Four Dimensions for the State

In the first post of this series, we explained why public administration is the ideal candidate for self-driving: it is rule-based, procedural, and bound to document everything. But that only establishes that it can be done, not where it should lead. In our book "Der selbstfahrende Staat," which Springer Gabler publishes this year (in German), we therefore transfer the four dimensions of the self-driving company to the state: autonomy, sustainability, resilience, and a human focus. This post shows what each of these dimensions means in the public sector and why none of them stands on its own.

Where the four dimensions come from

The picture originates in "The Self-Driving Company" (Springer 2023; German original "Das selbstfahrende Unternehmen", Springer Gabler 2021). There I built the company of the year 2035 on four fundamental dimensions that form the backbone of every sustainable and resilient organization. Autonomy is the core competence: much of the decision-making is delegated to software and artificial intelligence. The other three dimensions are the conditions under which that autonomy is accepted and keeps working in the long run.

In the book on the self-driving state we deliberately return to this model because self-driving companies and organizations run ahead of the state. By around 2035 they will demonstrate how autonomy, sustainability, resilience, and a human focus interact in a digitally transformed world. The state can learn from them, but it cannot simply copy them. What is a competitive advantage in a company is a question of fundamental rights, separation of powers, and democratic legitimacy in the state. That is why the weights shift as soon as the four dimensions are applied to public administration.

The four dimensions in the public sector

Autonomy: the software administers, politics shapes

The stages on the way to the self-driving state correspond to those of the self-driving company and are modeled on the autonomy levels of self-driving vehicles: analog, digital, automated, and self-driving. We speak of a self-driving state when around 80 percent of the decisions in public processes are made autonomously, that is, by learning software systems without human intervention. We deliberately set this threshold in the book. The annual tax assessment that is delivered automatically, together with the instruction to refund the tax credit, is the book's example for the stage just before, the automated state.

What matters is the boundary of that autonomy. It is not politics but exclusively the administrative processes that are digitized, automated, and finally made self-driving. Decision rules continue to be made by elected representatives; the software executes them. Politics provides funding, infrastructure, and the legal framework, monitors implementation through key figures, evaluations, and reporting duties, and adjusts where necessary. And wherever a decision has significant consequences for the lives of individuals or for society as a whole, people in politics, legislation, and the judiciary remain the final authority. Autonomy in the state therefore means administering rather than shaping. That is not a limitation of the model; it is its core.

Sustainability: from reacting to anticipating

In a company, sustainability mainly means anticipating the long-term consequences of decisions and reducing the ecological footprint. For the state, a second layer is added: it is itself the largest bearer of long-term responsibilities, from infrastructure to the social system. Precise data analysis identifies inefficient processes and lowers the administration's consumption of resources. More importantly, the self-driving state can predict developments and trends and act preventively before problems turn into crises.

In the book we describe this for families, for example: if data from the education register is linked with social data and information on housing, problem situations can be recognized in time and support can be distributed in a targeted way instead of waiting for someone to file an application. For legislative proposals, simulations make it possible to estimate the effects and side effects of a measure before it is adopted. Sustainability in the public sector is therefore less a question of footprint than a question of whether the state switches from a reactive mode to an anticipatory one. The precondition always remains the same: the self-driving state acts as a neutral tool without a political agenda of its own and supports the overarching goals of the state or of a confederation of states such as the EU.

Resilience: staying capable of acting when it counts

Resilience is gaining importance in light of an unpredictable global situation, and that applies equally to companies and states. In the book we describe an example that many administrations still remember vividly: the aid for businesses that had to close during the coronavirus pandemic. In the analog state, the aid was defined by regulation, applied for on paper, checked manually, and paid out by hand. The digital state brought online forms and status portals. In the automated state, the entire process from application to payment runs on rules. The self-driving state goes one step further: it determines the individual need for support from the existing data pools, takes assets, income, and benefits already received into account, and grants the aid without any application. That is resilience in the sense of citizens: the state remains capable of acting while its counters are closed.

