Who's Allowed to Approve a New Denomination

Opening Scene

Anyone with sufficient metalworking skill could, technically, physically strike a coin that looks reasonably convincing. Almost nobody is actually authorized to do so. That gap between raw capability and genuine authority is exactly what keeps a currency trustworthy — the value of a coin depends entirely on it having come from a legitimate, authorized source, not merely on it looking correct.

Governance over who can create and approve metric definitions in a semantic layer needs this same clear authority, and its absence undermines the whole system’s trustworthiness.

In Plain English

Metric governance means having clear, defined rules about who can propose a new metric definition, who reviews and approves it, and what standards it needs to meet before being published to the semantic layer where everyone will trust and rely on it. Without this, a semantic layer’s core value proposition — that its definitions are authoritative — quietly erodes, because “authoritative” requires a real, trusted approval process behind it.

The Old Way

Some semantic layer implementations, especially in their early stages, allowed relatively open, unreviewed publishing of new metric definitions, on the reasoning that lowering friction would encourage adoption. This made short-term sense but created a longer-term problem: a semantic layer’s credibility depends entirely on its definitions being genuinely trustworthy, and unreviewed definitions undermine that trust just as surely as unauthorized coins undermine a currency.

Once a few poorly-reviewed or simply incorrect definitions made it into a semantic layer, and were subsequently discovered to be wrong, the damage extended well beyond those specific metrics — it eroded confidence in the semantic layer as a whole, with users reasonably wondering what else might be similarly unreliable.

What’s Changing (and Why AI Is the Reason)

AI-assisted review can lower the friction of proper governance without sacrificing genuine rigor. Rather than governance meaning slow, entirely manual review that discourages legitimate metric creation, AI-assisted review can handle much of the mechanical verification — checking for double-counting risk (Article 3), consistency with naming standards, alignment with existing related metrics — leaving human reviewers to focus specifically on genuine business logic judgment calls.
AI-assisted approval routing can direct a proposed metric to the right reviewer based on its actual domain and complexity. Rather than a single bottlenecked review queue, AI-assisted classification of a proposed metric’s domain and risk level can route it to an appropriately knowledgeable reviewer, speeding legitimate approvals while still maintaining real oversight.
AI-assisted governance auditing can verify that published metrics actually went through the required approval process. Similar to the governance themes covered in this site’s orchestration-workflow-tools topic, AI-assisted analysis can periodically audit a semantic layer’s published definitions against its own governance records, catching cases where the process was skipped or bypassed.

The Metaphor, Fully Extended

Mint Element	Metric Governance Concept
Only an authorized mint being permitted to strike official currency	Only an approved process being permitted to publish a metric definition
Unauthorized coins undermining trust in the entire currency system	Unreviewed metric definitions undermining trust in the entire semantic layer
Royal inspectors verifying every new die meets the mint’s standards	AI-assisted review handling mechanical verification of proposed metric definitions
Different regional mints handling routine versus significant currency changes appropriately	AI-assisted approval routing directing proposals to appropriately knowledgeable reviewers
A royal auditor verifying every coin in circulation actually came through proper channels	AI-assisted governance auditing verifying published metrics went through required approval

For Beginners: What to Actually Do

Practice checking, for a semantic layer you rely on, whether there’s actually a defined governance process for approving new metric definitions, or whether publishing is effectively open and unreviewed.
Get comfortable with the idea that governance friction, done well, isn’t bureaucratic obstruction — it’s what makes a semantic layer’s definitions genuinely trustworthy in the first place.
Notice how your own confidence in a metric changes once you learn whether or not it went through a real review process before being published.
Understand that a semantic layer without genuine governance is at real risk of becoming just another source of unreliable, ungoverned numbers, defeating its entire purpose.

For Practitioners and Leaders: The Deeper Layer

Establish a clear, defined governance process for metric definition approval, specifying who can propose, who reviews, and what standards apply before publishing.
Use AI-assisted review to handle mechanical verification efficiently, freeing human reviewers to focus their judgment specifically on genuine business logic decisions rather than routine consistency checks.
Use AI-assisted approval routing to avoid a single bottlenecked review queue, directing proposals to reviewers with genuine domain expertise while still maintaining real oversight.
Use AI-assisted governance auditing periodically to verify your process is actually being followed in practice, not just documented on paper.

Quick Recap

Metric governance means having clear rules about who can propose, review, and approve new metric definitions before they’re published to a trusted semantic layer.
Open, unreviewed publishing undermines a semantic layer’s core value proposition, since its trustworthiness depends entirely on definitions being genuinely reliable.
AI-assisted review and approval routing can lower governance friction without sacrificing rigor, handling mechanical verification while directing genuine judgment calls to appropriate reviewers.
AI-assisted governance auditing can verify that published metrics actually went through the required approval process, catching cases where it was skipped or bypassed.

Where This Fits in the Series

Article 12 covered handling legitimate contextual variation. This article covered clear authority over what gets published as trustworthy. Article 14 looks at reading the mint’s ledger a century later, when the original authors are long gone.