organisations that ask, store and verify the same data. An interim manager who does not see this layer repairs symptoms.

The second reason lies in the connection with the Aiki method. Aiki distinguishes itself from forcing by redirecting energy towards outcomes that serve the collective interest. Infrastructure-level bureaucracy has the property that it embodies forcing design without any individual exercising the force explicitly. No one compels; everyone follows. That makes it vulnerable to interventions that consist not of confrontation but of redesign. A civil servant who refuses within the existing architecture loses her job. A designer who can re-route the architecture changes the outcome without conflict needing to arise anywhere. That is, in its most literal sense, an Aiki intervention at system level. It is no coincidence that the most productive thinkers about digital rule of law in the Netherlands work precisely there.

Conclusion

Dutch public administration is good at detecting symptoms and bad at naming causes. That pattern is known and is developed elsewhere in this Statecraft publication series. But there is no domain in which that distance is so literal, and so concretely felt by citizens, as in digital execution. The Kafkabrigade names the cause. They do so with empirical precision, conceptual weight, and constructive proposals. And they have done so long before the political domain was willing to listen.

Whoever wishes seriously to reform youth care, social assistance, building permits, the healthcare allowance or filing returns, and stays at the level of the organisation without looking at the architecture, does work that within a few years will be undone by the same infrastructure that produced the original problem. Whoever does look at the architecture finds in the work of Widlak and his colleagues the most thoughtful instrumentation available in Dutch. It is, as far as Statecraft is concerned, required reading. Not because it is comfortable, but because it is true.

---

Disclosure. The author chairs the Supervisory Board of the Kafkabrigade Foundation. This publication has been prepared on his own initiative, without commission or financial contribution from the foundation, and reflects the substantive position of Statecraft.

Sources. Widlak, A. & Peeters, R. (2018), De Digitale Kooi, Boom Bestuurskunde. Widlak, A. (2022), Volwassen Digitale Overheid, Boom Bestuurskunde. Peeters, R. & Widlak, A. (2023), Public Administration Review. Widlak, A. & Peeters, R. (2025), A theory of the infrastructure-level bureaucracy, Government Information Quarterly 42(2). Think tank Back of Government (2026), Nieuwland: a design for the digital rule of law. Council of State, unsolicited advice on digitalisation and the legislature (2018). Robben, L., Peeters, R. & Widlak, A. (2024), Burdens on the gateway to the state, Journal of Policy Analysis and Management.