Extracted from media publications, “event data” are used in academic studies and in strategic conflict prediction. They make it possible to produce knowledge on the behavior of actors in international relations and provide decision-making support for rulers. Since their creation in the 1960s, the artisans of this knowledge have constantly renewed their promises to optimize political decision-making in a context of crisis, in exchange for funding their research. However, by exploiting what is published by the news industries, these peace and war studies uncritically echo the selective representations shaped by journalists. This article shows that the promises of decision optimization can not be kept. It also highlights the extent to which knowledge using event data contributes to the reproduction of relations of domination, initially in the field of international information, in the academic field of crisis analysis and prediction.