You know, when you start with a foregone conclusion that everyone and everything is racist, you end up with the same foregone conclusion. But it might not be the correct conclusion.
First of all, I and the sources I’m drawing on rely on data about those guilty of crimes (whether through pleas or convictions), not merely those arrested, so it shouldn’t matter (for this purpose) whether blacks are arrested and released at a higher rate than others, as you say. But let’s assume that, to some extent, black people are probably overrepresented even among unjustified pleas/convictions (a natural consequence of the fact that black people are also overrepresented among those committing crimes). The question is: to what extent do you really think innocent black men are being convicted of crimes they didn’t commit? This undoubtedly happens, and it happens with white people as well, but what’s your guess about what percentage of black people accused of crimes are actually innocent?
Most criminologists estimate the wrongful conviction rate as a whole at about 0.5%-1% of all convictions, while some suggest a number closer to 2%–3%. (Here and here are two arguments in support of the higher number, for instance.) I don’t have data on how many of those wrongful convictions are specifically of black people, but since blacks represent 28% of criminal convictions, obviously whites, who are 69% of criminal convictions, are going to also be the victims of mistakes more often than blacks, though blacks could still be overrepresented within the rate of errors. According to the second article I linked to, “[t]he main cause [of wrongful convictions] in more than half of the cases — 52.3 percent — was eyewitness misidentification,” and it’s pretty unlikely, of course, that someone is going to mistake a white man for a black man. Most people who are victims of crimes actually really want vengeance against the person who did it, not against someone else, so if they misidentify the criminal, the mistake is most likely going to be a mistake that stays within the same racial category.
When you put all this together, my point is that the rate of wrongful convictions of blacks for crimes committed by whites or someone of some other race is a very, very small number when compared against the overall disparity in the rate of criminal acts by blacks vs. whites, so that using that overall rate of crimes committed in making the comparison against rates of people of various races killed in the course of encounters with police is still a very valid measure and certainly, far, far more valid than just doing a strict per capita comparison of killings by police by race vs. percentage of persons of that race in the general population.
Which brings me to my larger point: being anti-racist is all well and good, but when you’re careless in throwing around assumptions of racism left and right, you actually wind up ignoring the empirical data, endangering the lives of police officers through irresponsible accusations, blackballing people who don’t deserve to be blackballed and, all in all, doing more harm than good.