MERKABAH: An Architecture for Correctable Intelligence
Year: 2026
Type: preprint
Abstract: MERKABAH: An Architecture for Correctable Intelligence — Research Release 0.1 MERKABAH is a proposed architecture for correctable distributed intelligence that investigates how autonomous computational components can coordinate without establishing permanent centralized authority. The architecture comprises an always-operational eight-dimensional computational substrate (the Cube), a temporary event-triggered coordination mechanism (PigeOn), and an independent distributed challenge mechanism (Anti-PigeOn). A central design principle is that Pause constitutes a legitimate decision outcome, preserving the ability to reconsider, challenge, or decline intervention while independent computation continues. This foundational research manuscript introduces MERKABAH’s 14 proposed design laws, its philosophical framework of Eden, Wisdom, NOW, and shadows, and its approach to temporary authority, distributed governance, memory, recovery, and preserved correctability. The research proposes falsifiable hypotheses concerning coordination efficiency, resistance to authority capture, adaptive decision-making, and recovery under changing system conditions. Research status: This is an early-stage conceptual and architectural research proposal. The complete MERKABAH architecture has not been independently validated or production-certified. The philosophical propositions are presented as hypotheses rather than established scientific findings. Author: Shervin Ghorbany Version: 0.1 Keywords: Artificial Intelligence, Distributed Intelligence, Multi-Agent Systems, AI Safety, Corrigibility, Distributed Governance, Decision Theory, Event-Triggered Coordination, MERKABAH.
Author: Shervin Ghorbany
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225817689
Cites:
Cited by:
Related to: