This work contrasts several modeling strategies and proposes a new modeling approach. It demonstrates the existence of a new account to explain the first episode of skilled problem solving observed in Anzai & Simon (1979). The resulting characterization of the learners strategies is stronger than the classical Selective Search account based on heuristics that avoid move repetitions (Anzai & Simon, 1979; VanLehn, 1991; Ruiz & Newell, 1993). The new interpretation passes the tests of exhaustive protocol coverage, lack of conflict with the data, and parsimony contrary to the Selective Search heuristics and to modeling accounts based on those heuristics. This paper emphasizes the need for more research on strategy identification. It also calls attention to the complexity of modeling as a cognitive task an important issue given the central place occupied by modeling in cognitive science research.