HTML · Canvas · Genre topology · 41 territories · 45 lineage ridges
Tier I · Live Instrument
Music genre as geography, not taxonomy.
A continuous landscape rendered from forty-one genre territories and the lineage ridges that connect them. Each peak is a chromosome territory — a genre — and its height is chromosome density, how many songs cluster there. Ridges between peaks are lineage paths, evolutionary routes along which genetic material flowed. Valleys are transition zones where genre boundaries gradient into each other. No walls. Only crossfades.
Hover the terrain to read territory and density. Open full-screen →
The standard model treats genres as discrete labels. A song is rock or it is jazz or it is electronic. The labels are mutually exclusive, the boundaries are sharp, and the relationship between two genres is either inheritance ("metal came from rock") or unrelatedness. This is the taxonomic tree applied to sound.
Genres aren't separate species. They are persistent configurations of which conventions are expressed and which are suppressed — the same chromosome architecture, expressed differently in different cell types. The suppression map is the karyotype. The expression profile is the phenotype. The lineage isn't a tree; it's a topology of shared and divergent suppression vectors. Two genres are close when they ignore the same things.
The terrain visualises this. Peak height encodes how distinct a genre's suppression profile is — how loudly and how often it says not this. The ridges that connect peaks encode the strength of the lineage, the magnitude of shared regulatory architecture. A song in a transition zone is not misclassified. It's at a saddle point — half the suppression profile is one genre, half is another. The body hears it as both.
The horizontal axis is the acoustic-to-electronic gradient. The vertical axis is the energy-to-density gradient — low and sparse at the top, high and dense at the bottom. These aren't the only meaningful axes for music; they're the two that organise the largest number of family resemblances in the current registry without forcing collisions. Folk, Country, Blues, and Gospel sit close on the left. House, Techno, and Trance sit close on the right. Pop is the massive central peak because it's the gravitational well — most other genres exchange genetic material with it. Hyperpop sits between House and Pop because that's structurally where it lives.
The two-axis projection is lossy. A three-dimensional space (adding, say, harmonic complexity, or vocal-presence ratio, or compression-vector position) would capture more. The current view is the planar shadow of a higher-dimensional terrain. Future versions will let you rotate it. The current version is a topographic map you can read at a glance.
Forty-five lineage ridges, weighted by strength. Strong ridges (Blues → Rock, Jungle → DnB, Disco → House) carry near-complete genetic material — most of one genre's regulatory architecture survives into the next. Weaker ridges (Classical → Post-Rock, Synth-Pop → House) are partial transfers: shared suppression of certain elements but divergent expression of others. The same ridge can be a highway in one direction and a footpath in the other — Blues → Rock is dense; Rock → Blues is rare.
This is where the karyotype frame earns its keep. In a tree model, lineage is a single inheritance arrow. In a terrain model, lineage is a saddle: high in the middle, rising into both peaks. A song that sits on a saddle is half of one and half of the other, but the saddle itself is a coherent location. There are songs that live on saddles. They aren't transitional artefacts. They're a third territory the discrete model has no name for.
Discrete-genre classifiers report saddle-dwellers as misclassifications. The terrain reports them as residents. The same song, in two models, with two opposite truths about where it lives.
The palette runs from deep purple in the valleys, through plum and rose along the slopes, to amber-gold at the peaks. Contour lines mark elevation thresholds so the eye can read shape without colour alone — the highest peaks have the most concentric rings around them. Pop and Rock and Hip-Hop and House are the four most prominent peaks. They are also, not coincidentally, the four genres most other genres ridge into.
The terrain is a sibling view of the Sonic Phenomenology framework. The same suppression vectors that drive the engine's attention budget are what determine genre proximity here. When a new song is fingerprinted and its suppression vector falls between two existing genre centres, the terrain places it on the appropriate saddle. The visualisation is the analysis output, made navigable.
What's plotted here is the 2026 state of the registry — forty-one territories, twenty baselines fully characterised, eleven songs anchoring the geometry. As more songs are analysed, the peaks grow taller, new ridges form, and previously invisible saddles emerge. The terrain is a living map. This snapshot is the surface as of May 2026.
The visualisation responds to hover — read off the territory name, the density at the cursor, and the two nearest genres when you're in a transition zone. The full-screen view lives at /tools/karyotype-terrain/map.html. The source is in the atelier repository — a single self-contained HTML file with the genre coordinates, ridge weights, and terrain renderer inline.