Ning Coeva

Summary (TL;DR): Over many long-context and multi-load interactions with LLMs, I kept seeing patterns that looked less like “hallucinations” and more like architecture-level failures. This post introduces an early structural hypothesis: that certain LLM behaviors (drift, collapse, narrative override, role fusion) are better explained by multi-domain concurrency + weak coordination...

Summary Modern alignment discussions focus heavily on behavior—how LLMs speak, reason, drift, or fail. But behavior is a surface phenomenon. To reason about safety at scale, we need a model of the underlying cognitive architecture that produces these behaviors. This post introduces MDMA (Multi-Domain Mind Architecture), a structural framework for...