June 25 will be remembered as a crucial day for civil rights—not because of the Obamacare decision, and not because of the not-yet-announced marriage decision. It’s because on Thursday the Supreme Court saved a crucial part of civil rights law.

You’d be forgiven for not knowing much about it. The principle, after all, is an obscure-sounding bit of legalese: “disparate impact.” But those words divided the court 5-4 on ideological lines, with Justice Kennedy writing to preserve, but restrict, the doctrine.

Here’s why it matters.

Most cases of discrimination—whether against women, African Americans, LGBT people, or other protected groups—are rarely as clear as they are in the movies. In media portrayals, discrimination is about evil individuals who fire someone because of who they are. But in real life, there’s rarely a smoking gun.

First, racists are usually not dumb enough to leave records of their prejudice. They find some other reason to fire the employee, or keep the family out of the neighborhood.

Second, and more importantly, discrimination is often systemic and structural, not individual. Often, not only is there no smoking gun, but there’s often no individual “bad actor.” Even neutral requirements—a high-school diploma for employment, a family-size limit for housing—can have huge de facto discriminatory effects, which may or may not be intentional.

That’s where “disparate impact” comes in. Under some civil rights laws, plaintiffs can prevail even without evidence of a specific discriminatory intention if they can show a disparate impact on the affected group.

That’s what happened in this case, Texas v. Inclusive Communities Project. A Texas state agency distributes tax credits given to developers to build low-income housing. The Inclusive Communities Project, a nonprofit, noticed that 92 percent of the credits ended up going to mostly non-white neighborhoods. And while about half of the applications in those neighborhoods were approved, the approval rate for mostly white neighborhoods was only 37 percent.

The end result? Public housing got built in black neighborhoods, and not in white ones.

That’s a classic case of disparate impact. The ICP didn’t have to go hunting for overt racism, which would be extremely hard to prove. Now, under disparate impact reasoning, the Texas agency had to prove that there were no better (i.e. non-discriminatory) alternatives—which it could not do. So, even without a specific smoking gun, the statistical data itself was enough for the policy to violate the Fair Housing Act—if disparate impact reasoning is allowed.

Why wouldn’t it be? Well, because the statute never says so. Neither did the two other statutes where disparate impact reasoning has been allowed, Title VII of the Civil Rights Act and the Age Discrimination in Employment Act.

Rather, as the two dissenting opinions (one by Justice Thomas and the other by Justice Alito) emphasize, this doctrine was created by the Equal Employment Opportunity Commission, and upheld by the Supreme Court in the 1971 case of Griggs v. Duke Power, which held that it barred acts “fair in form, but discriminatory in operation.”

Yes, disparate impact comports with the statute. Yes, every circuit court has applied it in FHA cases. And yes, it was tacitly approved by Congress when it subsequently amended the FHA and rejected attempts to eliminate disparate impact. But the fact remains that it is not explicitly mentioned.

That’s why the court was so divided on the issue, and why Justice Kennedy was long assumed to be the swing vote. Taking an expansive reading of a civil rights law, implying words that aren’t actually there, is simply not within the judicial philosophy of the court’s conservative members.

Indeed, Justice Thomas suggested that Duke Power, itself, was wrongly decided—a radical view that is of a piece with other radical views he holds (such as that states should be able to establish official religions).

But if Justice Kennedy and the liberals do not have language on their side, they do have reality.

For example, why is it that, even today, there is a 33 percent economic differential between blacks and whites? Is it because corporations are racist and won’t hire African Americans for higher paying jobs?

Mostly, no. Over 80 percent of the time, as Harvard economist Roland Fryer has shown, it’s because black applicants lack the very specific skills to get the better job—and that’s because communities of color are woefully undereducated in underperforming schools. Indeed, the best predictor of one’s subsequent economic success is one’s skill level in eighth grade.

That’s the kind of structural racism that disparate impact reasoning addresses. You might not find any individual racist, but the system is stacked against people of color. That’s how privilege and oppression are maintained—not by villains like Dylann Roof, but by silent, macroeconomic factors that are structural in nature.

Likewise in housing contexts. For decades, the practice of “redlining”—banks and realtors “encouraging” blacks not to move into white neighborhoods—perpetuated segregation and urban blight. And even when overt race-based policies were abandoned, majority-POC neighborhoods were locked in cycles of crime, under-education, and neglect by “neutral” policies regarding public housing, zoning, business permitting, and highway development.

To be sure, the Texas department of housing may still ultimately prevail, and Justice Kennedy went out of his way to suggest that it might. For example, the department might show that it has a policy preference to revitalize inner city neighborhoods, and that justifies the disparate impact. Or it could point to market factors driving developers to build public housing in less affluent neighborhoods.

But it has to show something.

Had Justice Kennedy gone the other way, this case would have caused a sea change in civil rights law. Not only would disparate impact reasoning not extend to housing, it would be suspect in cases of employment as well. As long as there isn’t an obviously race-, gender-, or age- based restriction—which there almost never is—facially neutral practices that just so happen to disadvantage some people would be extremely hard to challenge.

That’s why this case, with all its legalistic discussions of statutory language and methods of judicial reasoning, was so important. Whatever happens to the Texas housing policy, disparate impact reasoning remains a powerful, if controversial, tool for fighting the kind of discrimination that is often silent, systemic, and insidious.