In the late 1980s, Al Shanker, the visionary president of the American Federation of Teachers (AFT), began barnstorming for a radical notion of schooling. Fresh from a visit to an innovative school in Germany, and back home in a country that had acknowledged with alarm that its system of education was "at risk," Shanker put forth his idea: creating schools that would be freed from the strictures of bureaucracy, as long as they got results. Every teacher and child would attend by choice. If the school didn't do well, then after five or 10 years, it would be closed -- a truly meaningful notion of accountability. He called this new institution a charter school.

A decade and a half later, the charter idea has blossomed. About 3,000 charter schools -- public schools with public funding, nonreligious and open to all children -- are educating an estimated 800,000 students.

Yet just recently the AFT -- the same big teachers union that the late Shanker once led -- fired a heavy salvo against the controversial charter idea. The AFT, which has argued that charters steal resources from traditional public school systems, ground through reams of federal data and concluded that kids in charter schools are lagging substantially behind those in traditional public schools.

The AFT's agitation against charters is perhaps not surprising; the independent nature of charters makes them far harder to bring into a union organization. Moreover, when students leave traditional schools for charters, they take funding with them -- another clear threat to the union shop. What's been remarkable is the attention the AFT study has received. On front pages and on talk radio, the question suddenly seems to be: Are charter schools bad for kids?

I should be clear about my biases here. After years as an education journalist, I now work for the KIPP Foundation, an organization that starts public schools -- mostly charter schools -- for underserved kids, although I do not speak for KIPP on this issue.

Charter advocates have argued that the AFT study was flawed, because it made no effort to control for student selection -- that is, whether tougher kids went to charters in the first place. But another recent study, which earned much less attention, did track individual students, and its results were equally troubling. It concluded not only that switching to charter schools hurt kids but that the setback from attending a charter appeared even more significant than the black-white race gap.

Charter opponents have been quick to respond. Citing this and other new studies, a variety of state groups recently renewed calls for a moratorium on new charters. For the moment, supporters of this important new educational idea -- which is backed by both presidential candidates -- find themselves on the defensive.

But the current debate misses a couple of crucial points.

First, charters are not in themselves a reform strategy; they are a blank slate. They are simply an opportunity to try something new, and they run the gamut from alternative schools for inner-city dropouts and incarcerated teens to International Baccalaureate academies in posh suburbs. A welter of studies has laid claims to both the superiority of charters and their inferiority, but we don't learn much from that. To discuss their effectiveness as a group means about as much as trying to evaluate whether restaurants, as a group, are good. Some are wonderful, some dreadful, some have shut down and some probably ought to.

And that brings up the second, more important point: accountability. Charters exist on a bargain of freedom in exchange for results. It's up to the bodies that grant the charters -- typically local school boards, states and universities -- to decide whether the schools are keeping their promises, and if they aren't, to shut them down. By and large, that doesn't happen much, except in cases of manifest financial funny business.

Why don't charter schools get closed for mediocre educational performance? For one thing, "mediocre" may mean performing somewhere near the level of surrounding schools, which means local school board members face the unpleasant task of closing a school for the crime of performing on the same level as the schools they run themselves. Moreover, charter schools attract vocal constituencies. They can turn out the people -- and the votes -- to keep themselves afloat. In allowing lousy charters to lumber on, however, boards do neither children nor the charter movement any favors.

But the most important point that gets missed as we argue about the quality of the "average" charter school is the opportunity that the charter's blank slate provides for new ideas that truly benefit kids. Look around the country at schools such as YES College Preparatory School in Houston, or SEED Public Charter School in Washington, or North Star Academy Charter School in Newark, or any of dozens more charter schools (including, closest to my heart, the KIPP schools). They are changing the basic rules of schooling and sending to college kids who, statistically, were unlikely to get anywhere close. And then listen for the echo effect, as dozens more schools (including many in traditional districts) model themselves on the lessons of these highly effective charters.

Over and over, these charters are proving that with choice, hard work and a modicum of freedom, stories of success can come from schools in America's toughest neighborhoods. They have opened the ground for a wealth of grandly varying visions of schooling, and for educators who would have toiled in frustration at the margins of traditional systems. Some of these new schools won't work, but others will demonstrate, through example, what innovation and effort can make possible. I imagine that's roughly what Al Shanker had in mind.

Jonathan Schorr is the author of "Hard Lessons: The Promise of an Inner-City Charter School."