How could government prevent the next benefits scandal? ‘Be vulnerable and share your data’
Government decision-making is unfathomable and hard to check. Many decisions are not made public and their layout makes it difficult for computers to analyse them. Professor Anne Meuwese is therefore calling on governments to show more vulnerability when it comes to providing information about how they function.
‘The crux of the problem is that a lot of data that plays a role in government decision-making is sensitive’, says Meuwese, Professor of Law and Governance of Artificial Intelligence at Leiden University’s Department of Constitutional and Administrative Law. ‘Take administrative decisions, for instance, government decisions directed at a person or a business. It’s no surprise that they’re not public. You don’t want your tax return or the size of your DUO loan to be publicly available. But there are ways to anonymise the information in the decisions and share it with researchers. The government could be more assertive about this.’
Synthetic data sets
Meuwese believes the government could, for instance, use synthetic data sets: data sets generated from real data sets with the sensitive data removed. ‘This kind of data set is already used in research and with computational textual analysis such as Natural Language Processing (NLP) can give a great deal of insight into how an authority works’, says Meuwese. ‘Yet it is barely used at all in empirical legal research into government decision-making. That is because we are used to looking mainly at court decisions, so at the result of that decision-making and the ensuing problems. But with the child benefits scandal and the discriminatory fraud checks by DUO that have recently come to light, it has become clear that we need much broader access to government decision-making so we can identify such problems sooner. And then a synthetic data set is ideal.’
Personalised digital environment for citizens
Meuwese emphasises that not only researchers but also citizens should gain more access to data that plays a role in government decision-making. ‘There’s currently a huge information disparity between the government and its citizens. In fact, as a citizen you practically only have access to the decision you have received. The decision-making behind it is completely inaccessible when you as a citizen want to know why, for instance, you are not entitled to benefits and what algorithms and criteria were used to reach this decision.’
Meuwese thinks a personalised digital environment might be a solution to this problem. ‘Citizens would be able to access information that concerns them personally. They would then be able to have more control over the data used for decision-making.’
Data budget
Ideally, the government itself should also take more responsibility for the data its decision-making is based on, says Meuwese. Not just by using more advanced methods to analyse this but also by reporting the results. ‘We agreed at some point that financial data would be the backbone of all sorts of processes and that that you need to be accountable for this so you can check whether the organisation is running well. We should also do the same for data and should report what exactly has been used along with how and when. We should also report on what isn’t known. This information obviously makes a government vulnerable it’s needed to prevent big systemic failures in the future, such as the child benefits scandal.’
Text: Sabine Waasdorp