When I joined the administration of George H.W. Bush in 1991, I had no preconceived ideas about choice and accountability. "Choice" meant vouchers, a cause that had been rebuffed repeatedly in state referendums and by the courts; the issue had never gotten my attention. "Accountability" was one of those platitudinous terms that everyone used admiringly but no one did anything about. My abiding interest, then and now, was curriculum—that is, the knowledge that is purposefully taught in subjects like history, geography, the arts, literature, civics, science and mathematics. I believed that American schools should have a coherent curriculum so that teachers would know what they are expected to teach and children would have continuity of instruction, no matter where they lived.

However, after I left the administration in 1993, I supported the nascent charter school movement, even going to Albany, New York, to urge legislators to adopt a law permitting such schools to be created in the state. I supported merit pay as a form of accountability, on the assumption that teachers whose students are more successful should be paid more than their peers. I supported testing, expecting that better information would help to pinpoint where improvement was needed. I was affiliated with conservative think tanks, including the Manhattan Institute, the Thomas B. Fordham Foundation and the Hoover Institution. When Congress passed the No Child Left Behind (NCLB) legislation in 2001 and President George W. Bush signed it in 2002, I applauded.

In my new book, The Death and Life of the Great American School System: How Testing and Choice Are Undermining Education, I describe how I came to repudiate my support for choice and accountability, though not for curriculum reform, which I still believe is necessary and valuable. Some news accounts have said I did a U-turn, but in fact I was really reverting to the time before I jumped on the bandwagon of organizational change and accountability, the time when I knew that the only changes that matter are in the classroom and in children’s lives. Reaching this conclusion was not an overnight conversion but rather the result of watching how the policies of choice and accountability played out in reality. I began to re-evaluate my views as early as 2004, as I watched the implementation of mayoral control in New York City, with its heavy emphasis on accountability and choice.

Many people have told me that I should have known better, and they are right: I should have. But I didn’t, and I am trying to make up for it now.

NCLB made accountability the nation’s education policy. It used to be the case that educators could more or less ignore federal education policy, because it seldom touched their classrooms. Thanks to NCLB, this is no longer the case. Now federal policy affects every school. In my book I define the governing philosophy of NCLB as "measure and punish." I conclude that this approach, which uses accountability as a stick to threaten schools, has failed.

The law requires that every state test every student from grades three to eight in reading and mathematics, then disaggregate each school’s scores by race, limited English proficiency, disability and low-income status. The law mandates that every student in every group must reach 100 percent proficiency by 2014. Every state is left to choose its own test and define proficiency as it wishes. If only one group in a school fails to make steady progress toward that goal, the school faces increasingly severe remedies and sanctions. First, the school will be put on notice; then all students in the school (including those who are doing well) will be offered the choice to go to a different school. In the third year, low-income students will be offered free tutoring after school. If the school does not meet its projected target over five consecutive years, it may be privatized or handed over to state control or charter managers; its staff may be fired, it may be closed or it may be restructured in some other way. Currently, about one-third of all public schools in the nation—more than 30,000—have been stigmatized as failing because they did not make what the law calls "adequate yearly progress."

At the same time that NCLB told states to set their own standards, Congress directed them to participate in the federal tests, known as NAEP (National Assessment of Educational Progress), which serve as an external audit of their claims. (Some cities also take the NAEP exams but do so voluntarily, to learn how they compare with the rest of the nation.) Since NAEP is administered to samples of students, no one knows in advance which students will take it, so no one can prepare for it and no one has any incentive to cheat or game the system.

By demanding that all students reach proficiency by 2014, NCLB incentivized states, districts and schools to cheat and game the system. That is the direct outcome of high-stakes testing. Some states have lowered their testing standards, thus making it easier for students to be rated "proficient." Consequently, many states now claim dramatic improvement in their test scores, but these gains are not reflected on the tests given every other year by the federal government. In Texas, where there was supposed to have been an educational miracle, eighth-grade reading scores have been flat for a decade. Tennessee claimed that 90 percent of its students were proficient in 2007, but on NAEP only 26 percent were.

