We like to imagine that injustice arrives with a bang - maybe a dramatic siren, a stern letter, or at least a mildly concerned tweet. But in Gothenburg in 2020, it arrived as a sleek algorithm promising efficiency. The city, tired of the administrative headache of school placements, handed the job to a machine. After all, what could possibly go wrong when you let software decide where hundreds of children should spend their formative years?
Plenty, as it turned out. The algorithm optimised distances, preferences and capacity, but apparently forgot that Gothenburg has a river - a major one, with fjords and highways for good measure. Instead of calculating walking routes, it used “as the crow flies” distances. Parents stared at their children’s placements in disbelief: schools across rivers, over major highways, in neighbourhoods the kids had never visited. A 13-year-old walking that route in winter? Not a chance. The schools administration offered calm, unhelpful advice: just appeal if you have an issue with your placement - as if the problem were a matter of personal taste and not a systemic malfunction.
Enter Charlotta Kronblad, a researcher in technology, a former lawyer, and the mother of one of the 700 affected children. She didn’t just appeal her son’s placement; she sued the city, arguing the entire decision-making system was unlawful. The city’s defence? The algorithm was merely a “support tool,” and they provided no code, no documentation, no explanation. The court placed the burden of proof squarely on Kronblad. Without access to the algorithm’s black box, she couldn’t meet the evidentiary threshold. The case was dismissed. The algorithm won.
This isn’t an isolated Swedish oddity. Kronblad draws painful parallels to the UK Post Office scandal, where the Horizon IT system falsely accused hundreds of operators of theft, and the Dutch childcare benefits scandal, where an algorithm wrongly flagged thousands of parents as fraudsters, plunging families into debt and foster care. In all cases, the machines operated behind a veil of technical complexity and institutional defensiveness. Errors multiplied. Harm deepened. Accountability lagged.
Kronblad’s conclusion is stark: when courts defer to technology rather than interrogate it, and when those harmed must prove the system’s flaws without access to the code, algorithmic injustice thrives. She calls for shifting the burden of proof to the parties that actually have access to the algorithm, and for procedural rules that allow systematic redress. Until then, she warns, we’ll stumble from scandal to scandal - because when injustice is delivered by code in near silence, accountability must answer at full volume.