Surely however that sort of defense against specific manifestations of bigotry can be evidence (assuming proof is practically unobtainable). For instance, if I were accused of hiring practices that are biased against black people, pointing out that I have only hired black people would be evidence (if not proof, in this hypothetical and silly black and white example...) that the accusations are false.
> For instance, if I were accused of hiring practices that are biased against black people, pointing out that I have only hired black people would be evidence [...] that the accusations are false.
Without information about the external constraints on your hiring, I'm not sure it would really constitute meaningful evidence (e.g., if the entire available employment pool was black, hiring only blacks wouldn't be evidence that the process you had implemented wasn't biased, just that the bias has no opportunity to manifest.)
(This is the kind of thing that would work mostly to burden shift, but then the burden would already, IMO, be on the person making the accusation from the start.)
OTOH, it would be very challenging (though not impossible) for anyone to have evidence that your system was biased against blacks with a hired-all-blacks outcome.
I suppose my hangup here is what I perceive to be the difference between evidence and proof.
It is entirely possible to find evidence of things that turn out to be false. For example, you could take the seemingly chaotic retrograde motion of the planets in the night sky as evidence of the existence of gods with self-determination. We know know that isn't true, we have better explanations for that phenomenon (notably the notion of heliocentricity).
The planets moving differently from the stars was never proof that there were gods up there, but evidence? Sure, it was evidence. Evidence can be of various qualities, evidence of thins has to be judged critically, and weighed against the alternatives.
Evidence for or against bias on a particular axis is something that tends to indicate the presence or absence of differential treatment basis on that axis.
Facts which do not provide a basis for comparisons of how different values on that axis are treated are not much evidence for or against bias, this include facts about outcome (e.g., all the people I hired are black, etc.) without additional information from which the expected outcome in the absence of bias can be inferred.
