This paper examines a novel type of multi-agent problem, in which an agent makes multiple identical copies of itself in order to achieve a single agent task better or more efficiently. This strategy improves performance if the environment is noisy and the task is sometimes unachievable