This article identifies three dominant models for democratic dynamics, based on battle, competition, and game, respectively, and assesses their strengths and weaknesses in the debate on democratic self-protection. Scholars typically describe democracy as competition or game, but antidemocratic threats and remedies in terms of battle, which causes unclarity of analytical vocabulary. However, some of these conceptual commitments, such as Karl Loewentein’s original militant metaphors and the assumed incompatibility between competition and active protection, are non-necessary and rather result from how these ideas originally entered the debate. The model of democracy protection as market regulation emerges in the German intellectual history of militant democracy and independently of laissez-faire assumptions. Critical assessment of each model suggests that, once clarified of misinterpretations, the competition model is the most coherent single framework to develop analytical terminology in. Particularly, it evades ‘the transition problem’ (when to switch from conciliary to military rhetoric and how to do that without aggravating political oppositions). The competition model also implies a relatively compelling and intuitive justification for democracy protection: it is justified to protect democratic competition from attempts to destroy the market structure altogether in favor of political monopoly. While the competitive model is the strongest, none of the models, however, fully capture the unique nature of democracy and its self-annihilating logic: democratic politics as an activity has an intrinsic relation to its constitutive rules and, consequently, can be utilized for antidemocratic ends in a way difficult to capture with models derived from adjacent fields.