In contrast, progress on the NAEP tests has been meager. Billions have been invested at the federal and state levels in testing and test-preparation materials. Many schools suspend instruction for months before the state tests, in hopes of boosting scores. Students are drilled on how to answer the precise types of questions that are likely to appear on the state tests. Testing experts suggest that this intense emphasis on test preparation is wasted, because students tend to learn test-taking techniques rather than the subject tested, and they are not likely to do well on a different test of the same subject for which they were not prepared.

Despite the time and money invested in testing, scores on NAEP have increased slowly or not at all. In mathematics the rate of improvement was greater before NCLB was passed. In reading there have been gains in fourth grade, but the national scores for eighth graders were essentially the same in 2009 as they were in 1998.

It is not only the sluggish improvement in test scores that is troubling. Nor is it the frequency with which states and districts manipulate the scoring of the tests to produce inflated gains. The biggest victim of high-stakes testing is the quality of education. As more time is devoted to reading and math, and as teachers are warned that the scores in these subjects will determine the fate of their school, everything other than reading and math gets less time. This is what doesn’t count: history, literature, geography, science, the arts, foreign languages, physical education, civics, etc.

So, the emphasis on accountability for the past eight years has encouraged schools to pay less attention to important subjects and inflate their test scores by hook or by crook. NCLB’s remedies don’t work, its sanctions don’t work and the results are unimpressive. Why members of Congress and Washington think tanks continue to defend this toxic law is a puzzle.

The other popular nostrum of our day is "choice," which has captured the imagination of big foundations and many wealthy business leaders. Vouchers still have fervent proponents, even though only 30,000 students use them, and there is slight evidence of their effectiveness. Vouchers have been replaced by charters as the vehicle for promoting free-market reforms. What was but an idea in the late 1980s is a full-blown movement today, with 1.5 million students enrolled in 5,000 charter schools.

Charter schools receive public money but are privately managed. Unlike regular public schools, they operate free of most rules and regulations. More than 95 percent of charter schools are nonunion. When the state comptroller in New York sought to audit the state’s charter schools, they sued to block him, claiming that they should be trusted to do their own audits.

Charters vary widely in quality. Some are excellent, some are abysmal, most are somewhere in between. The only major national evaluation of the charter sector was carried out by economist Margaret Raymond at Stanford University. Her study was funded by the staunchly procharter Walton Family Foundation, among others; yet she found that only 17 percent of charters outperformed a matched public school. The other 83 percent were either no better, or they were worse. On the NAEP exams in reading and mathematics, students in charter schools perform no better than those in regular public schools, whether one looks at black, Hispanic or low-income students, or students in urban districts.

Yet charter schools have passionate advocates, certainly on the right and also from a group called Democrats for Education Reform. Some charters are run by for-profit firms, some by nonprofits, and some are managed by community-based organizations. Their business model often involves a high turnover of teaching staff, because teachers are expected to work long hours, sometimes sixty to seventy hours weekly, plus be available by cellphone at all hours to their students. This works because so many charters are nonunion schools, but it is difficult to see how this model could be replicated. Not only does it preclude teachers’ unions; it precludes a teaching profession in which teachers expect to make a career of teaching and have families.

The media like to focus on a star charter school, as though one extraordinary school is typical. The teachers are young and enthusiastic; the children are in uniforms and well behaved, and they all plan to go to college. But such stories often overlook important factors about charters: one, the good charters select students by lottery, and thus attract motivated students and families; two, charters tend to enroll a smaller proportion of students who are limited–English proficient, students with disabilities and homeless students, which gives them an edge over neighborhood public schools; and three, charters can remove students who are "not a good fit" and send them back to the neighborhood school. These factors give charters an edge, which makes it surprising that their performance is not any better than it is.

