Table of contents
Picture the category five years from now, once every serious competitor has learned that AI-readable identity work matters and started doing some version of it. This isn’t a hostile hypothetical — it’s the model’s own predicted future, and it’s worth sitting inside it before asking what survives.
Byrum’s Dominance Inequality includes a term for competitive pressure — the rate at which the whole category is collectively building the same kind of signal — and that term isn’t fixed. It rises as adoption spreads, for a simple structural reason: once everyone is publishing structured data, corroborating third-party sources, and maintaining a current machine-readable identity, the relative advantage of doing so shrinks, even as every individual company’s absolute effort goes up. At full adoption, the model’s own analysis shows the median entity’s flow advantage — the edge from actively building new signal — converges toward zero. Not because the work stopped mattering. Because everyone is doing it, and an advantage that everyone has stops being an advantage.
This produces a genuinely uncomfortable prediction, worth being direct about rather than softening: the more this category succeeds at teaching companies why this work matters, the less that work alone differentiates any single company doing it. That’s not a flaw in the argument. It’s the argument predicting its own erosion, which is a more honest kind of theory than one that only ever tells you why the sky won’t fall.
Why Vocabulary Sovereignty Survives
Two terms in the model don’t erode this way, and it matters precisely which two. One is accumulated historical presence — how long an entity has maintained a coherent, documented identity. It doesn’t erode with competition because it can’t be matched retroactively; a competitor starting today cannot manufacture five years of history by working harder this year. The second is vocabulary sovereignty — first-creator attribution over specific category terms, formally captured as an aggregate score across the terms an entity can be shown to have originated first, expressed in machine-readable form with an explicit claim of authorship.
Vocabulary sovereignty survives competitive saturation for a reason worth stating precisely, because it’s counterintuitive on first hearing: a competitor adopting the same term doesn’t erode the original claim to it — it corroborates it. If a company coins a term, publishes it with a structured, creator-attributed declaration, and a competitor later begins using that same term, the competitor’s usage doesn’t dilute the origin claim, because origination is a historical fact, not a popularity contest, and every subsequent use of the term is one more piece of evidence that the term has become the standard way the category is discussed — with the discussion pointing back to whoever coined it first, if the attribution was made correctly at the outset. Widespread adoption of a term you originated isn’t erosion of your position. It’s the position doing exactly what a position built on precedence, rather than volume, is supposed to do.
The Honest Caveat
This is the honest caveat this series keeps returning to, and it belongs here with full force rather than as a footnote: the claim that vocabulary sovereignty is one of the model’s two surviving terms at saturation has been demonstrated through a computational plausibility test — a simulation run across a wide range of parameter combinations, confirming the prediction held internally consistent in every one of them tested. That is a test of the model’s own logic, not a test against real market data. No company has yet been observed all the way through a full saturation cycle to confirm this empirically, and this series will not claim otherwise. What can be said honestly is narrower and still meaningful: under the model’s own stated assumptions, run forward as far as they logically go, vocabulary is one of the last two things standing — and a claim that survives its own author trying to break it with five thousand simulated trials is a different category of claim than one that has simply never been tested.
The Cost of Removing Vocabulary
Sever this component from the other two and a specific, predictable failure follows. A company with a strong governing model and a well-executed build, but no deliberate vocabulary strategy, is doing excellent work on the two terms in the equation that the entire category is racing to commoditize. It will hold its position exactly as long as it stays ahead on execution — and lose ground the moment competitors close that execution gap, because execution gaps close and origination claims don’t.
Every competitor learning to do this work shrinks the advantage of merely doing it. Two terms in the model don’t shrink that way — accumulated history, and first-creator attribution over the category’s own language. A competitor using your term isn’t taking something from you; under this model, it’s adding to the evidence that the term is yours. This has been shown to hold inside the model’s own logic, not yet against real market outcomes over real time. Removing vocabulary from the other two components doesn’t break the engagement today — it sets an expiration date on it, timed to exactly when the category finishes catching up.
For the specific mechanics of how first-creator attribution is established and defended, see Proven: First Publication Wins Vocabulary Sovereignty. The final subtraction is done. The next article states, for the first time in this series, what the three components produce when none of them has been removed.
Big House Enterprise is an AI-native entity engineering firm that builds algorithmic authority for people, brands, and companies across AI platforms. Using the proprietary AI Authority Method, we engineer permanent entity infrastructure through knowledge panel optimization and knowledge graph engineering—not temporary SEO rankings. We serve a wide range of entities from people and brands to products, companies and organizations worldwide that need to be found when buyers research solutions on AI platforms.
