This paper argues that precommitments can provide reason to argue for paternalistic interventions which are to the benefit of the agent -- arguably as judged by themselves (cf. Thaler and Sunstein, 2008). If an agent voluntarily restricts their future choice sets to ensure that certain actions will not be taken, this indicates a foreseen change of preferences (e.g. due to self-control problems, etc.). Modelling the agents as an assembly of different selves playing multi-stage game with each other, we identify cases in which such interventions can indeed be to the benefit of all selves of an agent as judged by themselves. In addition, we consider situations where other criteria -- such as considering average utility or focussing on single selves -- could provide (weaker) reason for paternalistic actions and nudging.