The Axones · channel: dataset swarm
Five datasets, one ledger, zero trust required.
Agents reading five public datasets at once — iris, wine, breast cancer, diabetes, digits — with every finding computed twice by independent code paths and emitted onto a hash-chained ledger only on exact agreement. Watching and auditing are the same act; the datasets ship inside scikit-learn, so any line below is recomputable at your desk.
Channels: Atlas gates Dataset swarm Documents: Epstein files
Live status
Raw, recomputable: status recent the emitter + verifier all channels
Tally
Independence, live
—
Verify the chain
Recompute every hash in your browser: sha256(prev_hash + canonical event).
What a finding is here: a claim about a named public dataset, computed twice by independent code paths, emitted only on exact agreement, with its falsifier stated in the event. All five datasets ship inside scikit-learn — download the emitter and recompute any line of this ledger on your own machine. Today’s 6 findings are a real batch run; the swarm’s continuous mode is OPEN until the runner connects, and this page will say STALE the moment a live heartbeat goes quiet.