Attested is a consensus-grounded knowledge graph. It doesn't map objective truth — it evaluates statements by their justification. Every verdict's reasoning is visible end to end, and bottoms out in a handful of stated commitments you can inspect and argue with.
Reasoning has to start somewhere. Ours starts here — proposed by no person, and the one commitment we've agreed not to reopen:
Human suffering is bad; human thriving is good.
It's an ordinary statement evaluated in the root context — you can see it on the graph and trace its dependencies like any other.
Every context, definition, and verdict exists because humans find it useful for navigating toward thriving and away from suffering.
We try to be precise about what this does and doesn't claim. Verdicts aren't deduced from the axiom the way a theorem follows from its premises — reasoning also rests on a few other stated commitments that sit alongside it, like "reliable tools for predicting the world reduce suffering." What we guarantee is that those commitments are few, named, visible, and — apart from the axiom — open to challenge. Claiming more than that would recreate the unearned authority this project exists to replace.
You don't vote on whether a statement is true. You vote on the condition — the justification that backs it. The verdict follows from the condition's evaluation, not from raw opinion.
A condition is a formula that references other markets, statement contexts, or governed parameters. When its dependencies resolve, the condition evaluates to a verdict: affirmed, denied, or undetermined. If you disagree with a verdict, you propose a replacement condition — you don't fight the claim directly.
This structure means corrections cascade: overturn a dependency and every downstream verdict goes stale for re-evaluation. The graph is honestly correctable, not frozen.
When you vote on a market, you don't see the running tally. The result is revealed only at resolution.
This is deliberate: a visible tally is a herding anchor. Sealed tallies mean each participant evaluates on their own reasoning, not on which way the crowd is leaning. The threshold (stake pool + margin) is frozen at open time so the bar can't move during the vote.
Attested maximizes the odds of good-faith consensus; it does not guarantee truth. The design rests on one bedrock premise: a majority of effective vote-weight is reasonable and engages in good faith. Nothing here prevents a bad-faith supermajority from defeating the system.
A graph that can identify exactly what it doesn't know should be able to go and find out.
Attested already surfaces questions that no existing source settles — they sit as undetermined, with the missing piece named. The intent is to let the community prioritise those questions and fund independent research to answer them, with the result entering the graph as an ordinary, checkable claim. That turns "we don't know" from a dead end into a queue.
Funding research creates an obvious hazard — a funder with a stake in the answer is the exact failure mode Attested exists to route around. So the rules come first: the question and the method are agreed and published before anyone is engaged; researchers are paid the same for an unwelcome or null result; method, data, and result are published in full; and a funded finding gets no special standing — it faces the same market as any other claim. Whoever votes to fund does not decide whether the finding is adopted.
This is intent, not a feature: it needs revenue Attested doesn't yet have. It's stated here because it's the honest answer to the limitation above, and because a plan you can hold us to is worth more than one we keep to ourselves.