And what is it Mr. Meadmaker

?
We must decouple school attendance from geography. If where you live dictates where you go to school, the situation will persist.
I'll requote the paragraph again, for context, and I'll probably quote portions of it again. I have never seen it summarized quite so well.
The heart of the problem with our urban schools is neither the education bureaucracies nor teachers unions, as Chubb, Moe, and many other voucher advocates have contended, flawed though those institutions may be. Instead, as the sociologist James S. Coleman found in the 1960s, a student's family's income and the collective social and economic background of his classmates are by far the most important influences on his academic future. Not only do lower-income students tend to score relatively poorly, children of any background who attend high-poverty schools are far more likely to produce worse test results than they would in schools with primarily middle-class students. America's urban school systems remain almost universally dysfunctional, primarily because the country as a whole is about as segregated by race and income as at any time since the civil rights revolution.This means that in the existing voucher programs, which have been confined to city school districts, students have had only a limited choice between public schools in low-income neighborhoods and private institutions—mostly parochial schools—that serve almost identical populations.
Let's start with the single most important observation from that paragraph.
Instead, as the sociologist James S. Coleman found in the 1960s, a student's family's income and the collective social and economic background of his classmates are by far the most important influences on his academic future.
Does anyone want to disagree with this? It certainly fits the available data.
To add emphasis:
Not only do lower-income students tend to score relatively poorly, children of any background who attend high-poverty schools are far more likely to produce worse test results than they would in schools with primarily middle-class students.
Now, place yourself in a position of a middle class parent with a four year old child. Presumably, you want your child to do well in school. You want high test scores, which hopefully are correlated with high real knowledge. The single most important thing you can do to help your child to be a higher achiever is to get him into a classroom with other high achievers, and that means getting him into a school with primarily middle class students.
If you live in the City of Detroit, that means you move. If you can afford to move, and you have a child, you move. Or, you send that child to private school, where only (or mostly) middle class children can afford to attend. However, for the amount of money you need to invest, most people decide it's more cost effective to move.
And that is why our cities will remain segregated as long as the public schools exist in their current form. When my kid turned four, I moved. I was willing to live in a low income neighborhood, but I was not willing to send my son to school with our low income neighbors, for exactly the reasons Mr. Coleman identified fourty years ago. Sending him to school with those kids would have made it more likely for him to get low test scores than sending him to a school with primarily middle income kids.
It's that simple, and you quoted the research yourself.
So what's the solution? Obviously, you have to give the city kids the chance to get away from their low income schools. If they did that, more of them would choose to continue to live in the low income neighborhoods. That means less segregation in the neighborhoods, and in the schools.
The only way I can see to do that, realistically, is with schools of choice that cross school district lines. Vouchers are the most flexible program, but multi-district charter schools would work also.