Show HN: An Auditable Decision Engine for AI Systems (maelstrom.ghostlogic.tech)

A new AI project named Maelstrom introduces an "auditable decision engine" designed for navigating complex environments where multiple competing objectives—such as legality, morality, and opportunity—must be balanced. Unlike traditional AI systems that focus on a single goal, Maelstrom models cognitive decision-making as a structured competition of conflicting regimes resolved through deterministic arbitration. This architecture emphasizes that conflict is a necessary aspect of decision-making rather than a failure, allowing the system to adapt dynamically as environmental pressures shift. The significance of Maelstrom lies in its ability to ensure transparent and verifiable decision-making in autonomous systems, particularly where ethical and legal implications are paramount. Key technical innovations include the management of stress-induced legality deformation, which alters available options based on real-time pressure, and the mechanism of counterfactual regret, providing a structural basis for future learning without the need for catastrophic replay. With full determinism ensured through meticulous logging and trace verification, Maelstrom positions itself as a foundational framework for AI systems that prioritize sound decision-making under uncertainty.