Dispute Resolution
Handling disputes and remediation in the Nosh network
A buyer needs to be guaranteed that when they pay for a service they will receive it. If service is not delivered, the provider must not get paid, and may be additionally penalized.
A dispute arises when the buyer and provider disagree on whether the service was provided, in which case we introduce two dispute resolution mechanisms.
Undisputed service
A buyer, makes a payment for the service they buy from provider . Out of this payment, will be taken as a protocol fee. The remaining goes as payment to the provider.
When the buyer makes the payment, the are kept locked in a contract until the service is provided, and the buyer confirms service was provided, by exchanging their pin. The are only forwarded to the provider, once the user pin is received.
In addition, The provider may have to lock an additional amount as a commitment to the buyer that the service will be provided, and if not this will get slashed.
The amount may depend on
- The reputation (graph value) of the provider (which can reduce the amount they need to stake as they have already built trust), so decreases with higher graph value
- The reputation of the buyer. A high reputation buyer is able to demand a higher level of guarantee so increases with higher graph value (note that we also need to define a graph value for buyers).
- The size of the order, should increase with the price of the order .
If the buyer does not give their pin, agreeing that the service was completed, they are returned their payment, or (to add a cost to submitting malicious claims), and the locked is burned. Alternatively, an amount equivalent to could be substracted from the graph value of the provider.
Disputed Service
A dispute arises when the buyer does not give their pin, but the provider files a claim that they did fulfill the service.
- In the case that the provider purchased insurance, the Automated Insurance Market Maker provides coverage of the lost collateral.
- In the case that the provider did not purchase insurance, the Statistical Court resolves the dispute.
Our design focuses on the following mechanisms for ensuring a healthy dispute resoluion process:
Consequences for Losing Party The losing party in a dispute can have the edge weights in their transaction graph slashed or burned, harming their future earning potential in the network. This mechanism reduces the frequency of false claims by economically rational actors.
Reputation Reputation of participants is stored in a series of on-chain attestations.
Reputation Incentives and Consequences Malicious producers who lose disputes frequently or perform lazily will acquire lower reputation scores, making them less likely to be discovered or prioritized in search requests. Service purveyors who renege on their commitments or provide sub-par services can face consequences such as reduced eigenvector centrality scores, higher insurance costs, and potential slashing of their edge weights in the transaction graph. Lazy providers who consistently fail to meet quality of service (QoS) guarantees will see their reputation scores and eigenvector centrality decline, leading to reduced visibility, higher transaction costs, or blacklisting.
Reputation-Based Insurance Fees When a Producer and a Buyer create an escrow contract, they can optionally choose to insure their transaction in the case of a dispute. The Automated Insurance Market Maker prices risk in a reputation-based model, where participants with a strong reputation and low dispute history tend to be eligible for a lower cost of insurance. Over time, good-faith actors are rewarded with lower fees. This incentivizes good behavior and reduces the likelihood of false positive claims.