Better matches versus honest choices 

Systems that allocate school places are an example of one-sided matching: people list their preferences, and an algorithm assigns them to the available places. However, these systems face a difficult trade-off. Some systems are designed to make truthful reporting the best option. In other words, pupils and parents have no reason to change their preference list to improve their chances. This is known as strategyproofness. However, these systems can sometimes leave pupils with a place at a school far down their list. 

Other systems can give more pupils a place at one of their preferred schools overall. But they may also make it worthwhile to report preferences strategically: ranking schools differently from one’s genuine preferences in the hope of securing a better outcome. 

Amsterdam’s secondary-school system illustrates this challenge. The city uses a lottery-based method called Random Serial Dictatorship, or RSD. It encourages truthful reporting, but some pupils have been assigned to schools much lower on their lists.

Rules can create unequal advantages 

To examine what is gained and what new challenges arise when a system gives more weight to pupils’ preferences, Tasnim studied rank-minimising matching. Unlike Amsterdam’s current RSD, the lottery-based system, this method aims to give more pupils a place at a school high on their preference list. Her research shows that it can improve outcomes overall, but it also creates a new risk. Applicants may be able to improve their own outcome by changing their preference list, using only information about which schools are popular as other pupils’ first choice. This means that families with more information or confidence may benefit more than those who simply report their genuine preferences. 

Strategic reporting can also be encouraged by additional policies. In Amsterdam, a placement guarantee was introduced to help pupils who did not receive a place at one of their preferred schools. Tasnim’s analysis found that this policy can also give pupils and parents a reason to adapt their preference lists strategically. If many families respond in this way, the guarantee may no longer be able to deliver on what it promises. 

Clear information can make a difference 

Having an incentive to act strategically does not mean that everyone will do so. Tasnim surveyed 140 Amsterdam parents about their school-choice decisions. The study found that explaining the risks of strategic reporting can discourage people from using it. 

These findings show that clear communication is part of designing a fair system. Policymakers need to consider not only what an algorithm does, but also how people understand the choices it creates. 

AI for public decisions 

Tasnim also examined whether AI methods that learn from examples, including Deep Q-Networks and Generative Flow Networks, could offer another way to allocate school places. Her research found that these methods can sometimes make better matches while remaining more resistant to strategic reporting than established matching methods. She stresses, however, that a system should not be judged only by its technical results. It is equally important to consider how people may respond to it and how it works within the public institution that uses it. 

Tasnim’s research does not point to one simple replacement for Amsterdam’s current system. Instead, it shows that proposed changes should be tested more broadly before they are introduced. Policymakers should consider not only whether a new rule gives more pupils a preferred school, but also whether it creates advantages for families with more information, what happens if many families change their behaviour, and whether the policy can still deliver what it promises. 

More information

Read Mayesha Tasnim’s thesis , One-Sided Matching Under Strategic Reporting: Insights from Amsterdam School Choice in UvA Dare