The increasing complexity of the State, in particular the use of new information and communication technologies have led to the emergence of new types of bureaucracies such as system-level bureaucracy and screen-level bureaucracy. The first is composed of specialists who develop information systems for public policies, while the screen-level bureaucracy is formed by middle-level and street bureaucrats who act at the end of the provision of public services, depending on the systems to perform the work. The present work has as object of study the analysis of the relations between these bureaucracies in the implementation of public policies, with emphasis on the coordination problem. In this inter bureaucratic relationship, communities of practice, discretion and expertise are important analytical elements, since they allow the creation of strategies of action by the actors and the coordination of the policy. The theoretical references on bureaucracy and implementation of public policies were constructed not only in classical texts such as Weber (2007), Wildavsky and Pressman (1973) and Lipsky (1980) that discuss the relations between bureaucracies in the implementation of But also in the more recent work by Guy Peters (1998) and Bovens and Zouridis (2002) on the coordination and role of ICTs in policies. From a methodological point of view, it is a case study comparing two public policies that differ in the way in which the relations between system level bureaucracy and screen level bureaucracy occur. In the two informal, mutually dependent, close and complementary, informal learning networks allow actors to create and use action strategies that contribute to coordination. The distinction between policies has occurred in the identification of greater coordination problems in education policy. The explanation first relates to the substantive difference of the tax collection policy, which is of greater importance to the government and has a greater maturity in the development of information systems than other secretariats, which provides greater expertise and more collaborative relationships. Empirical evidences related to the organizational characteristics of the secretariats contribute to explain this distinction, such as: the greater turnover of the bureaucrats of the Department of Education (SEED); The separate physical location of the education systems and teams development team, different from what occurs at the Treasury Department (SEFA).