The Entity Home Is Native, Borrowed, or Up for Grabs
Status: Original concept, first publication 12 July 2026.
I coined “entity home” back in 2015 to name the place an entity speaks from, the canonical spot a machine treats as the entity itself rather than a page about it. For years I described it as though every entity could have one, and working through the three classes has shown me that was too neat. An entity home isn’t a thing every entity gets. It’s a thing an actor has by nature, a tradeable earns only if it’s special, and a concept never truly possesses at all, which is exactly why concepts are the interesting case and, right now, the opportunity.
An Actor has an entity home natively, because it’s a single agent with one canonical presence
An actor is one thing that acts, so there’s a single place the acting comes from, and that place is the home. You have one. Your company has one. The machine wants a canonical node for every actor it models, and an actor obligingly supplies one, because a single agent with a single presence is precisely what a home is. This is the clean case, and it’s the reason personal and organisational brands are where entity-home work pays off first: the slot exists by nature, and the job is just to make it unmistakable.
Actors nest, so the home is a hierarchy, not a single point
I used to say “one Toyota”, and that was lazy. Toyota is a parent in Japan, and there’s a Toyota France, a Toyota in nearly every country on earth, each one an actor in its own right, each with its own entity home, each able to sign contracts and answer in its own courts, and each linking up to the parent above it. So the home isn’t a point, it’s a tree of actors, and resolution runs up the branches: the national actor first, the parent behind it. The model holds, it’s just richer than a single node, and that nesting matters the moment a tradeable needs somewhere to attach.
A Tradeable’s home defaults up to its owner, unless it’s unique enough to earn its own
A product can’t act, so on its own it has nowhere to speak from, and its home defaults upward to the actor that makes it. That default only breaks when the tradeable is unique enough to hold a canonical presence by itself: a specific car with a chassis number, a specific building, a named mountain can each be homed, because there’s exactly one of them. A model can’t. Ask where the entity home of the Corolla is and you fall straight into the trap, because the Corolla isn’t one thing, it’s model times country times trim times year, thousands of variants with no single canonical instance, so the home collapses back up to Toyota the actor and every variant hangs off it as an attribute. Uniqueness lets a tradeable stand on its own. Replication and variant-explosion push it back up to its owner.
Google’s own markup rules are a tell, not a rulebook
You can watch Google sorting the world this way in its own structured-data guidance, if you read it as a sign rather than a law. A page about a person or an organisation can be marked as the entity itself, the mainEntity, its home, while a page about a product or a concept is marked only as what the page is about. Google doesn’t police that, there’s no inspector, nobody gets fined: it’s a signal, a window onto how Google already divides the world into things that can be an entity’s home and things that can only ever be described. And the part that matters more than the markup is what the signal does to everyone else. When Google declines to recognise a type of thing for its richer results, it quietly discourages the whole web from marking that thing up as an entity in its own right, and discouragement at that scale changes what gets published, and what gets published is what the next machine trains on. So the rule isn’t only a hint about how Google thinks today, it’s Google nudging the shape of tomorrow’s training corpus toward its own model of what an entity is. Watch what the machine rewards, and you can read what it’s teaching itself to become.
A concept has no home, so the machine builds one, and today that’s Wikipedia
A concept can neither act nor be owned, so it has no native place to speak from at all, and yet the machine still demands a canonical node for it, because an engine can’t reason about a thing it can’t point at. So it manufactures a home, and today, overwhelmingly, the manufactured home for a concept is its Wikipedia page. That’s a borrowed home, rented from an editor pool, standing in for a presence the concept doesn’t naturally have. Which means the slot is occupied by convention, not by right, and anything occupied by convention can be occupied by someone else.
The vacancy is fillable, and it’s closing
Here’s the opening, and here’s the honesty that has to come with it. Because a concept’s home is a manufactured slot, you can move into it: publish the definitive, dated, corroborated account of a term, and the machine may well resolve the concept to you, deserved or not, because it isn’t awarding merit, it’s resolving a reference to whoever occupies the slot. This works best for new terms, where no internal definition exists yet and the vacancy is wide open. I don’t expect it to last, because Google and OpenAI have little reason to keep renting their definitions from Wikipedia or from you. The likely path is that they internalise Class 3, define concepts themselves, and reclaim the slot. Up for grabs is only ever up for grabs for now.
Occupy the slot, then convert it to authorship before the window shuts
For me, this is the whole move, and it’s why the work is urgent rather than merely clever. You occupy the vacancy while it’s open, and you use the occupancy to bank the one thing the engines can’t nationalise: the dated, authored, corroborated record that you originated the concept. The home is transient, and the authorship is permanent. When Google absorbs the definition into itself, the Wikipedia-shaped slot disappears, and what survives the absorption is the provable chronological fact that you coined the thing, first, on a date, with independent voices confirming it. That’s temporal authority landing exactly where the ontology says it should: occupancy is how you get read today, authorship is how you’re still credited after the machine takes the concept in-house. Occupy for now. Author for keeps.
The Three Entity Classes series
- The Three Entity Classes: Everything Either Acts, Gets Owned, or Defines the World
- Classify the Entity, Not the Word: Why “France” Is Three Different Things
- Every Class Collides: The Same Name Names a Thousand Things, and Uniqueness Wins
- I Own Brand SERP, and I Don’t: How a Trademark Spawns a Second Entity
- Acquired Distinction: How a Common Word Gets Dragged Down a Class
- The Entity Home Is Native, Borrowed, or Up for Grabs
- E-E-A-T Lives at Class 1, Because Trustworthiness Is Something Someone Confers on You
- The Agent on the Boundary: Who Counts as an Actor Is About to Be Worth a Fortune
Publication note: this article is part of the Three Entity Classes series by Jason Barnard, first published on jasonbarnard.com on 12 July 2026. It corrects the earlier “only actors can have an entity home” framing: actors home natively and nest, unique tradeables can home while models collapse up to their owner, and concepts occupy a manufactured, closing vacancy that should be converted into permanent dated authorship.