That doesn't add up though. If you guess that black people are more likely to have guns and you search them far more often, more evidence will turn up, even if they are/were as likely to have a gun than any other group.
If you never search white people for drugs, guess what? You never find drugs on white people.
Yes that's correct. However, in that case, while your total number of black drug crimes will be high, your "drug crimes per interaction" rate will still tell the true story.
(Assuming no straight up fraud/planting drugs on the part of the police, which is also very possible.)
The "drug crimes per interaction" rate doesn't work for this case either, if the number of interactions is just too low. And if you look at real numbers, in the US, for example, we're talking a difference in the order of two magnitudes.
So if any person has a 10% chance to carry an illegal drug and you search 100 black dudes and 1 white dude, chances are your drug crimes per interaction will be 0 with the white dude and about 10% with the black dudes.
Now add to that, that the white dude is much more likely to get off with a warning that's not committed to protocol and there we are.
If you never search white people for drugs, guess what? You never find drugs on white people.
It's a self-fulfilling racist prophecy.