There are two answers. First we can steelman the idea and imagine that, given millions/billions of debtors we could aggregate enough data to notice trends and t

There are two answers. First we can steelman the idea and imagine that, given millions/billions of debtors we could aggregate enough data to notice trends and then be confident in relying on those trends without understanding causality. More dangerously, we could imagine that these data would be used to allocate more favorable lending terms to those individuals who spent their time browsing the "correct" websites.