The cutoff probably varies per city, but I imagine they're only trying to get rid of the lowest percentiles of the distribution. Honestly though, a 4.7/5 gener

The cutoff probably varies per city, but I imagine they're only trying to get rid of the lowest percentiles of the distribution. Honestly though, a 4.7/5 generally sounds good. If a 4.7/5 is near the (very) bottom of the distribution then it sounds like the rating system is quite flawed and should suggest to Uber that they need to replace it (perhaps with just a "would you use this driver again" question, and if no, then whether the cause was behavioural (e.g. odour or some form of rudeness) or dangerous driving), then rank drivers using this. Vice versa for riders. If the /5 rating system was accurate, then 'good' drivers/riders should be ~3.5-4, excellent drivers 4.2+, and poor drivers 1-2. But since some people have different definitions of 5/5 compared to others the rating system ends up generally turning into a 5/5 for above satisfactory with no real issues, and random usage of the rest of the scale.