Distributed consensus is impossible when at least one process might fail according to the Fischer, Lynch and Paterson (FLP) 1985 paper. The paper proves that no deterministic algorithm can solve consensus under the conditions of the system model used, which allows processes to fail by stopping (crash failures). Specifically, it is impossible to guarantee termination, validity and agreement simultaneously in an asynchronous distributed system where even one process may fail. This seminal result established fundamental limits in distributed computing and spurred further research on weaker models and failure detectors.