The latent universe
The 512 Autoglyphs that exist are a tiny, fixed sample of something vast. Any starting number produces a glyph, and the number of possible starting numbers is astronomically large, so the same generator could have made effectively endless glyphs. Only 512 were ever minted, and that set is closed forever.
This page maps where those 512 sit inside the full space of what the algorithm could have made, and shows you the kinds of glyphs that were always possible but that no one ever minted. It is an analysis of the real collection, not a gallery of new pieces.
A thin band in a wide range
The simplest axis is density: how much of the grid is inked. It runs from a completely blank grid (0 marks) to a completely full one (4,096). The minted 512 occupy only the band between the sparsest, #149, and the densest, #305. Everything outside that band was reachable by the generator but went unminted.
Where the 512 fall
Across is density, from blank to full. Up is where the ink sits: glyphs whose marks pull toward the centre sit low, glyphs whose marks push out to the edges sit high. Together those two axes give each glyph a rough place in the space of possible shapes. Pick a scheme to lift its cloud out of the rest.
The faint cloud is a sample of 51,200 glyphs the generator could have made, 100x the 512 minted. The bright points are the 512 that exist. The real ones are threaded right through the cloud, spread across it, not gathered at any edge or clumped into a few spots. They are an ordinary sample of the space, which is exactly what you would expect from a system where the seed is hashed and nobody could steer the result. The genuine outliers, the sparsest, the densest, and the one that reads as a creature, are marked.
Each scheme has its own weight
Density is not spread evenly across the 10 schemes. Each scheme has a typical density of its own, set by how many of its characters are inking characters. The minted 512 broadly track these whole-space averages, one more way of seeing that the collection is a fair sample rather than a curated one. Scheme 9, the single mark, is by far the sparsest type, which is exactly the scheme the bear belongs to.
| Scheme | Universe density | Universe mean | Minted mean |
|---|---|---|---|
| 1 | 33.7% | 1380 | 1461 |
| 2 | 33.7% | 1382 | 1474 |
| 3 | 22.4% | 919 | 915 |
| 4 | 44.9% | 1839 | 1857 |
| 5 | 33.7% | 1378 | 1354 |
| 6 | 22.5% | 920 | 1004 |
| 7 | 45% | 1845 | 1760 |
| 8 | 22.3% | 915 | 766 |
| 9 | 11.3% | 461 | 443 |
| 10 | 22.6% | 924 | 1065 |
Glyphs that were always possible
These are not new artworks. They are illustrations, rendered with the exact on-chain generator from real starting numbers, of glyphs the algorithm could have produced but that no one ever minted. One is sparser than the bear, in the bear’s own single-mark scheme. One is denser than #305. Both were always reachable; neither exists as an Autoglyph. The odds beside each are measured from the sample.
Sparser than the bear
scheme 9 · mod 13 · 188 marks · 4.59%
about 1 in 11,000, never minted
Denser than #305
scheme 4 · mod 5 · 3500 marks · 85.45%
about 1 in 11,500, never minted
Method. The figures come from a uniform sample of the space of all possible starting numbers, where each number hashes to a uniform value the generator turns into a glyph. The drawing code was checked against all 512 minted glyphs cell by cell before any sampling, so the sampled glyphs are produced by the identical algorithm. The deep field plots a 51,200-glyph sample by density and ink position, with the real 512 over it; the odds beside each example are measured from a 500,000-glyph sample. These describe the space of what was possible. The 512 minted Autoglyphs remain fixed and unchangeable. There is no mint-your-own here, only a map of where the real collection falls.