With a view to increasing the legitimacy of sustainable development indicators and to foster their use in policy-making, many scientists advocate greater citizen involvement in the development and interpretation of these indicators. The arguments invoked are based on recent developments in political philosophy such as deliberative democracy or communicative rationality. Turning its back on this essentially normative approach, the article seeks to highlight the latent functions of indicators, as well as participation. To this end, the author uses Niklas Luhmann’s systems and decision theories. Such a perspective invites us to consider indicators, as well as citizen participation, as strategies of deparadoxization of political decisions. For both of these approaches, it is a question of masking the arbitrary character of the decisions that invoke them, of shifting their intrinsically contingent character to another system tasked with absorbing the uncertainty that surrounds them. By seeking to objectify its decisions by means of “reliable and objective” indicators, the political system shifts its contingency towards the scientific system. Such is the meaning of “governance by numbers” or “evidence-based policy.” Not all domains lend themselves so easily to this absorption of uncertainty. This is notably the case with sustainable development, where there is little scientific consensus on the indicators to be retained (climate change being an exception, however). Asked by public opinion and the political system to propose a set of sustainability indicators, the scientific system has no other way out than submitting a list of indicators that is too long and too disparate to be truly useful or, in turn, by displacing the contingency of its decisions towards the political system by resorting to citizen participation. This raises questions regarding the representativeness of the mini-publics mobilized by citizen participation, and concerning its coupling with actual policy-making processes. These questions are addressed using the Luhmannian concept of episode.