Information search in a distributed environment is an interactive process that involves both retrieval and the processing of information across users and artifacts. How the information is distributed across internal representations and external representations affects the efficacy of information search. Using a human-centered method called UFuRT, we developed an information search model and a taxonomy of search tasks. We further developed several prototypes of information search interfaces with different patterns of distributed information and investigated the relations between search tasks and interface types. The results from the analyses show that UFuRT is a useful process that not only provides design guidelines but also generates estimates of representational efficiencies, task complexities and user behavioral outcomes.