The other thing to note is that a lot more black Americans than white Americans are victims and perpetrators of homicide.
If killings happened with current perpetrator counts, but were hitting victims randomly across the black and white US populations, then we'd expect to see a
lot more white victims of black murderers than is actually the case.
This is the FBI counts for 2013, this data will correspond to the graph above's data points for 2013:
If we, for simplicity, ignore other racial groups and distribute just the count of black/white victims of black/white killers randomly by population shares, we get this:
Most murders occur within races because of the degree of community sorting that occurs in the US. Most people kill people they know.
But the rate at which white folk are killing different racial groups is actually a lot closer to what we'd expect from a fully random distribution of victims, than the rate at which black folk are killing different racial groups. This distribution is all almost certainly reflective of degrees of community contact and interaction and also of the dysfunction of particular communities, rather than having a directly racial explanation.
I mean, it's not like this data will include killings by cops (the actual core of BLM grievances), since it's about unlawful homicides and very few killings by police get treated that way.