This includes a change of roles that we describe using tax calculation as an example: today, even the smallest company has to work through the applicable legislation itself. In the self-driving state, the state provides an open-source tax calculation core as an interface that is continuously updated in line with current legislation. Especially in a crisis, when rules change quickly, responsibility for keeping them up to date thereby moves to the state.

From our project practice I add a second reading: a state that leaves 80 percent of its decisions to software must also be resilient against the failure and manipulation of that software. In the book we name complete logging and historization of all decisions, encryption to high standards, and mutual review by the legislative, executive, and judicial branches. Any manipulation of the logging would immediately become visible through the historization, and vice versa. Resilience is therefore not a feature that can be added later; it is an architectural decision.

A human focus: people as the yardstick and the corrective

The fourth dimension places people at the center. I consider the fear of being controlled by machines unfounded, because technologies are instruments that are meant to reflect our values. In the state, however, the human focus is more demanding than in a company, because public administration cannot lose anyone as a customer and must not lose anyone.

Two premises of the book make this dimension concrete. The first is the avoidance of administrative discrimination. Practices have crept into traditional bureaucracy that arise from understandable human behavior: heavily loaded officials prefer simple cases, applicants with good language skills receive more attention, and complex cases involving uncertainty are delayed or passed on. Intelligent software treats everyone equally, but only if bias is avoided in programming and training and if the legislative and judicial branches continuously review data and algorithms. The separation of powers thus becomes a control loop in which each branch checks the others.

The second premise reads: the transparent state prevents the surveillance state. It is not the citizens who become transparent; the administration discloses its own processes, decisions, and data while protecting privacy at the same time. Anyone who can trace how a decision came about has no reason to fear covert surveillance. The AI regulation adopted by the European Parliament in March 2024 points in the same direction by requiring human oversight, documentation, and traceability for many applications in public administration.

Finally, the human focus also concerns the people working in administration. When software takes over routine work, documentation, and control, leadership shifts from the "3 Cs" (command, control, correct) to three different priorities: support, challenge, and give feedback (the "3 Fs" of the German original: Fördern, Fordern, Feedback geben). Case workers gain time for the cases that require empathy and discretion. Psychologists and ethicists take on a larger role, because someone has to ensure that the state's systems act for the good of people.

Why no dimension stands on its own

The four dimensions are not a ranking. They are a balance, and unlike a company, the state cannot pick and choose among them. Whoever relies on autonomy alone gets efficiency without acceptance. Whoever refuses autonomy out of caution stops at the digital form and keeps paying the cost of bureaucracy. The following overview shows how the guiding question of each dimension shifts when moving from the company to the state.

In the book we tested every statement against five premises: a focus on the human and the sustainable, technological progress frees us from bureaucracy, the self-driving state administers and politics shapes, the avoidance of administrative discrimination, and the transparent state prevents the surveillance state. Reading these premises alongside the four dimensions makes the relationship visible: autonomy is the means, the other three dimensions are the purpose.

A company may pursue autonomy for its own sake. A state may not.

I deliberately put this more sharply in the public sector than in the book on companies. A company that automates too far loses customers and corrects itself through the market. A state that automates too far loses trust, and there is no market that brings it back.

Four questions for every digitalization project

The dimensions yield a simple test that we already apply in public-sector projects today, long before an organization reaches the self-driving stage:

1. Autonomy. Which decisions should the software make, which should it only prepare, and where is a final human decision mandatory? That boundary belongs in the specification, not in operations.
2. Sustainability. What data does the administration need to act in an anticipatory rather than reactive way, and which data silos stand in the way?
3. Resilience. How does the procedure work in a crisis, when the system fails, or when rules change quickly? Who keeps the rules up to date?
4. Human focus. How do we recognize administrative discrimination, how do decisions become traceable, and how much time do the people in the administration gain for people?

Whoever answers these four questions for a specific procedure has broken the vision of the self-driving state down into a project that can be commissioned, specified, and accepted.

The next step

The four dimensions are the coordinate system in which the further posts of this series move; the starting point is the post The self-driving state: why public administration is the ideal candidate. We present the complete model with the five premises and the service catalog on our book page: About the book. How we support administrations in specifying and implementing procedures along these dimensions is described on our page on public administration.