Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Political Science, 2002. / Includes bibliographical references (p. 395-406). / This dissertation explores the causes of, and possible remedies for, extremely violent ethnic conflict. It starts from a robust yet under-explored finding in the literature: Most groups that fall victim to genocidal violence actually trigger their own demise by launching armed secessions or revolutions against state authorities that only then retaliate with genocide or forced migration ("ethnic cleansing"). Accordingly, the dissertation asks why groups that are vulnerable to genocidal retaliation would provoke that very outcome by launching such "tragic challenges." To explain this phenomenon, the dissertation employs three case studies to test three hypotheses drawn from rational deterrence theory. The cases focus on three subordinate groups whose armed challenges provoked genocidal retaliation: Bosnia's Muslims in 1992-95; Rwanda's Tutsi in 1990-94; and Kosovo's Albanians in 1998-99. To gain further insight by adding variation on the theory's dependent variable, the dissertation also examines an earlier period of the third case during which the subordinate group did not launch a violent challenge, despite having substantial grievances, and thereby avoided genocidal violence (Kosovo's Albanians in 1989-97). he three hypotheses are as follows: (1) the group did not expect its armed challenge to provoke genocidal retaliation; (2) the group expected to suffer genocidal violence regardless of whether or not it launched an armed challenge; (3) the group expected its armed challenge to provoke genocidal retaliation but viewed this as an acceptable cost to achieve its goal of secession or revolution. The dissertation confirms the third hypothesis: subordinate groups launch tragic challenges when they expect to prevail and are willing to civilians as the cost of doing so. / (cont.) Most surprisingly, the dissertation finds that a key cause of the optimism leading to tragic challenges is the expectation by subordinate groups of receiving humanitarian military intervention if they provoke genocidal retaliation against themselves. This reveals that international policies of humanitarian intervention create moral hazard, encouraging vulnerable groups to launch armed challenges and thereby potentially causing the tragic outcomes that these policies are intended to prevent. The dissertation concludes by exploring prescriptions to mitigate this newly discovered "moral hazard of humanitarian intervention." / Alan J. Kuperman. / Ph.D.