IP addresses are logical entities — assigned by authorities, governed by institutions, limited by design. They tell computers how to find each other, but they know nothing about where things actually are.
The Earth already has 569 trillion unique points. Why invent addresses when geography gives them to us for free?
Welcome to GEP.
Click two places on Earth. Watch GEP route between them — no servers, no DNS, no lookup tables. Just math.
GEP uses Uber's H3 hexagonal grid to divide the Earth into cells at 16 resolutions (R0–R15). Each resolution is exactly 7× more precise than the previous — a mathematical property of hexagonal tessellation.
Where c(r) is the number of cells at resolution r. At R0 (continent scale), Earth has 122 cells. At R15 (sub-meter), it has 569 trillion.
| RES | CELLS | SCALE | CELL SIZE | ANALOGY |
|---|---|---|---|---|
| R0 | 122 | Continent | 1,107 km | Like a country code (+1, +39) |
| R5 | 2,016,842 | City | 8.5 km | Like an area code (212, 06) |
| R7 | 98,825,162 | Neighborhood | 1.22 km | Like a zip code |
| R10 | 33.9 billion | Street address | 65 m | Like a street number |
| R15 | 569 trillion | Sub-meter | 0.5 m | Like a desk in a room |
GEP routes packets between any two points on Earth by finding their Lowest Common Ancestor in the H3 hierarchy — the coarsest resolution where both points share the same parent cell.
The algorithm ascends from the sender's resolution to the common ancestor, then descends to the receiver. Zero external state. Zero routing tables. O(R) time complexity.
Each H3 cell's index is SHA-256 hashed to produce a permanent, collision-resistant address called a GEA (GeoEpoch Address):
These hashes are chained across resolutions into an Epoch Chain — a Merkle-like structure where each resolution's hash commits to all cells at that level. The genesis hash anchors the entire system:
Five innovations identified. Related to Provisional Patent #63/948,788.
github.com/GNS-Foundation/gep-core ·
Create a GEPSite →