Maybe people with brown eyes are more than likely to commit a violent crime than someone who has hazel eyes?
OK, a little hard to parse what the counter argument is here, but I'll try. I believe it is a restatement of the old contention that 'color is just skin deep', that we assign race based solely on superficial trivial aesthetic characteristics, and we could just as rationally do it based on other traits like eye color or as Jared Diamond said, ear wax texture. But this is sophistry. Race being referred to as a color of skin is merely a useful shorthand, choosing one of a large group of characteristics by which populations differ. The main idea is that populations, long separated by geographic and cultural barriers, become separate breeding pools and diverge, a process that eventually leads to separate populations.
In other words, a person is 'white' not because their skin is literally white, or even because it is pale, but because they come from a shared ancestry and shared genetic pool of which one aspect is the genes for pales skin that easily absorbs vitamin D in low sunlight.
The other argument appears to be that because the racial membership was not the immediate cause of any individual committing a crime (because others sharing that race did not commit the crime), it should not be viewed as a risk factor. I think this confuses risk management, which inherently involves prejudice of some sort (one must prejudge before a crime) and assigning guilt (culpability after crime).
So no, all black people are not guilty for the crimes of some black people. But, think of it this way. Men commit about 10x more violent crime than women over all (about the same ratio as black men to white men). No one man is guilty of the crime of any other one man. But it is rational of women or men to manage risk by being more wary of men in general.... and therefore also to be even more wary of black men in general