This extended abstract presents the core argument of the book project *Maschinenethik: Architektur der Begrenzung technischer Wirksamkeit im irreversiblen Raum*. The study argues that machine ethics begins at the wrong point when it asks how machines can be made moral. Moral is not treated here as an input parameter, a rule set, or an optimization target, but as a response to a more fundamental condition: irreversible effectiveness in reality. The central claim is that technical systems operate within a space in which actions can generate consequences that are not fully retractable, even when the underlying architecture remains reversible at the level of code, simulation, or model design. From this perspective, the argument develops in six steps. First, it reconstructs irreversibility as the structural condition of binding action. Second, it distinguishes decision from mere selection and argues that irreversible binding cannot be delegated in the strong sense. Third, it examines how delegation in complex technical systems leads to fragmented attribution and requires architectures of responsibility. Fourth, it analyzes scaling as a structural shift that pressures the singular case and reveals a non-scalable dimension of reality. Fifth, it formulates the problem of power as a question of proportion between technical effectiveness and the capacity to bear binding consequences. Sixth, it approaches machine dignity not as an anthropomorphic status claim, but as a limit concept marking the breakdown of total instrumental availability. The text does not propose a virtue ethics for machines and does not reduce the problem to regulation or alignment. Instead, it offers a philosophical framework for understanding the limits of legitimate technical effectiveness under conditions of irreversibility, delegation, scaling, and systemic power. This publication is an extended abstract of a larger German book project in philosophy of technology and machine ethics.
Michael Kübler (Sun,) studied this question.