00 / ORIENTATION

Research, systems, and essays.

Current work on AI, continuity, authority, and adaptive systems.

01 / CURRENT WORK

Current Work

Recent research and essays.

01DARK AGENCYRESEARCH PAPER · V0.2.0

The Internet Is Already Built from Places Where Agency Can Disappear

Machine action can remain locally visible while its origin, authority, objective, and continuing effects become globally unreconstructable.

An action may be visible to every system it crosses while the operation remains invisible to all of them. Read paper ↗
02SOURCED CASE STUDYGOVERNED AGENT SYSTEMS

The AI Agents Didn’t Survive. Their Institutions Did.

Short-lived agents passed work, practices, and state across generations. The governance problem is not only whether an agent persists, but what survives it.

What survives the agent? Read case study ↗
03ESSAYGOVERNED ADAPTATION

Stratified Plasticity

Intelligence can change at many layers without treating every correction as a reason to rewrite the deepest machinery beneath it.

High behavioural plasticity. Low constitutional drift. Read essay ↗
04ESSAYCONTINUITY / EVIDENCE

Show Me How You Changed

Continuity is not a story about growth. It is a record of what changed, why, and what happened differently afterward.

Accountable change, not retrospective autobiography. Read essay ↗
05ESSAYLOCKED · 17 SEP 2026

The Subconscious Interface

AI as an external surface for making recurring patterns in thought, reaction, and expression inspectable without pretending the machine can read a hidden self.

Reflection → inspection → authorship. Read essay ↗
SELECTED WORK

Longer-form publications.

READER DRAFT SEP 2026 3 PARTS / 10 CHAPTERS

Something Worth Choosing

A Theory of Legitimate Continuity

What may legitimately cross from what was into what comes next: memory, rupture, inheritance, judgement, refusal, and the future's right to choose.

Page four of Something Worth Choosing showing chapter one, The Persistence Mistake.
ARTIFACT / PAGE 4The Persistence Mistake
Title page of The Human Time Dividend, canonical dissertation version 1.0.
PUBLICATION / V1.0Canonical dissertation
CANONICAL DISSERTATION V1.0 25 JUL 2026

The Human Time Dividend

A Measurable Theory of Human Return from Automation

A measurable test for whether automation returns meaningful human time rather than merely relocating institutional burden.

Enter publication ↗ ePDF ↓
CLAIMARTIFACTDEMONSTRATIONOUTCOMERECORD

Claim the state that exists. Show what supports it. Leave the unresolved parts attached.

02 / SYSTEMS

Systems

Operational projects and methods.

PUBLIC INTRODUCTION / 001CONSENT-GOVERNED

ONE STEP

A request and offer network that begins before requester and provider already know one another.

DECLAREBOUNDROUTEMATCHVERIFYCONSENTTRANSFERCLOSE

Compatibility is not permission. No stage automatically authorises the next.

OPERATOR PRACTICE

Methods for making action and self-observation inspectable.

METAFRAMEWORKPROPOSED · V0.1

Operator Execution Metaframework v0.1

Govern the transition from recognised intent to warranted execution without turning every move into a bespoke ceremony.

SEE → WARRANT → MOVE → OBSERVE → CONTINUE Read metaframework ↗
ESSAYLOCKED · 17 SEP 2026

The Subconscious Interface

How AI can make the patterns beneath conscious thought inspectable without pretending the machine has privileged access to a hidden self.

Reflection → inspection → authorship Read essay ↗

03 / CURRENT EDGE

Governed Agent Systems

Active inquiry, not a finished portfolio object.

STATERESEARCH SERIES · FIRST ARC
OBJECTArtificial systems that act
STATUSActive research / non-canonical

A five-piece research arc on capability, delegated authority, escalation, receipts, and revocation in artificial systems that act, now joined by sourced case studies of what these problems look like in the wild.

Capability is real. Permission is separate.
The machine can.May it?
CAPABILITYAUTHORITYESCALATIONRECEIPTREVOCATION
CS-001 / SOURCED CASE STUDY The AI Agents Didn’t Survive. Their Institutions Did. The next agent may be new. Its environment is not. →
Read series ↗ Start with 001 →
Steler-OS04 ↘

04 / OPERATING INFRASTRUCTURE

Steler-OS

The infrastructure used to organise and operate the work.

SELECTED TOP-LEVEL ESTATES / 00–37 STELEROS / PUBLIC ABSTRACTION
00–05
01CROWN02CODEX03ATLAS04FORUM05COURT
06–15
06WORKSHOP07COCKPIT08OBSERVATORY09FOUNDRY10PORT12GARDEN14THRESHOLD15ARMOURY
16–20
16BASEMENT17UNDERWORKS18MENAGERIE19ARCHIVE20MUSEUM OF MISTAKES
22–29
22BACK ALLEY23TAVERN24RITES25WEATHER26NIGHT SKY27ROADS28RUINS29QUIET ROOM
30–37
30SYSTEMS31CANVASES32RESIDENTS33MEMORY ATLAS34STUDIO35MACHINE ROOM36PUBLICATIONS37OPINIONS
OPERATING INFRASTRUCTURECONTINUITY / GOVERNANCE

Steleros / The Constellation

A working operator environment for preserving provenance, current state, decisions, authority boundaries, research, evidence, and continuity across the work.

STATE LIVES HERE. AUTHORITY IS CHECKED BEFORE CHANGE. EVERY CHANGE LEAVES A RECEIPT.