Shadbolt, Hall & Berners-Lee's 2006 Semantic Web retrospective names ontology development and maintenance cost as 'an often quoted concern'

semantic-webontologyknowledge-graphcodificationai-strategycross-time-bridge

Five years after Berners-Lee, Hendler & Lassila's 2001 Scientific American paper introduced the Semantic Web — a vision of machine-readable, codified knowledge layered over the web — three of the field's own principals wrote a retrospective in IEEE Intelligent Systems. In it, they name the cost of building and maintaining ontologies as "an often quoted concern about the Semantic Web." The source's own wording is a hedge, not a verdict: it reports that the concern is widely raised, not that it was the decisive cause of the vision's limited uptake. Treating it as the single confirmed culprit would overstate what this quote, on its own, supports.

The claim matters as a 25-year-old anchor point for a live 2026 question: enterprise "semantic layer" pitches for agentic AI (claim-tung-roussiere-2026-codify-tacit-knowledge-into-private-knowledge-graphs-as-moat) are, by name and mechanism, a rerun of the Semantic Web vision this retrospective is looking back on. Whatever answer the 2026 revival has for ontology maintenance, it is answering a concern the original team was already naming in 2006 — not a new problem. A later, more specific first-person account of why ontology-maintenance projects historically failed is argued in claim-reynar-three-2001-2005-knowledge-systems-died-on-schema-population-maintenance, which this note does not itself establish; the 2006 source only names the concern, not its causal weight.

This bridges to the vault's existing codification thread (observation-codification-is-the-shared-hinge-of-automation-and-imitation), which has so far treated codification as roughly binary (written down or not). This note's source instead treats codification as something with an ongoing cost, separate from the act of writing knowledge down in the first place.