The Method
Controlled Vocabulary is a daily lexicon of trustworthy data. It is built for the corporate hostage — the practitioner held accountable for data they are given neither the access to reach nor the authority to decide. The satire is on their side.
- 01
Collect
Each morning we read a few hundred RSS and Atom feeds across governance, quality, contracts, stewardship, lineage, modeling, analytics, and AI readiness. We keep snippets, summaries, metadata, and links — never the full article. The work belongs to the people who wrote it.
- 02
Cluster & score
Items are normalized, de-duplicated, and clustered into the day's actual stories. A score weighs cross-source confirmation against vendor framing, novelty against noise. Most of what trends is filed, not promoted.
- 03
Select — by a human
An editor reviews the clusters as pills, promotes the few that matter, files the rest, and strikes the terms that can't be defined yet. Nothing is published automatically. A draft becomes an issue only when a person approves it.
- 04
Map every claim
Each entry carries a claim-to-source map: the specific sentences we stand behind, and the sources that support — or contradict — them. If we can't map it, we don't print it.
Short, opinionated, useful. We translate the day’s data news into work: what changed, why it matters, what a data leader should do next, and which part of the corporate theater is pretending the problem is a tooling gap.
We don’t write “leverage,” “unlock,” “transformation journey,” or “rapidly evolving landscape.” We don’t punch down at stewards, analysts, or the people holding the pipeline together at 3 a.m. The target is always the ritual, never the practitioner.
A data product without decision rights is a dataset with better branding.