About

Why this exists

Bittensor wrote its organization in code. A subnet's incentive mechanism says what work is valued, who judges it, and how the reward flows, and it enforces all three without a manager. Since every subnet earns emission in proportion to what the market thinks it is worth, every subnet now has to explain itself to the people who fund it, work for it, buy from it, and encounter it. Most of that explaining is nobody's job.

This index measures the gap. Not the quality of any mechanism, which the market can judge, but whether a first-time reader can find out what a subnet is, who it is for, how it holds up, and whether its front door is open, in five minutes, from the subnet's own public materials. A network that allocates on perception should be able to see what there is to perceive.

The voices are the second instrument. Perception is made of words, and the words come from people: the founders, the foundation, the investors, the analysts, and the subnet teams, each with a way of saying what Bittensor is. The map records those words verbatim, dated and sourced, reads each narrator through one framework so they can be compared, and traces every metaphor from its first use to everyone who picked it up. It is a genealogy of the network's self-description, by voice.

Who built it

Mikyö Clark. I built the community function at EigenLayer from zero through its token launch, ran go-to-market for Ritual, an AI-native Layer 1, and for Inversed, an agent governance layer, and now advise companies on organizational redesign around agentic AI via Compression. The longer version is at mikyo.one. The argument behind this index is on the home page.

How to use it

On the map, pick a narrator and read them by era, or pick a metaphor and follow it across narrators. The framework view lays narrators side by side, slot by slot. Every statement carries its source and its confidence, and unverified statements are hidden until you switch them on. On the index, sort by an audience and read the deltas: a subnet that moves a long way between its emission rank and its buyer-legibility rank is telling you something about who it has been talking to. Open a subnet page for the evidence behind every score. Every subnet page opens with a plain-language explainer, and dotted terms open a definition where you are. If a rating or a quote is wrong, the method page says how to fix it.

Cite

Legible, version 2.0, Mikyö Clark, legible.network, 2026. The index (scale version 1.0) and the map are both CC BY 4.0; the code is MIT.

Contact

m@cmprssn.xyz, or an issue on the repository. I read everything; corrections get answered first.