Augur is the oldest name on this list and the one that least resembles a venue. There is no application to open, no market to trade and no balance to deposit: the project's own FAQ answers “Can I use Augur today?” with “Not as a new Augur-native prediction-market application: one has not yet relaunched.” What augur.net offers is project documentation, an archived fork record and a monitor, and what the protocol offers is a resolution mechanism that spent 2026 proving itself in public.
That makes Augur a different kind of entry from the licensed exchanges and the live on-chain venues on this list. It is worth reading anyway, because almost every other venue's biggest unresolved problem — who decides what actually happened when real money is riding on the answer — is the only thing Augur has ever been about.
What Augur is today
The project is mid-reboot. The Lituus Foundation began rebuilding Augur from the ground up on 31 March 2025, and it now funds two parallel teams deliberately working different angles on the same problem: Lituus Labs, on oracle architecture, cryptoeconomic security and protocol design, and Dark Florist, on alternative oracle structures, user-facing applications and incentive models. The Foundation's stated endpoint is threefold: finalise a new oracle architecture with academic-style documentation, widen REP access across more exchanges and L2s, and relaunch an Augur-native market protocol once oracle security and fee-routing are settled.
So the honest status is neither “dormant” nor “live”. The legacy v2 protocol is deployed on Ethereum mainnet and its dispute machinery ran a real fork this year; the application layer that would let you use it is still being designed.
A protocol from Ethereum's early days
Augur ran one of Ethereum's first crowdsales in 2015 and launched on mainnet in 2018, which makes it a genuinely different artefact from everything else here: it predates the licensed event exchanges by years and the on-chain venues by a decade of tooling. Its founding position, stated in its own account of the reboot, was radical then and is still rare — “no company or committee should hold the final word on who won”, with disputed outcomes settled through an open economic process in which participants stake value on competing answers and lose it when they back the wrong one.
Stewardship has changed hands more than once since. The work is currently funded and coordinated by the Lituus Foundation, which writes the project's updates in the first person. That matters for anyone judging the token: the same organisation holds treasury REP and is the primary source for what the project is doing.
REP is a dispute token, not money
The single most misunderstood thing about Augur is the role of REP. It is the reputation token used in reporting and disputes, and the protocol glossary is explicit that it “is not the currency used to trade market shares”. Holding REP is a claim on the obligation to resolve markets correctly, not a share in trading activity, and the project treats it that way: “REP is not a passive asset and every REP holder has a duty to show up when called.”
Protocol thresholds are measured against theoretical REP, the supply basis the contracts use for percentage tests. The genesis-universe reference is 11,000,000 REP, which makes 2.5% equal to 275,000 REP — but the site is careful to say that a live deployment's own configured value is the authoritative one, not the widely quoted genesis number.
How a v2 market resolves, and when it escalates to a fork
Resolution starts quietly. After an event ends, a designated reporter has an initial reporting window; a report makes that outcome tentative. Any REP holder can then challenge it by staking REP on a different outcome. The first dispute round can run up to 24 hours, and later rounds up to seven days.
The amount required to dispute is not a fixed multiple of the last bond, which is a detail worth understanding before assuming a dispute is cheap or expensive. For dispute round n and outcome w the bond is:
B(w, n) = 2An − 3S(w, n)
where A is the total stake across all outcomes at the start of the round and S is the stake already on the proposed outcome. Bonds are crowdsourced, so several participants can fill one together, and the target is engineered to pay successful disputers about a 40% return drawn from the losing side.
| Filled dispute bond | What the protocol does next |
|---|---|
| Under 0.02% of theoretical REP | The new tentative outcome enters another dispute round immediately. |
| 0.02% up to 2.5% | The new tentative outcome waits out a window before another dispute round. |
| 2.5% or more | The market enters the fork state. This is the backstop. |
Most disagreements never get that far. When a market finalises without a fork, stake on the incorrect outcome is forfeited: the protocol burns 20% of it and redistributes the rest to the REP staked on the final outcome.
The fork: one universe per outcome, one way
A fork is Augur's last resort, and it behaves like an election in which only people with money at risk can vote. The protocol splits into a child universe for each possible outcome of the disputed market — including Invalid — and every REP holder must migrate their tokens into the universe they believe reflects reality. A child universe is created lazily, only when migration first targets that outcome, so an outcome nobody backed may never have a contract at all.
While the fork runs, the parent universe is locked: no new markets, no new REP staking, and disputes in other unfinalised markets are put on hold, though those markets can keep trading and simply cannot finalise. Migration is one-way — a holder's choice cannot be reversed or moved to a sibling universe — and the reference mechanics allow a forking period of up to 60 days. The child that receives the most migrated REP when the period ends becomes the winning universe, and its outcome becomes the forking market's final answer. That result cannot be disputed through the ordinary process.
