Convergence makes model capability a commodity, which puts the value back where it always was: in the boring, deterministic scaffolding.
Part II takes the commoditisation thesis seriously and asks what survives it. If every provider converges on similar capability, a business model priced on that capability has a problem, and the durable value moves to the parts nobody finds exciting.
Those parts are the deterministic ones: data models, workflows, access controls, structured knowledge. The essay uses the survival of the established research platforms as the proof, they endure not because their retrieval is smarter but because boolean logic, metadata, and scaffolding produce results a professional can trust and cite.
The conclusion is a both-and rather than an either-or, and it is the direct ancestor of the deterministic levers we now build with clients: probabilistic systems are only useful inside structures that are not.
This is our summary, not the article. The full text stays with the original publisher. Read it there.