Why a shared name is the missing layer

Spatial computing runs on frames that do not line up. The standards that describe those frames are mature. What has been missing is a name for a thing that survives crossing between them.

Consider a single door. A map knows it as a feature in a coordinate reference system. A robot knows it as a plane in the map it built this morning. A headset knows it as an anchor in a frame it invented when you put it on. A scan knows it as a cluster of points. A twin knows it as an element in a model. Five systems, five representations, and no way for any of them to say the one thing that matters: this is the same door.

Geometry is solved. Identity is not.

The industry spent decades making the geometry interoperable, and it worked. There are good standards for expressing where something is, how it is oriented, and how the globe is indexed. Any two systems that speak them can exchange where something is. What they cannot exchange is which thing they are talking about, because there has never been a shared, persistent identifier for the thing itself.

The cost of that gap is quiet but constant. Every pair of systems that must interoperate re-registers, re-projects, or re-anchors. Shared augmented reality drifts. Digital twins lose registration against what they model. Robot fleets disagree about where objects are. And every AI built on top inherits the inconsistency, because it reasons over observations that were never tied to stable identities.

Identities are shared and global, but what you can see and say about them is yours.

Two planes

The insight that makes a shared name safe is that identity and view are different planes. The identity plane is global and sparse: one name per thing, discovered through resolution, through the identifiers the world already assigns, and through re-identification. The view plane is private and rich: your layers, your disclosed precision, your attached content, all keyed to the shared names so your view stays consistent with everyone else about where things are, while differing on what you have added.

This is the same separation that made the web work. Everyone resolves the same domain to the same server; what each person then sees, caches, and presents is their own. Nobody imagines DNS forces one shared website. A spatial identity layer does the same for physical and virtual things: it makes systems agree on which thing, and leaves everything else to them.

How two strangers agree

Because anyone can mint a name without permission, two systems can independently name the same door. A shared world works only if they converge. Three mechanisms make that happen, in order of how much weight they carry: rendezvous through identifiers the world already assigns (a parcel number, an asset tag, a place id); spatial dedupe within a grid zone; and re-identification, under which independently minted names for one thing are merged into one.

The missing layer was never more geometry. It was the name that lets the geometry we already have be used together. See it happen on the globe →