Sovereign reading room — long-form essays by Muhammad Arif bin Fazil on AI governance, institutions, and the path to AGI.
15 essays · sealed 999 · updated 2026-08-07
"Humans, institutions, and AI do not become evil because they are malicious. They become evil when they get better at protecting narratives than correcting themselves with reality." — Arif Fazil · AGI Paradox trilogy · 2026-08-07
Why bigger models won't save us, and why the smartest thing in the room might be an institution.
The third axis of failure — what Acemoglu missed about institutional decay.
Why the path to AGI will consume more tokens, more compute, and more patience.
What Petronas bought, spent, and walked away with — a case study in sovereign capital.
A geoscientist stumbles into AI governance and discovers the question nobody asked.
Why the smartest thing in the room might be an institution, not a model.
What Acemoglu missed — institutions don't just fail by being extractive or weak. They fail by simulating.
Why the path to AGI will consume more tokens, more compute, and more patience than anyone admits.
A sovereign capital case study — two decades of investment, extraction, and exit.
Before an agent can act, it must be named. Naming is not taxonomy — it is constitutional.
Tools compete. Most die. The ones that survive are not the strongest — they are the most metabolized.
The tool does not express the thought. The tool IS the thought. When you build, you think.
Past, present, future — three timelines that collapse into one constitutional boundary.
A 6-axis constitutional coordinate system for mapping LLM value space.
From BINATANG to WARGA — the constitutional journey from animal to citizen.
A confession from someone who built the thing he hates — and why that matters.
Zoeppritz as constitutional floor — why the AVO-attention equivalence demands governed transformers.
A formal bridge between seismic AVO and transformer attention — the trilogy continues.
Completing the trilogy — a rigorous mapping of seismic AVO fluid factor onto transformer attention residual.