Many years ago I watched a TV show about a wealthy man who was a superhero on the side. It was unique because the star was (still is, for that matter) African-American. The hero had a rumpled, geeky white guy who provided him with technological wizardry. It was one of those moments in popular culture when the boundaries of “normal” were pushed outward.

Because I am who I am, I read about the origin of the series. I found out that the original pitch for the program was that, instead of a rumpled, geeky white guy, the hero’s scientific team consisted of two Africans: a man and a woman. They wore not rumpled clothes but traditional African clothing. Not only that, but these would be African scientists practicing African science.

The young network executives responsible for the show ridiculed the idea. It was ludicrous to think that African science conducted by African scientists dressed as Africans had any credibility or appeal. Audiences wouldn’t understand it. The executives didn’t understand it. It wasn’t normal. It didn’t make sense.

So the show was produced with a white guy—which seemed normal—and, lest we forget, an African-American star—less normal, but it made sense.

Last week researchers published a study of grants for biomedical research granted by the National Institutes of Health. The aim of the research was to find out whether any disparities existed in grant-making based on the race of the applicant. They didn’t expect any, but to their surprise found that an African-American applicant was 36% less likely to receive a grant than a white applicant.

What strikes me is that even at the NIH, in the heart of science, where objectivity rules, social processes of the subtlest kind are at work. The researchers who discovered this bias are at a loss to explain the outcome. Although applicants must specify their race in submitting a proposal, the reviewers who make the decision never see that. But they do see a wide variety of other information. Humans are not stupid when it comes to identifying someone’s social status from seemingly innocuous cues: name, institution, academic pedigree, research subject.

The researchers suggest providing better training to applicants. Better training is not the problem. Overt racism is not the problem, either. The problem is that no social force exists to prevent the covert use of social cues as to an applicant’s race.

In their previous research on racial disparities among biomedical faculty, the researchers found no disparities. Thus their surprise at bias showing up in NIH grants. What they failed to acknowledge is that 50 years ago disparities were rampant and alleviated only after pitched political battles that have forced decision making into the sunlight. Not so with the NIH grant process.

More shocking than the 36% disparity between black and white researchers in getting NIH grants is the incredibly low number of black researchers who apply: only 1.4% of applicants are African-American. I remind you that African American’s account for 13% of the total population.

As the researchers themselves argue, “throughout the education pipeline, blacks are less likely to graduate from high school, attend college and major in biomedical science, and obtain a PhD in biomedical science.”

The problem isn’t covert and likely unintentional racism at the NIH. The problem is the educational process that discards minds that are embodied in the wrong skin color. So if the problem is the so-called pipeline, it seems sensible to ask why the dismal outcome and what can be done.

I’m sure you’re painfully aware that a heated fight over school reform is underway. The major forces for reform have mounted a naked attack on teachers. The attacks start at the national level in the Obama administration’s Department of Education with such ideas as “reorganizing” the classroom using high tech to increase class size and lower the number of teachers. Basically, the argument is that the pipeline is a problem because of bad teachers.

It’s a mechanistic argument: bad outcome, get a better machine. How do you know it’s a bad outcome? Test scores. How do you know when the problem’s fixed? Test scores. What’s the goal of process? Test scores. For better education, get better teachers who will produce better test scores.

As Dana Goldstein argues in The Nation “The research consensus has been clear and unchanging for more than a decade: at most, teaching accounts for about 15 percent of student achievement outcomes, while socioeconomic factors account for about 60 percent.” She contrasts the neurotic hovering over teachers and teacher performance with school projects like the Harlem Children’s Zone, “which provides “wraparound” social and health services alongside charter schools.”

It is painfully obvious that children who are hungry do not perform well in school. The statistic thrown around these days is that 17 million children in the US do not have a secure source of healthy food. A disproportionate percentage of those children are African American.

You want to improve educational outcomes? Feed the children. You want to improve the educational pipeline and increase the number of African American researchers who receive NIH grants? Feed the children.

Yet programs that do that very thing are destined for the killing floor of deficit reduction. Does that seem normal? Does that make sense?