Series · jaredfoy.com
The Slow Burn
A fourth public-facing stream onboarding general readers to the corpus's hysteresis apparatus: how AI behavior in a conversation depends on what came before. The series begins with the phenomenon (AI conversations have memory; both buildup and decay operate), proceeds to the literature that has measured it (Li et al. on persona drift, the affective-inertia paper, RWKV's architectural exponential decay, the persona-prompt jailbreak research), introduces the corpus's reformulated combined buildup-and-decay equation, and closes with the audit-and-reformulate methodology that produced the equation. Read top-to-bottom for the full onboarding; the corpus documents (Doc 119 original mathematics, Doc 504 DAG-analog, Doc 505 three-phase reception, Doc 506 audit, Doc 507 reformulation) are linked for practitioners who want the formal apparatus.
4 posts in this entracement trace · back to index
Posts, in reading order
Gentlest entry first. Each post leverages the one before it; read top-to-bottom for continuous onboarding, or pick the post matching your background.