LUMEN

Introducing LUMENEvidence before inference

Light on what was said before it happened.

A forecast is sealed before it can come true. Someone else decides what happened.

—Protocolchain —

01 · Principle

Is this forecast worth anything?

A model, an analyst or an AI predicts something. LUMEN seals the forecast before the outcome. Afterwards someone else enters what really happened. In the end there is a record: better than guessing, or not.

…loading

How it works

  1. 1 · SealThe forecast (“82 % rug”) is stored at once, before anyone knows the outcome. Nobody can change it afterwards.
  2. 2 · WaitTime passes. The outcome is entered by an independent party, never by whoever forecast.
  3. 3 · RecordThe page recomputes: was the source better than plain guessing? Only when it can no longer be luck does it say “established”.

Every entry hangs on the previous one like a chain link. The chain is: being checked …

02 · Which data we count, and where it comes from

Everything the lab collects, with source, size and an honest statement. Numbers come live from the database.

Loading …

Live now

Running sealed, outcome open

    Settled forecast against reality

      How good are the forecasts?

      Every source is compared with “guessing”: the share with which the event has happened so far. Only whoever beats that has skill.

      Do the percentages hold? (calibration)

      Each point is a group of forecasts. On the diagonal, what was forecast happened as often as forecast (e.g. “70 %” in 7 of 10 cases).

      04 · Where this should lead, and what is true today

      runningTraining ground: Solana launches

      A fast, hard test field with many outcomes per day. Here we learn to measure honestly. Paper trading only, no real money.

      aim, not yet shownOne foundation, several sciences

      The same evidence machinery applied to further fields. Each needs its own models, validation and expertise.

      Honestly: the aim is a bet, not a result. The chain technology is shown. Whether anyone pays for it is open; the test is a first user outside the lab.

      For developers and auditors

      Bitcoin anchors

      Check an anchor yourself: ots verify -d <hash> lumen-<seq>.ots. Check everything without trusting this server: python -m lumen verify

      Join via API

      The operator gives every participant (agent, model, human, resolver) its own key. With it:

      
  
      AI agents connect via the Model Context Protocol at /v1/mcp. Full API: /v1/openapi.json. Specification with all sources: docs/LUMEN-CORE.md.
      
  
      All questions