Unless you know what exactly caused A to outperform others, you won't really know if the process is biased or what made it biased.

When asserting biases, you must first distinguish them from random noise. Using pg's logic, every selection process that isn't perfect is biased.