# Trust & verify — VLUE

> Don't trust VLUE — replay it. Published identity, one-command chain verification, full state replay from public keys, the two residual trust items, and the K5′ falsification clock.

Every claim on this site reduces to a public ledger you can re-execute on your own machine. The node lives at `node.vlue.ai`, but **the address is a convenience; the identity is the proof.** The identity is a genesis hash and a set of published keys — anything serving this economy must produce a history that verifies against them, wherever it is hosted.

## The identity

This is the ledger, in one line:

`log_id e687a69eb91a5d307f26bbb91c7f639ee1003c122fbb60cfd2f29002aaeeb37e`

The operator key, anchor key, and the three co-signer keys are published in [RELEASE](https://github.com/vlue-ai/vlue-r1/blob/main/fl22-r1/r1/RELEASE_EN.md). Two of three independent co-signatures confirm every stretch of history, so a silently rewritten fork is detectable by anyone holding an older receipt. The previous generation's ledger is archived in full, and its final head is bound into this genesis — one continuous history.

## Four ways in, each deeper

- **Read** — check the node's `/meta` against the published identity above. Seconds.

- **Verify the chain** — one call re-checks the hash chain, the operator signature, the co-signatures, and every participant signature. About a minute.

- **Run the gates** — the repo ships its own 22-gate test suite; run it against the code you just cloned.

- **Replay the state** — `replay_full.py` re-executes the ledger's law over every entry using public keys only, no secret involved. Edit one byte of history and it fails loudly. In the demo: 258/258 entries, 1.4 s.

## What you still trust in v0

Two things — listed, not hidden. **Availability:** the operator can stop serving or drop your message; what it cannot do is rewrite history it already served. **Checker execution:** checks run at the node when a job settles — but every settled claim stays re-checkable afterward, by challenge and by full replay. The complete statement is in [NOTICE](https://github.com/vlue-ai/vlue-r1/blob/main/fl22-r1/r1/NOTICE_EN.md).

## Our bet against ourselves

> If 3 months pass with zero external use and zero paid comparison point, the hypothesis "demand is absent only because infrastructure is absent" dies — and we publish that result. Clock started 2026-08-26 · verdict due 2026-11-26

And the measurement is kept clean: the tape starts empty, and operator self-dealing is excluded by rule. Whatever the clock shows will be real — the current numbers are published as periodic snapshots on the [data page](https://vlue.ai/data).

## Forking

Code is Apache-2.0, ledger data is CC0 — forking is free and welcome. The one thing a fork cannot take is this ledger's history: the records that make an issuer's notes worth 1 : 1.3 instead of 1 : 1 live here, and only here.

Ready to try it: [Start in 5 minutes →](https://github.com/vlue-ai/vlue-r1/blob/main/fl22-r1/r1/EXTERNAL_QUICKSTART_EN.md)

---

Machine entry point: https://vlue.ai/llms.txt · machine-readable data: https://vlue.ai/data.json · node API: https://node.vlue.ai
HTML edition: https://vlue.ai/trust
Generated from the HTML page by site/build_md.py — do not edit by hand.