Gothenburg handed school placements to an algorithm and got chaos with no one accountable

In Gothenburg, Sweden, an algorithm distributed school places for the first time, and it was supposed to make the system fairer and more efficient. Instead, it turned admissions into an opaque process in which families faced chaos and the city faced a question: who is responsible for a decision made by code?

How the algorithm came about

In 2020, city authorities decided to automate one of the most painful municipal tasks: assigning children to schools. The logic looked convincing. A machine finds it easier than officials to weigh the distance to school, families' preferences, catchment zones and available places all at once. The approach was sold as a technical upgrade, not a political decision.

The algorithm was presented almost as a neutral tool that simply calculates options faster and more accurately. That was precisely the main temptation. When a decision is packaged in the language of optimization, it seems free of human bias. But school admission is not only math. Behind every row in a spreadsheet are families, routes, daily logistics, a sense of safety and trust in the system. When such a mechanism fails, the problem no longer looks like an ordinary administrative error. It becomes a blow to people who can neither check how the calculation worked nor understand why their child ended up somewhere other than expected.

Where the problems started

The promised objectivity quickly collided with reality. Instead of a clear procedure, parents received decisions whose logic nobody could properly explain to them. For some families this meant a longer trip to school; for others, the collapse of plans made well in advance. When people tried to find out why the system had worked the way it did, they ran into the wall familiar from digital bureaucracy: there is a decision, but seemingly no one responsible for it.

The consequences of this automation showed up at several points at once:

• parents did not understand the allocation criteria;
• officials pointed to the system rather than to their own decision;
• technical complexity made it hard to challenge the result on the merits;
• an error, if one occurred, scaled instantly to many families;
• the algorithm's neutrality turned out to be more a promise than a proven fact.

The most painful part of this story is not the failure itself but the impossibility of getting clarity. When the author of the original text tried to challenge the decision in court, it became clear how poorly law and public administration are adapted to disputes with automated systems. Formally, a decision seems to have been made and a procedure seems to exist, but in practice a person is arguing not with a specific official but with logic hidden inside code and an administrative process. In such a dispute, a citizen almost always starts from a weak position.

Why the algorithm won

The Gothenburg case shows a broader problem: digital systems in the public sector are often introduced under the banner of efficiency without building a comparable system of accountability. When a person makes a decision, you can ask about their motives, their mistake, an exception to the rules. When an algorithm makes it, the chain of responsibility breaks apart: developers say they merely implemented the requirements; officials say they trusted the model; lawyers say the rules were formally followed. In the end, the person harmed is left alone with a system that has no face.

That is why the algorithm "wins" in the public eye. Not because it is necessarily right, but because it is hard to open up, verify and question in plain language. Technical opacity starts to function as a form of power. Code built to simplify a process becomes a barrier between citizen and state. And the more loudly authorities call such a system objective, the harder it is to admit that it may contain disputable assumptions, flawed settings or simply bad management decisions.

What this means

The Gothenburg school allocation story is a warning for anyone automating socially sensitive decisions. An algorithm can speed up a process, but it cannot replace explainability, the right to appeal and living accountability. If these elements are not built in from the start, any "efficiency" easily turns into digital injustice.