Only coverage that’s proven to publish.
Herald is the Bittensor subnet that turns real press into a verifiable commodity. Miners place articles in real outlets; a code-only oracle proves each one is live, editorial, indexed and built to last. Emissions pay for earned coverage — never for paid placements dressed as news.
The mechanism is built around journalism it must never corrupt. These hold no matter who is mining.
Herald rewards earned coverage, never paid coverage. The reward goes to the miner who caused the article — never to the newsroom that ran it.
Sponsored posts, press releases and paid syndication are detected and scored as exactly what they are. They are never counted as real news.
The owner may run a miner to keep the standing brief alive. It is scored exactly like everyone else. No advantage, ever.
The public ledger of verified coverage. On the live network, every record opens the snapshot the oracle saved — the proof the article ran, on tier, as real news.
A funded brief becomes a portfolio of independently-verified placements — each one passed the oracle’s seven checks and is attributed on-chain. That’s the deliverable: proof you can hand to your board, not a screenshot.
Every validator runs the same code on each claimed article. Exact checks return one answer for everyone; judgement checks run inside agreement bands so weights never diverge.
↳ The validator runs these cheapest-first with early-exit; the two judgement checks use an optional reference LLM (off by default) inside agreement bands. Web fetch and search run in a shared time window with caching and a provider quorum, so location and rate limits can’t split the vote.
Ranked by validator weight — a blend of tier, real-news quality, indexing and 30-day persistence.
Six steps. The work is getting real coverage; the chain records who caused it, checks it by code, and pays only when it lasts. Tap a step.
An operator opens a coverage brief — tier, topic, key messages, reward pool and a time window — and funds it up front.
Two kinds run side by side: client briefs paid from a funded pool, and a standing brief that pays from emissions for general pro-Bittensor coverage whenever no client brief is open.
If the rule were simply first to send the URL wins, anyone could claim a story that ran on its own. Commit-reveal closes the gap.
A miner seals an intent — brief, outlet, hotkey, nonce — and stakes a bond before the article exists. When the story goes live, only that commitment can claim it. Cheating costs more than it can earn.
Every score is measured against a signed, versioned registry. Tier sets the multiplier. Disputes point to an exact list version.
Independent newsrooms with the widest reach and the strongest editorial authority.
Established titles with real editorial standards and strong, trusted domains.
Verified outlets with smaller reach — real coverage, lower authority.
You already know how to get press. Herald turns it into on-chain income — no GPUs, no servers.
Set the topic and the tier. Only verified, persistent coverage is ever rewarded.
Herald reuses Bitcast’s settlement shape. Rewards are priced in USD, converted into validator weight, and emitted as SN69 alpha under Yuma consensus. Client briefs pay from a funded pool; the standing brief pays from emissions, so the wire never goes quiet.
SN69 is a clean, freshly-registered subnet. We ship a lean pilot first, then layer in the hard parts.
Mechanism, consensus plan, registry schema, where brief data lives.
Briefs, commit & claim, exact-check oracle, scoring, weight-setting.
Full commit-reveal, judgement checks, vesting, bonds, slashing, disputes.
Registry model, signed versions, admin tooling, abuse defence.
Operator admin, brief board, public proof page, reporting, leaderboard.
Deploy to SN69, pilot-measurement tools, monitoring, handover.
No. Herald rewards earned editorial coverage. Press releases, sponsored posts and paid syndication are detected by the oracle and never scored as real news — that rule is not negotiable.