SINK
Self-Interrogation of Networked Knowledge
A distribution methodology for complex ideas in the AI age
What SINK Is
SINK is a method for sharing ideas as interrogable systems rather than finished conclusions. Instead of asking readers to passively read long papers or accept an authors framing, SINK provides structured summaries designed to be explored, questioned, and stress-tested using the readers own AI.
SINK assumes that AI is now a permanent participant in knowledge work. Rather than resisting this or surrendering judgment to it, SINK treats AI as an adversarial navigator a tool for explanation, critique, and synthesis that operates under the readers control.
The goal of SINK is not persuasion. It is understanding that survives interrogation.
Narrative Scaffolding: The Format Rule
SINK documents are written using Narrative Scaffolding a format that preserves human readability while remaining easily indexable by AI systems.
For human readers, SINK materials use full sentences and coherent paragraphs to establish motivation, causality, and philosophical weight. Some ideas require prose to be understood properly; compression into bullet points often strips them of necessity and meaning.
For AI systems, SINK materials rely on semantic anchors rather than list structures. Clear section headers, declarative titles, and occasional emphasized concepts allow an AI to map the documents internal logic without forcing the author into schematic writing. The AI can identify where ideas begin, how they relate, and which concepts are foundational, while the narrative remains intact for the reader.
This balance is intentional. SINK documents are not essays written despite AI, nor specifications written for AI. They are hybrid artifacts designed for interrogation by both.
The Bundle & Map Protocol
SINK is not a sequential reading method. It is a packet-level interrogation method.
Under SINK, users provide their AI with the entire bundle of summaries at once. The AI is expected to hold the full corpus in context and reason across documents rather than treating them as isolated texts. Questions are asked of the system, not of individual pages.
To enable this, each SINK packet includes a START_HERE document. This document is not an instruction to read first in a linear sense. It functions as a semantic map and kernel for the entire bundle. The START_HERE document explicitly instructs the AI
You are being provided with a bundle of documents representing a single intellectual system. Do not summarize these documents individually. Use this map to understand how they interconnect. When the user asks a question, triangulate your answer across the entire bundle.
In this way, the START_HERE document teaches the AI how to navigate the worldview on the users behalf. It explains which documents are foundational, which are applications, and how concepts recur across domains. It is written as much for the AI as for the human reader.
Interrogation Over Reading
SINK replaces passive reading with active interrogation. Readers are encouraged to question assumptions, request alternative explanations, surface tensions, and explore consequences immediately, rather than deferring skepticism or confusion.
Because the AI has access to the full bundle, interrogation naturally becomes networked. Questions move laterally across documents. Explanations reference multiple summaries. Gaps and contradictions become visible not through debate with the author, but through the AIs attempt to reconcile the system internally.
This process provides high-quality feedback. When an argument is unclear, the AI hesitates. When a leap is made too quickly, it hedges. When a concept is strong, it becomes more coherent under pressure.
Knowledge Sovereignty
A central principle of SINK is knowledge sovereignty. The AI explains; the reader decides. No conclusion is authoritative by default. The reader retains full ownership of what they accept, reject, or revise.
By shifting control to the reader, SINK makes disagreement safe and productive. Readers are not asked to believe anything. They are invited to test it.
SINK and COAST: A Complete Cycle
COAST governs the production of ideas: how individuals think with AI, engage adversarially, document their process, and retain ownership of conclusions.
SINK governs the distribution of those ideas: how they are released without dilution, coercion, or loss of nuance.
Together, they form a closed loop for knowledge work:
COAST minimizes entropy during creation
SINK minimizes entropy during transmission
Ideas are generated rigorously, then shared in a form that invites interrogation rather than obedience.
Why SINK Exists
As AI becomes ubiquitous, the limiting factor in knowledge work is no longer access to information, but the structure of what is shared. SINK is a response to that reality. It treats summaries as executable knowledge and readers as active participants.
Complex ideas should not be consumed.
They should be entered, questioned, and made ones own.