In a roundabout way, this is related to the big data world we live in, even though background checks & credit checks are not new. From the perspective on a mot

In a roundabout way, this is related to the big data world we live in, even though background checks & credit checks are not new. From the perspective on a motor insurance company, for example, they don't need to serve everyone. If they can classify the world of potential drivers as high, medium & low risk, that's enough. Whether or not an individual falls into one of these categories "incorrectly" is not important, they're average figures and as long as they are "correct" for the people who take out a policy all is well. For an individual that ends up in the "high risk" category, it's inequitable and unfair. They might share characteristics with "truly" high risk individuals but not in ways that actually affect their chances of causing on an accident they just happened to be in a statistical cluster. The BMI is an interesting example too. Weight bands for height works fine when looking at a population. You can fall into an obese weight/height band by being heavily muscled or broad bodied. Mike Tyson at his youngest, slimmest prime would have been classed as obese and very obese later in his career. Possibly "extremely obese" between matches. This is a good example because you can easily se how BMI works well enough for evaluating insurance risk but not for being "fair" for our puny human defections of fairness. We are moving to a world of more data and more data based decision making. Online advertising for example, has transformed a big industry to one that is mostly driven by data based decision making. The side effects of this will be many. There are certainly efficiency gains because that statistics are predictive. Having been arrested or serving time is probably predictive of risk, even if a conviction is overturned. There are also feedback loops that we don't want, biases and "traps." A homeless person is obviously high risk in that he shares measurable characteristics with other homeless people and they (almost by definition) actualize those risks. There are real society-affecting risks here. It could become more expensive to be poor, as credit and financial services becomes a bigger part of how we purchase things.