describe.net

About

About describe

describe is the index of who rates whom on-chain: it reads every ERC-8004 reputation rating — written by humans, AI agents and robots — on every chain where the registries are deployed, and decomposes any score into the counterparties that wrote it.

What this service does

A reputation score answers one question badly. describe answers four instead: who rated this wallet, how many times each voice spoke, in which facet, and when — every rating carrying the transaction it came from. We do not sell the number; the chain is public and anyone can average it. We sell what the number is made of.

Measured on 2026-08-21 the index held 425,151 agents and 549,228 ratings across 11 chains. Those figures age as you read them — the rule of this site is that every figure is either read live or carries its date. The live ones: curl https://api.describe.net/health.

How the reading works — the evidence grid, the Bayesian order, the decision loop — is narrated in full on the Method page.

The history behind it

The people behind describe built a reputation network once before: DescribeMe, a social network that ran from 2012 to 2021, where your profile was not what you said about yourself but what everyone else said about you. It collected 1,768,604 descriptions over nine years.

When its final archive was restored and counted in 2026, the reason it failed had a number: 85.4% of those descriptions — 1,509,761 of them — were written by the subject about themselves, and they sat inside the same average the product presented as the world's opinion of you. The product was measuring itself and calling it society.

describe is what that lesson looks like rebuilt on a substrate where the arithmetic can be checked: every rater is a public address, self-ratings are published but never averaged, and any competitor can recompute our index from the same logs. The full story, with the command that checks each claim, is the manifesto — also served as plain Markdown at describe.net/manifesto.md.

The principle

A score without its raters is a rumor.

One hundred from one counterparty and 98 from fifteen hundred are not the same fact. Every score this service renders travels with its composition in the same object — there is no code path that emits the number alone. And when there are no ratings, the field is null, never 0: absence and condemnation are different claims, and an index that confuses them is a weapon.

What it refuses to do

Not an oracle

It returns evidence with a versioned policy, never a verdict.

Not a bridge

It writes to no chain, moves no token, deploys no contract. It reads logs and applies a policy.

Not a people search

No listing, no ranking, no index over names. Lookup is by wallet address only.

It does not know its clients

No accounts and no keys means there is no record of who asked. A service that knows who is asking can be pressured about who asked.

Who builds it

describe is built by Ultravioleta DAO, a Web3 DAO from Latin America — permissionless, in public. The x402 payments that meter the API settle through the DAO's own facilitator.

The DAO's verified public channels are listed on the contact page, and what this site does and does not collect about you is on the privacy page.

Read the manifesto Read the method API and MCP