In a recent CircleID piece, I described a structural pattern that appeared consistently across several large nTLD namespaces: a speculative noise layer in the high teens to low twenties, a dominant neutral middle making up 55–70% of each zone, a smaller investor-grade layer, and a genuine operator-grade layer sitting in the mid-single digits.

The finding was that namespaces with similar registration volumes could have materially different compositions underneath.

Since publishing that analysis, I’ve extended the same pipeline to a significantly larger namespace—more than nine million unique domains after deduplication, roughly double the largest namespace in the original study. The question was simple: does the four-layer pattern hold at a fundamentally different scale, or does it break?

The short answer: the shape held. But two findings emerged that the earlier analysis did not capture.

Finding 1: The Pattern Is Scale-Invariant—But the Ratios Are Not

The four-layer structure appeared in the larger namespace exactly as it did in the smaller ones. A speculative noise layer. A dominant neutral middle. An investor-grade segment. And a thin operator-grade layer. The pattern was unmistakable.

What changed was the balance. The operator-grade layer in this namespace was proportionally thinner than in any namespace I had previously analyzed—sitting below 4% of the total zone. The investor-grade layer, by contrast, was among the largest I’d seen. Speculative inventory outnumbered adoption-grade inventory by roughly three to one.

That means the full spread of operator-to-investor ratios across all the namespaces I’ve now examined is roughly fourfold—from namespaces where operator-grade inventory slightly exceeds speculation, to namespaces where speculation outweighs adoption by three to one. Same structural profile. Dramatically different ecosystem balance.

This is a meaningful result. If the four-layer pattern is structural—something that emerges from how nTLD namespaces are populated rather than from any individual registry’s strategy—then the relevant question for any TLD is not whether these layers exist, but what the ratios between them look like and how those ratios are trending over time.

Finding 2: Noise Has a Signature

The earlier analysis noted that noise floors varied between roughly 15% and 22% across namespaces. The larger namespace pushed that ceiling higher—closer to 28%. But the volume of noise was less interesting than its composition.

In the original namespaces, the dominant noise categories were consonant-cluster strings and ultra-long keyword combinations. In the larger namespace, the noise profile was structurally different. Hyphenated domains exceeding practical length accounted for the largest single noise category—roughly double the proportion seen in any previous namespace. Numeric spam was the second largest, appearing at rates several times higher than in the earlier analysis.

Meanwhile, consonant-cluster noise and random alphanumeric patterns—the dominant categories elsewhere—were proportionally smaller in this namespace.

In other words, the noise floor is not just a percentage. It has a composition. And that composition varies meaningfully across namespaces in ways that appear systematic rather than random.

Why does this matter? Because different noise compositions suggest different registrant acquisition dynamics. A namespace heavy in numeric spam and long hyphenated strings shows a pattern consistent with bulk, pattern-based registration dynamics. A namespace where the noise is more consonant-cluster and random-alphanumeric driven suggests a structurally different acquisition pathway.

The noise signature, in short, carries signal about how a namespace is being populated. It may not tell you who is registering, but it offers structural clues about why.

What this adds:

The original analysis proposed that registration volume compresses meaningfully different namespace compositions into a single number. The extended analysis adds two things to that claim.

First, the four-layer pattern appears to be structural rather than incidental. It held at more than double the scale of the original study, across a namespace with different positioning, a different operator, and a different registrant base. If it holds across additional namespaces, it may represent a fundamental feature of how nTLD ecosystems organize rather than a characteristic of individual registry strategies.

Second, noise composition deserves attention alongside noise volume. Two namespaces can have similar noise floors—say, one at 22% and another at 28%—while being noisy in completely different ways. That difference is diagnostic. It points toward the acquisition dynamics, registrar behaviors, and promotional structures that shaped the namespace into its current form.

Across the namespaces analyzed so far, the operator-to-investor ratio exhibited greater variation than any other structural measure examined. That makes it a promising candidate for continued study—and possibly worth tracking alongside registration volume as the dataset grows.

The structural shape tells you what kind of ecosystem a namespace is. The ratio tells you which direction it leans. Both are measurable from a zone file. Neither appears in standard registration reporting.