How the Augur dispute resolution mechanism is facing its biggest test yet
Something interesting is happening over on Augur right now. There is an Augur market that, due to its confusing language, is currently in a shoot-out dispute round which may end up pushing the system to its limit. Given enough capital and patience, a few participants with a lot of REP may be able to push the market to a resolution that is not accepted by the majority of observers. Nick Tomaino covers the question in detail. Here’s a quick summary.
Market Resolution Refresher
Let’s refer back to the Augur whitepaper for how markets resolve.
- Once a market closes, the amount of ETH locked in the market remains constant, though the rights to those shares may be traded over time.
- The market is then sent to the dispute resolution rounds, where REP holders can stake towards outcome that they deem truthful.
- Let A(n) = the total stake over all of this market’s outcomes at the beginning of dispute round n.
- Let ω = any market outcome other than the market’s tentative outcome at the beginning of this dispute round n.
- Let S(ω, n) = the total amount of stake on outcome ω at the beginning of dispute round n.
- Let B(ω, n) = the size of the dispute bond needed to successfully dispute the current tentative outcome in favor of the new outcome ω during round n.
- Each dispute round has a window of ~7 days.
- The round is resolved when consecutive dispute rounds produce the same result (in other words, the opposing outcome does not meet the minimum bond requirement).
- The protocol pays out the REP staked to the losing outcome to the stakers of the winning outcome, pro rata to their REP stake.
- The protocol recognizes the outcome as final and pays out the underlying ETH accordingly to the shareholders.
Let’s go through a basic example where we calculate the REP and ETH payout for a particular Augur market dispute. Say there is ~$100 in open interest, with Outcome “X” holding 80% of the market. In this dispute, 10 REP have been staked by Outcome A. The disputes go through several rounds until Round 5, where the minimum bond required to flip the tentative outcome is not filled.
- According to the formula, the calculates such that the winning outcome should get ~50% ROI (in REP), as well as the upside of the open interest on the underlying market (in ETH).
- REP Payout: 80 REP is paid out to the stakers for Outcome “X”, pro rata.
- ETH Payout: market settles for outcome “X” and underlying ETH shares are paid out accordingly (~20% of the $100, pro rata).
How long can these disputes keep flipping? There is a special last case resort known as the Fork Phase, where all markets freeze and the entire universe of REP holders must vote on the market in question to determine the “correct” outcome.
- The Fork State initiates once the dispute bond size exceeds 275K REP (2.5% of all REP in existence).
- For the market in question, each outcome creates alternative “child” universes, where REP holders can migrate their tokens towards the outcomes they deem true.
- The child universe that receives the most REP is considered the “Winning Universe” and its corresponding outcome it deemed the true outcome— all existing markets can only port to this universe.
- Because the REP tokens may only be used for dispute resolution within the universes they reside within, the REP migrated to “losing” universes are effectively non-fungible with the “winning” universe
- The backers of the winning outcome earn the (i) market’s underlying ETH (ii) market’s opposing REP staked and (iii) the price effect of “the losing” universe’s REP tokens being effectively taken out of circulation
To The Limit
The House market is a fun one to observe because of how the dispute economics have been revised to deal with situations of asymmetric upside in the underlying market.
There is currently ~$700K USD in ETH open interest for this market, with 96% for Dems, 3% for Republicans, and 1% Tied.
If the market were to resolve in favor of the Republicans, the 3% who own the Republican shares would earn the remaining 97% of ETH in open interest ( ~$679K); conversely, there is a relatively limited 4% of ETH upside for resolving Democrats (~$28K.)
One could expect that Republican shareholders are therefore more inclined to aggressively stake REP for the Republican outcome. This introduces the intended game-theoretical dynamic — market observers would prepare for such behavior and realize that despite limited upside for the market to resolve “Democrat,” there may be interesting REP returns by outlasting the Republican REP tide.
Below is the projected path for the market resolution assuming disputes continue going back and forth. The market reporter put down an initial stake for the “Republican” outcome (Round 0). Since then, the outcomes have been flipping consistently, with a cumulative 728 REP staked so far (~$7,280 at current REP prices). The subsequent rounds are merely illustrative, and a Fork State would initiate at Round 14 given that the dispute bond filled is greater than 275K REP.
Because the monetary consensus implies the Democratic outcome, a crypto-trader could keep filling the bond for the Democratic outcome with expectation of ~50% ROI in REP terms. Extending the idea of “Keepers”, lenders may take out their deposits on products like Compound to maximize their returns (50% vs <2%) by taking advantage of this opportunity. In mature crypto markets with many networks and participants, this might exhibit some form of local equilibria.
Though forking should generally be avoided, it is entirely possible that this market could be taken to that end-game scenario. The variables at play for forking would likely hinge on the expected behavior of exchanges — specifically which child universe they would migrate their REP towards.
Some might make a bet that most exchanges may not participate in the migration vote, out of risk perception or general apathy. Others might grossly underestimate the role of exchanges in determining on-chain decisions, which would bolster the idea that crypto capital operators exert active network influence especially in capital-based security models of the near future (Proof of Stake.)
Of course, as this process unfolds, a savvy trader could play the underlying market (volume is high during the resolution periods) as the disputes go deeper and closer to the potential fork state.
Though most of this analysis is speculative, it is definitely one to watch in the first quarter of 2019. The dispute resolution is currently in Round 4, which you can track in real time. I predict that the dispute resolution mechanism will prove itself to be robust enough, and the market will probably does not reach the ultimate fork state and resolves itself before the REP stake becomes exorbitantly expensive (in the six-figures.)