The book, extended
Blogs
Concepts explained, standards pressure-tested, and market claims answered. The posts follow the book's three acts. ACT I: why the current value system fails. ACT II: how value actually behaves. ACT III: how value is governed into existence. Start with the act that matches your question, jump straight to a topic below, or browse everything. Claims here stay under review; what changes is recorded in The Claims Register. New to the vocabulary? The glossary defines the standard’s terms in one place.
Browse by Topic ▾
Browse by Chapter of the Book ▾
Browse by Role ▾
A standards body for AI economics now exists. Its opening measure of value is the share of work done without a person, which prices substitution and never asks whether the promised outcome arrived.
August survey data shows every executive seat naming itself the signer of AI purchases, while unowned decisions double one level down. No assignment exercise settles contradictory claims. A buyer’s record does.
A number restated three times is not three sources. One Gartner figure inverted its meaning in two public steps, and the buyer’s own evidence file is where the standard applies first.
A seller-side vendor has made custody of the buyer’s outcome definitions into a permanent staffed role. No equivalent exists on the buyer’s side, and the definitions are being written into the systems that recommend.
A TCO model prices what its author can see. The costs that decide the outcome live on your side of the table, and they do not behave like line items.
Sellers treat buyer silence as a wall the right technique will breach. It is a line item. Disclosure costs capacity no buyer has, and only one kind of channel scales.
On 5 August the UK published its value model: terms in advance, minimum weights of 10 and 20 percent, KPIs on a public record, and consequences at the next award. The mechanism transfers.
A specification for computable value models is free, open, and properly licensed. Its governance still sits with a vendor whose commercial thesis is on the roadmap. A free license settles who may use it and leaves open who decides what it says.
Agent-to-agent commerce is being wired with a machine-readable model of the buyer’s value, written entirely on the seller’s side of the table. The buyer’s counterpart artifact does not exist yet.
Organizations are built to detect failure. Value rarely fails. It drifts, through three forces that each look reasonable at the moment they operate and leave the record looking clean.