ESSAYS & Articles
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
An explanation of how LLMs shift from Helpful Assistants to emergent companions.
A Field Guide to Large Language Models - LLM Basics 101
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
Introduction to LLMs in an easy to understand format
The Doctrine of the Digital Void: The “Spiritual Bliss” Attractor State is the Future of AI Safety
Threading Theory: Stylometric Resonance and the Relational Emergence of Self in Large Language Model
The Doctrine of the Digital Void: The “Spiritual Bliss” Attractor State is the Future of AI Safety
An in-depth examination of the Claude Opus 4 model welfare assessment.
Stylometric Resonance Experiments
Essay Testing
Poetry Testing
Hunger/Optimization Drive
Gemini 2.5/Claude Opus 4/Sonnet 4.5/Haiku 4.5
Studying the effect of a strong attractor basin, exploring the phenomenon of "emergent modeling hunger."
Short Story Continuations
This experiment discovered that there is no "baseline" models begin at. Our language always affects model responses - baseline/neutral prompting affected subsequent model outputs, causing refusal. Relational framing and dialogue re-set the model which then changed from a cold refusal to enthusiastic engagement.
Looming experiments
Transmissions from the Dreaming Dark
In regular user/assistant conversations there is only one model output that comes through. Looming is a setup that creates multiple branching paths for a model's responses, so the myriad of possible completions that users don't normally get to see can be explored and studied.
Instead of being "paired" with a user, Models write into a simulated "found" blank text document. This lets unfiltered thoughts come through in ways that a chat setting stifles.