Elon Musk Cited Race Crime Data

I found this online erroneous comment:

They'll teach right-wing tribalism exclusively, and reflect Elon's Darwinist philosophy that you have to be useful to be society or perish. He's been peddling false narratives against Africans pretty hard. Elon Musk claims we have to decide between wanting to evolve and not wanting to be evil. ...

For example, Elon Musk recently posted that blacks are 37 times more likely to kill a white person than a white person is likely to kill a black person. That statistic is designed to make you think we'll have a Utopia if blacks are put into concentration camps. However it ignores the fact that there are 7.5 times more [whites than blacks] thab in the US. Therefore that 37 has to be divided by 7.5. And then when you adjust for socioeconomic factors the remaining 5x discrepancy goes away.

No, this is wrong. Another commenter explains:

Your politics are admirable comrade, but your math is poor. If murders were random, it would be like taking two random marbles from a bag with 100 black marbles, and 500 white marbles. In the cases where you chose one of each, it would be 50/50 black then white, or vice versa.

Same thing with sex. Men are far more likely to kill. Victims are mostly men too, but women still far more likely to be killed by a man than vice versa.
Can you accept that as fact? Then why not for race? People are not all the same, that would be very boring.

After some pushback, another explains:

Let's assume the killings are random and everyone has equal capacity for murder.

The probability of killer being black is 2/17. The probability that he kills white person in random killing is 15/16. Hence the probability of killer being black and victim being white is (2/17)*(15/16)=11%

The probability of killer being white is 15/17. The probability that he kills black person in random killing is 2/16. Hence the probability of killer being white and victim being black is (15/17)*(2/16)=11% - the same

If you are seeing discrepancy between black-on-white and white-on-black rates then our assumption that murders are random and everyone has the same capacity for murder does not hold.

That's right. If race did not matter, the incidence of Blacks killing Whites would be the same as Whites killing Blacks. It is not. Not even close.

THere is a symmetry here. Unless Blacks and Whites have some behavioral differences, we expect the incidence of Blacks killing Whites to be the same as Blacks being killed by Whites. Mathematicians like to translate these problems into taking marbles from an urn, but then "murders" is the same as "is murdered".