The original vision of charter schools in 1988, when the idea was popularized, was that they would be created by venturesome public school teachers who would seek out the most alienated students, those who had dropped out or those who were likely to do so. The teachers in these experimental schools would find better ways to reach these students and bring what they’d learned back to the regular public school. The fundamental idea at the beginning of the movement was that charter schools would help public schools and enroll students who needed extra attention and new strategies.

Now the charter sector sees itself as competition for the public schools. Some are profit-driven; some are power-driven. In some cities, charter chains seek to drive the public schools out of business. In Harlem, which has a heavy concentration of charter schools, the regular public schools must market themselves to students and families; they typically have a budget of $500 or less for fliers and brochures. The aggressive charter chain that competes with them has a marketing budget, according to the New York Times, of $325,000. The expansion of charters has been mightily underwritten by hedge-fund managers, the Walton Family Foundation, the Eli and Edythe Broad Foundation and other major benefactors.

Just at the point where I had made an ideological break from my past support of accountability and choice, the Obama administration came into office. I expected that Obama would throw out NCLB and start over. But, on the contrary, his administration has embraced some of the worst features of the George W. Bush era. Obama’s Race to the Top competition dangled $4.3 billion before cash-hungry states. To qualify for the money, states had to remove any legal barrier to the expansion of charter schools. States also had to agree to create data systems making it possible to evaluate teachers by their students’ test scores. And they had to pledge to "transform" or "turn around" low-performing schools.

Each of these elements is an echo of Bush’s policies. The expansion of charters fulfills the dreams of education entrepreneurs and free-market advocates, who would dismantle public education if given the chance. Judging teachers by test scores is wrongheaded because students’ scores are affected not only by what the teacher does but by such important factors as poverty, student motivation and family support. Yet only teachers will be held accountable. "Turning around" low-performing schools is a euphemism for NCLB-style punishments: if scores don’t go up, schools are closed, privatized, turned into charters or handed over to the state. When Rhode Island authorities announced their intention to fire the staff at the only high school in Central Falls, their decision was hailed by Education Secretary Arne Duncan and President Obama. I thought the decision was meanspirited and wrong. Last year the state commissioner of elementary and secondary education sent a team to the same school and said it was making progress. Why not build on that progress? Why fire the staff without evaluating anyone? (Recently a deal was reached to rehire the entire staff; it includes conditions such as a longer school day and more after-school tutoring.)

Obama’s emphasis on evaluating individual teachers by students’ test scores has set off a frenzied effort by states to rewrite their laws in hopes of snaring some of the federal billions. The legislature in Florida recently passed legislation that would have denied tenure to new teachers, based 50 percent of teachers’ salaries on gains in student scores, removed any credit for teachers’ experience and education, and funded testing of all subjects by cutting 5 percent from the budget of every school district. Teachers and parents rallied and persuaded Governor Charlie Crist to veto the bill, probably ending his career in the Republican Party. Similar measures are moving through other state legislatures, cheered on by think tanks in Washington.

In my book I gather the evidence that persuaded me that accountability and choice are not likely to improve American education. Accountability, narrowly focused as it is, dumbs down education. Choice may enfeeble public education by draining away the best students and leaving what remains to the remnants of the public education system.

None of the policies that involve testing and accountability—vouchers and charters, merit pay and closing schools—will give us the quantum improvement that we want for public education. They may even make matters worse.

We need a long-term plan that strengthens public education and rebuilds the education profession. We need better-educated teachers who have degrees in the subjects they teach; we need principals who are themselves master teachers, since they are the ones who evaluate and support the teachers; and we need superintendents who are knowledgeable educators, since they make crucial decisions about curriculums, instruction and personnel.

We must ensure that every student has the benefit of a coherent curriculum, one that includes history, literature, geography, civics, science, the arts, mathematics and physical education. And we must attend to the conditions in which children live, because their ability to attend school and to learn is directly influenced by their health and the well-being of their families.