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.

Who built it

Mikyö Clark (they/them). 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 run Compression, a firm for organizational redesign around agentic AI. The longer version is at mikyo.one. The argument behind this index is in The Network That Wrote Its Org Chart in Code.

I hold no position in any subnet token and have not been paid by any subnet, validator, or foundation for this work. If that changes it will be stated here.

How to use it

Sort the index 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 score is wrong, the methodology page says how to fix it.

Cite

Subnet Legibility Index, version 1.0, Mikyö Clark, legible.network, 2026. Data is licensed 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.