The Moon Fork, in numbers
All of this was theory until 2026. A community member, Micah Zoltu, raised a crowdsourcer specifically to drive a dispute over the market “Will the Artemis II Mission successfully liftoff in the first week of April?” all the way to the fork backstop, with existing REP holders contributing roughly 200,000 REP they knew would be spent, so that the mechanism could be watched end to end. The mission did launch; what was being tested was the machinery, not the answer.
| Moon Fork record | Value |
|---|---|
| Escalation began | 8 April 2026 |
| Escalation game | April to early June 2026 — stakes double each round; threshold 2.5% of REP supply; roughly 40% paid to winning stakers |
| Migration window | Early June to 3 August 2026, 01:00 UTC — closed permanently |
| Eligible supply | 11,000,000 REP |
| Total migrated | 6,547,546.66 REP — 59.52% of supply |
| To the Yes universe | 6,545,760.43 REP (winning universe, 0x2811…2A9f) |
| To the No universe | 1,786.23 REP |
| Invalid universe | Never created — no REP migrated to it |
| Independently observed | Ethereum block 25,677,103, read 3 August 2026 at 21:36:23 UTC by two separate RPC providers with the same result |
| Current REP | REPv2_Yes_1 at 0xCf6A0A7826fa124B7705d6f3c675eAD76f1e540D |
What the fork was really testing
Augur's own account of the result is worth taking seriously because it is unusually unsentimental. The mechanism worked — a system that had only ever existed as a threat carried a real dispute to a settled outcome, and there was no switch anyone could pull to force the result. But the more interesting claim is about why the fork helps security at all.
Augur's argument is that a protocol's defence depends on holders actually showing up to dispute bad outcomes, and that part of any supply never does — tokens sit in lost wallets or with people who stopped watching. That REP counts toward the security budget while contributing nothing to it, so the oracle looks better defended than it is. The fork forces every holder to act or be left behind, and around 40% of supply never moved. What came through is a smaller pool, but all of it belongs to holders who showed up.
The mechanics cut the other way too: an attacker who pushes a dispute to a fork ends up holding tokens in the universe that does not reflect reality, and as honest holders migrate away from it, the economic value of the system moves with them. That is the whole design — truth enforced by capital, not by an authority.
Why the project says resolution is the weak point
The reboot's central argument is that this problem is now demonstrated rather than hypothetical, and Augur makes the case using three episodes at the largest venue, chosen precisely because volume concentrates there. In July 2025 a market on whether Ukraine's president would wear a suit, carrying over $150 million in volume, flipped from Yes to No on an oracle token-holder vote. In May 2026 a Wall Street Journal review of that venue's disputes found that in nearly one case in five, accounts helping decide an outcome were tied to bets in the market they were ruling on. In June 2026 a market worth $14 million fought over whether a sale date or a disclosure date controlled the payout, and the ruling has since been challenged in court by traders on the losing side. Different markets, same weak point.
It is fair to read all three as Augur's own framing, and it is also fair to note that Augur is not neutral about the answer: its conclusion is that a platform should not have to act as judge every time a contentious outcome lands on an edge case.
Augur Lituus: the next design
The reboot's technical output is Augur Lituus, published as a whitepaper on 29 January 2026 and now in development with ChainSafe engineering the implementation. The design separates the oracle from prediction markets entirely: a decentralised resolution layer that markets, DeFi protocols and any application can call when a real-world outcome is disputed, with results bridgeable to any EVM chain. Augur's own framing is that it is meant to sit underneath platforms rather than compete with them — an admission, in effect, that the venue and the judge should be different things.
The whitepaper's headline claim is quantitative: under its stated assumptions, forcing a fraudulent outcome through Lituus would cost about 134% of the oracle's fully diluted valuation, against roughly 92% under Augur v2's mechanism — meaning an attacker would have to buy up and destroy more than half the token supply, then buy into the replacement supply a second time. The assumptions are published alongside the result so they can be challenged. Separately, the Foundation funds Dark Florist's own line of work on a deliberately non-finalising oracle, which is best read as a parallel project rather than a component of Lituus.
Who it suits today
Nobody looking for somewhere to trade. The honest version: Augur suits researchers and protocol historians who want to study what a completed algorithmic fork actually looks like, and any REP holder who was affected by the Moon Fork and needs to establish what they now hold — which is REPv2_Yes_1 if they migrated, or a position in the parent universe expected to be worthless if they did not. For anyone comparing prediction markets to put money to work, the live on-chain venues such as Zeitgeist and the licensed exchanges are the ones with something to log into; Augur's next market app is gated on an oracle architecture that is still being engineered.
Two practical warnings come straight from the site. Migration is permanently closed and cannot be reopened or extended — nobody can migrate on your behalf, and anyone offering to migrate, recover or unlock REP is running a scam. And the post-fork token is new enough that wallets and exchanges label it inconsistently, so before acting on anything presented as REP, match the contract address: Kraken lists REPv2_Yes_1 under the ticker AUGUR, while its on-chain symbol is unchanged.





































