Abstract: Within the domain of soft real-time interactive simulations the timeliness of a computation can be more vital than its accuracy in ensuring the success of the overall system. At the same time, designers wish to provide the most accurate answers possible within the given time constraints. Time complexity-based multilevel modeling provides an avenue for satisfying this trade-off. However, implementing and managing such models in traditional, third generation languages can dramatically increase the overhead and complexity of the code while limiting portability and code reuse. Many of these issues can be resolved by architecting the overall system around a data flow perspective rather than control flow. The SHADOW System, a language, compiler and run-time engine built around data flow concepts, has been used to demonstrate the viability of these concepts in both structured laboratory tests and large scale applications.