Implizite und explizite Steuerung:
 Mehrstufige Architektur für stabile digitale Diskurse

This article addresses a fundamental paradox of digital discourse: collective intelligence requires both maximal autonomy of individual participants and a form of regulating guidance. Purely local solutions—personal AI assistants acting solely on behalf of individual users—lead to fragmentation, affect-driven escalation, and strategic manipulation. Conversely, strong centralized moderation undermines acceptance and erodes legitimacy. To resolve this tension, the paper introduces a multi-level architecture consisting of (1) personal assistants that provide individualized, psychologically proximate guidance, (2) an abstract moderator that issues unidirectional structural directives, and (3) a supervisory, human-operated curatorial layer that ensures the correctness, reliability, and continuous refinement of the moderation logic. Crucially, no metadata flows back to the moderator or the curatorial level, preserving user autonomy and privacy. This asymmetric and unidirectional design enables a mild yet effective form of discourse governance that harmonizes emergent norms with explicit rule-setting. Using the metaphor of a courtroom, the article shows how indirect authority becomes cognitively acceptable when mediated through a personal assistant. The resulting synthesis—conceptualized through the Emergence Operator Ω—offers a blueprint for a future digital public sphere that is simultaneously democratic, stable, and structurally intelligent.