Let's start with what we don't know: the precise circumstances under which a white police officer in Ferguson, Mo., shot dead an unarmed black teenager named Michael Brown.
But here's what evidence does strongly suggest: Young black men in America suffer from widespread racism and stereotyping, by all society -- including blacks themselves.
Research in the past couple of decades suggests that the problem is not so much overt racists. Rather, the larger problem is a broad swath of people who consider themselves enlightened, who intellectually believe in racial equality, who deplore discrimination, yet who harbor unconscious attitudes that result in discriminatory policies and behavior.
Scholars have found that blacks and Hispanics treated by doctors for a broken leg received pain medication significantly less often than white patients with the same injury. School administrators suspend black students at more than three times the rate of white students. Police arrest blacks at 3.7 times the rate of whites for marijuana possession, even though surveys find that both use marijuana at roughly similar rates.
Two scholars sent out nearly 5,000 resumes in response to help-wanted ads, randomly alternating between stereotypical white-sounding names and black-sounding names. They found that it took 50 percent more mailings to get a callback for a black name. A white name yielded as much benefit as eight years of experience, according to the study, published by the National Bureau of Economic Research.
These doctors, principals, prosecutors and recruiters probably believe in equality and are unaware that they are discriminating. So any national conversation about race must be a vivisection of challenges far broader and deeper than we might like to think.
Joshua Correll of the University of Colorado at Boulder has used an online shooter video game to try to measure these unconscious attitudes (you can play the game yourself at http://www.csun.edu/~dma/FPST/study.php). The player takes on the role of a police officer who is confronted with a series of images of white or black men variously holding guns or innocent objects such as wallets or cellphones. The aim is to shoot anyone with a gun while holstering your weapon in other cases.
Ordinary players (often university undergraduates) routinely shoot more quickly at black men than at white men, and are more likely to mistakenly shoot an unarmed black man than an unarmed white man.
I'm typical. The first time I took the test, years ago, I shot armed blacks in an average of 0.679 seconds while waiting slightly longer -- 0.694 seconds -- to shoot armed whites. I also holstered more quickly when confronted with unarmed whites than with unarmed blacks.
In effect, we have a more impulsive trigger finger when confronted by black men and are more cautious with whites. This is true of black players as well, apparently because they absorb the same cultural values as everyone else: Correll has found no statistically significant difference between the play of blacks and that of whites in the shooting game.
"There's a whole culture that promotes this idea of aggressive young black men," Correll notes. "In our minds, young black men are associated with danger."
Further evidence for these unconscious attitudes toward race come from implicit association tests, a window into how our unconscious minds work. You can take them online at https://implicit.harvard.edu/implicit/.
One finding is that we unconsciously associate "American" with "white." Thus, in 2008, some California college students -- many who were supporting Barack Obama for president -- unconsciously treated Obama as more foreign than Tony Blair, the former British prime minister. Likewise, Americans may be factually aware that Lucy Liu is an American actress and Kate Winslet is British, but the tests indicated that Americans considered Liu as more foreign than Winslet.
Yet we needn't surrender to our most atavistic impulses. Prejudice is not immutable, and overall the progress in America on race is remarkable. In 1958, 4 percent of Americans approved of black-white marriages; today, 87 percent do.
There's some evidence that training, metrics and policies can suppress biases or curb their impact. In law enforcement, more cameras -- police car cams and body cams -- create accountability and may improve behavior. When Rialto, Calif., introduced body cams on police officers, there was an 88 percent decline in complaints filed about police by members of the public.
Yet an uncomfortable starting point is to understand that racial stereotyping remains ubiquitous, and that the challenge is not a small number of twisted white supremacists but something infinitely more subtle and complex: People who believe in equality but who act in ways that perpetuate bias and inequality.
Nicholas D. Kristof is a syndicated columnist with The New York Times. Contact him at Facebook.com/Kristof, Twitter.com/NickKristof or by mail at The Times, 620 Eighth Ave., New York, N.Y. 10018.