What is intelligence, at each level and in between?
A cell, a brain, an organisation, a model, an agent: each embodies intelligence at its own level, physical, biological, digital. And much of it is between the units rather than in them, as music is between the notes and an ant colony thinks in the traces its ants leave for each other. Lem warned that we recognise only the intelligence that looks like us; we have always imagined it as a version of ourselves, and even modelling a machine on our own brain is anthropocentrism of a high order. So the design question for AI is how big the acting unit should be, and what has to pass between units for the whole to be intelligent.
- The Right Size of Intelligence · essay, 6 September 2026
- Intelligence Was Never in the Machine · essay, 21 June 2026
- An Intelligence Made Entirely of Maps · essay, 11 August 2026
- Shouting into the Lake · essay, 22 October 2026
Who may let a machine act, on whose authority, and within what limit?
The question a financial institution has to answer this year, and the oldest question of delegated authority in a new body. A firm writes delegated authorities for its staff and runs its agents on entitlements. Between the two there has to be a mandate: who may delegate the act, what evidence the agent needs before it acts, when the firm may refuse, and who answers afterwards.
- A Risk Appetite Where the Act Happens · working note, 28 October 2026
- The Mandate as a Unit of Risk · paper outline, 28 October 2026
- Model Risk When the Model Acts · essay, 23 June 2026
- Human-in-the-Loop Doesn't Survive Volume · essay, 25 June 2026
What is the value of what a model knows, and who pays for it?
A token has a price. What the model is permitted to rely on has none, so it is treated as free, and the cost appears later as a decision nobody can defend. Governed knowledge belongs in the economics of an AI task as a variable of its own.
- The Value of What the Model Knows, working note · working note, 28 October 2026
- The Value of What the Model Knows, outline · paper outline, 28 October 2026
- Admissibility · essay, May 2026
What does AI do to the value of holding, binding, or predicting a person?
And how has binding of that kind come apart before, when it has come apart at all. The history of holding a person is a history of the instruments that made it pay, and AI is a new instrument.
- Slavery in the AI Age · essay, 28 July 2026
- The Quiet Stranding · essay, 31 July 2026
- Shouting into the Lake · essay, 22 October 2026
How does participation give rise to knowledge, and then to wisdom?
The question behind all the others. Civilisations learned to act on answers they could not inspect, and each time the fix went around the decision and left the oracle alone. Some things can be designed; some can only be grown, with the rules written into the places where people meet.
- Guided Emergence · essay, 21 July 2026
- The Emergence Layer · essay, 14 July 2026
- The Oracle Problem Has Been Answered Three Times · essay, 16 July 2026
- The Tower I Mistook for a Lighthouse · essay, 26 July 2026