SUPPOSE policy makers and politicians were required to report back to school for the 2003-04 school year. They would get their textbooks and resources, which for them would amount to reams of reports, studies, and blueprints about how to make No Child Left Behind (NCLB) work. They would have time lines and tasks, with no exceptions allowed (everyone must be on the same page at the same time). There would be tests of the policies they enacted and requirements that they show progress. And, at the end of the school year, they would be expected to be better able to answer the central question posed by NCLB: Can the federal government make students learn at higher levels? That would be a really new course of study. But, in fact, there isn't even a syllabus written for it yet. Beyond the details of rules and accountability plans that are making people grumpy and critical, NCLB has brought about an incredible shift in the federalism that traditionally governed education policy in this country. True, in 1994 the Clinton Administration's makeover of the Elementary and Secondary Education Act set national requirements for standards and accountability, but there was so much flexibility that federal/state roles remained about the same. The failure of most states to comply with that earlier legislation on a voluntary basis, especially during a time of economic prosperity, opened the way for advocates of low-income and low-performing groups of students to insist on mandates and consequences. Their allies, also waiting for a chance to change things quite a bit, were conservatives who supported choice as their goal. The cult of efficiency, as exemplified in annual testing, is the measure of how well this new NCLB experiment is working. The bargain was a stark one: more federal control for more money. But the latter has not kept pace with the former. There may be more total federal dollars for public education than ever before, but this is not the case in terms of the share of total costs of education, according to Gail Sunderman of the Civil Rights Project at Harvard University. So far, the syllabus for this new course in federalism seems to have too many open-ended questions. After failing to gain additional funding from the Bush Administration to carry out the NCLB mandates, the nation's governors are faced with major reductions in state funding because of budget shortfalls even as they are required to increase spending because of such NCLB mandates as testing and data collection. Will their support for education reforms hold? It's an open question. As accountability for adequate yearly progress moves into high schools, there is a clear data problem. The Houston school district's luster as an outstanding system, for example, was tarnished by an analysis this past summer that showed it had more than 5,000 additional dropouts. Yet NCLB requires that, as an indicator of academic performance at the high school level, states report the percentage of students graduating on time with a regular diploma. Almost everyone knows that dropout data are unreliable. Even the limited definition used by the National Center for Education Statistics (NCES) excludes 27 states, among them the largest (California, Texas, and Florida), because their methods of collecting the data do not align with that of NCES. The Urban Institute, relying on other sources of data, calculates that the real national on-time graduation figure is close to 66%. Using an on-time graduation rate of 75% as a goal, 20 of the 24 states in which data are available for all methods would meet the target under the NCES definition. Only eight or nine states would meet it under the Urban Institute's more comprehensive definition. …