m4chth1s
Why? Servers claim that black customers - on average - tip less.
I don't know how prevalent this belief on the part of servers is. And I don't know how much of a numerically accurate basis in reality it has.
But, as a thought experiment, imagine for a moment that there exists an identifiable group [black people, star trek fans, old people, dentists, ...] that are likely to lead a server to go home with less money at the end of the night.
There can be a really fine line between statistics/machine learning and "bias"/"racism".
Where it gets even more complicated is that there might be variables which are not expressed as "is the customer black" which ultimately predict/are highly correlated with the question "is the customer black" - but are also predictive of the question "is the customer profitable".