How a naif might think the distribution looks based on the perpetual cherry picking and gross omissions served up by the major media:
Piggybacking on the 2012-2013 NCVS numbers crunched by Heather Mac Donald, how the perpetrator distribution against black victims of non-homicidal violent crime actually looks:
A picture is worth a thousand words, and this is a handy one to pull out of a back pocket every time a SWPL or race hustler yammers on about how some individual case of white-on-black violence serves as a microcosm of American society as a whole (while ignoring the 27 instances of black-on-white violence that occur for each single case of white-on-black violence).
Just 1-in-10 black victims suffer at the hands of non-Hispanic white perpetrators of violence. This even though whites comprise a majority of the population.
Proximity, of course, plays a substantial role in shaping the distribution. The Obama administration's drive to push NAMs out of the inner cities and into white suburbs will, as one of its many aftereffects, lead to an increase in interracial violence. As a consequence, a few more blacks will suffer at the hands of whites and a lot more whites will suffer at the hands of blacks. Diversity + Proximity = War.
Still, if #blacklivesmatter, it seems curiously ineffective and inefficient to focus almost exclusively on a small minority of cases in which blacks suffer if the objective is to reduce said suffering.
Whites are, in fact, less likely to perpetrate acts of violence against any of the groups measured--whites, blacks, Hispanics, or others--than their numbers alone would predict under the (flawed) assumption that propensity for criminal violence is distributed evenly across racial groups.
Blacks, on the other hand, at 13.2% of the population, are more likely to perpetrate acts of violence against members of all other groups than would be predicted under an egalitarian assumption based on their share of the population alone.