First, using an arbitrary number of stars is part of the problem. Too much opportunity for legitimate differences of rating to come into play. Just make the question "would you do business with this driver/rider again?" and leave it at that. At the very least the rating should be market-specific, since there is so much cultural influence over such ratings.
Second, Uber should not be making the choice based on some arbitrarily chosen numerical value. Present the rating to the drivers, maybe with some comments, let them choose who they want to take. After all, if Uber isn't employing the drivers and is just the middleman, wouldn't they want to avoid being seen as a crucial decision maker?
Second, Uber should not be making the choice based on some arbitrarily chosen numerical value. Present the rating to the drivers, maybe with some comments, let them choose who they want to take. After all, if Uber isn't employing the drivers and is just the middleman, wouldn't they want to avoid being seen as a crucial decision maker?