> would you get into an Uber if you knew the driver had 5 previous DUIs? Well they somehow convinced the state they should have a driver's license still. If th

> would you get into an Uber if you knew the driver had 5 previous DUIs? Well they somehow convinced the state they should have a driver's license still. If they're that dangerous they shouldn't be on the road. Further, simply having 5 DUIs on record is itself not a sufficient data point to determine if there is any increased risk today of that person doing so again. For instance, if a person turned to drinking out of grief at one point in their life but has had many stable years since, how would you identify that risk against the risk of someone who hasn't yet lived through such a situation but were it to occur they would drink to excess and make poor decisions too? NOTE: I'm not saying those DUIs aren't a problem, I'm just pointing out that it can be more complicated than it first appears to reach an internally consistent, much less fair, approach.