The shape you didn’t decide

Below is a reading of one of my own tracks that I could not have produced by listening harder.

Not because the reading is clever — because it depends on something outside the track. A map of how my whole catalogue relates to itself, written before that song was finished. With the map, the song says one thing. Without it, it says something considerably smaller.

That map is what I’ve been calling a mythology layer. This is what it is, what it found in a bass single called Dope Ways, and why I think it’s the missing input for anyone trying to make generative tools produce something that sounds like them.

You already have one

The claim isn’t that you have themes. Everyone knows their themes — themes are what you’d list if asked. The claim is that underneath the themes you meant, there’s a structure you didn’t decide: things that recur across years and formats without your choosing them, related to each other in ways you’d never have described.

In my case one of those turned out to be a function rather than a subject. Nearly everything I make is organised around passage through darkness rather than escape from it. Not darkness as a topic — darkness as a route. It shows up in songs written a decade apart, in a game, in how I name things. I never decided that. It was simply true, and had been for a long time before anyone wrote it down.

Once it was written down, it stopped being a description and became a test.

The instrument, on one track

Dope Ways is a bass track with Squirrelybass that opens a three-single run. On the surface it’s a monster song — swamp creature, cosmic horror, a lot of very confident threats. It’s funny. It’s meant to be.

On the left is the whole track, with the annotated passages marked in place. On the right is the part usually left implicit: the actual graph the reading resolves against, as the vortex it is. Select a line and you travel to the entries it lands on; drag to look around the structure, or step through the resonance points in order. That lookup is the whole mechanism — the reading isn’t interpretation, it’s a query against a structure that already existed.

The readings

Each has the two fields the layer returns for any material: what it feels like from inside, and what it does in the system.

A note on how that diagram looks, because it is the shortest available demonstration of the argument. A relation graph would normally be drawn the way every tool draws one: bright nodes, even lighting, everything legible at once. This one isn’t. It sits on a dark field, most of it stays in shadow, and selecting a line carries a small light to the region that line resolves against — nodes just outside the reach go dim rather than dark, and the rest you have to trust are there.

That wasn’t an aesthetic preference. The graph’s own hub node is passage through darkness with a light you carry, and the mythology’s oldest recurring object is a lantern. A map of this particular body of work cannot honestly be evenly lit — even illumination would contradict the thing being mapped. The plainest mark on the whole diagram is the node called ordinary, for the same reason: the canon defines its namesake animal as ordinary until it decides to matter.

The shape is derived too, and this is the part I’d point a sceptic at first. The entries are strung on a vortex whose radius contracts to a throat and opens again — and the node sitting at the throat is safe passage. I didn’t put it there for the metaphor. Safe passage carries five of the graph’s nine relations, so it is the narrowest point by measurement, and the layout contracts toward whatever the topology routes through. That the most-connected node in a mythology about going through things turns out to be the place the shape narrows is not a flourish. It fell out of counting edges.

Any generic graph tool would have given me circles on a plane. The layer told me what the picture had to be, down to where it narrows — and that is the difference I am trying to describe.

Read straight through, that’s a monster brag with a redemption turn. A good one, but nothing unusual.

Read against the graph, it’s the whole dissolution-and-return sequence in miniature: hypervigilance, a sleeping thing that must not wake, escalation to cosmic scale, fire that purifies rather than kills, the thesis stated four times, a collapse from astral horror down to a kitchen, sunrise, then a plain declaration of self-determination landing on the most ordinary word available.

I did not write it that way on purpose. That’s the entire point — the graph found the shape, and the shape was already load-bearing.

What the graph then produced

The reason this is more than self-flattery is that the graph became an input to systems that make things.

Two recurring characters in our video work were derived rather than designed. One is a hooded figure carrying a lantern — straight from the passage function, made literal: the light you carry is small, and you go anyway. The other is a near-black, iridescent, entirely deadpan performer, from the founding emotional promise of the project — slightly disturbed, yet soothing — and the electronic-goth lineage the music descends from. Neither is a mascot somebody sketched. Each traces to a specific canon entry, and the trace is checkable by someone who isn’t me.

The rules they carry come from the same place: the lantern is never absent from frame; its amber is the only warm light in any shot that figure appears in; the other character’s purple-teal sheen never transfers to him. Those read as art direction. They’re inherited constraints.

The part that matters most: it says no

It’s also the honest answer to why the free instrument stops where it does. It walks you to the graph and the verdict and then stops on purpose. Everything past that — the forces organising the graph, what the recurring figures actually do, the structures they form — is the full instrument, and it’s paid. I’d rather say that here than have you discover it.

What I know, and what I don’t

• The evidence is n=1. I built the instrument and I’ve run it on my own catalogue. That is exactly as weak as it sounds, and it’s the reason for this piece.
• It has never been properly road-tested by an artist working alone, on material I know nothing about, with no shared context to lean on.
• It needs a frontier-tier model. On weaker models the interview degrades into agreement, which produces a graph that flatters and therefore can’t reject anything.
• I don’t know whether the structure generalises, or whether I’ve built a very precise instrument for reading exactly one person. Finding out is the only reason to hand it to anyone.