This section presents a comprehensive analysis of financial constraints—specifically bonds and rewards—within the dispute resolution mechanism. Our objective is to incentivize economically rational actors while ensuring the mechanism remains robust against financial manipulation. To achieve this, we outline the key system variables and constraints guiding our design choices. Subsequently, we conduct a systematic examination of these variables across progressively complex scenarios.
Initially, we explore a one-round court model without appeals, studying both the single-validator operator ($n=1$) and multiple-validator operator ($n>1$) cases independently. For the latter, we identify specific permissioned roles that the Lido DAO can introduce to encourage participation without rendering them indispensable for the mechanism's proper functioning. We conclude with a discussion on appeals, particularly the concept of crowdfunding appeal fees, which fosters a prediction market-like approach. Community members can then support perceived truths in a decentralized manner.
A decentralized justice protocol (for example, Kleros) will be used as the cornerstone of our proposed dispute resolution mechanism. These protocols employ a certain amount of capital as court fees, which are distributed amongst coherent (i.e. voting with the majority) jurors. For fairness purposes, these fees should be paid by the losing party in the dispute.
In our use case, disputes will happen between node operators and accusers, who will make cases against operators that appear to use white-label nodes. In order to guarantee that the losing side is able to pay the court fees, it is necessary that node operators post some amount of collateral, or bond, before any dispute happens in the first place. To prevent frivolous accusations, accusers will also be required to post a bond when presenting their case, which is subject to be forfeited if the case is lost.
Additionally, we want to incentivize accusers to procure evidence and make cases against suspicious operators, which will require a reward to be paid to them if their case is won.
Finally, note that it is possible that the decentralized justice protocol under use is imperfect, and so it may lead to a rate of false positives and false negatives—especially in the first round before appeals are considered. We want to quantitatively model the effect of these imperfections on the choices of a rational actor, lest it encourage any kind of financial abuse.
Let us define the following variables:
$C$: court fees that must be paid by the losing side in the first round of a dispute to the Kleros court, for redistribution amongst jurors. (Note that $C$ does not include redistributed PNK stake from slashed, incoherent jurors).
<aside> ℹ️ Remark: This parameter is selected a priori when setting up the decentralized court. Among other considerations, it requires taking into account the effort each juror must put into analyzing a typical case.
For more details, see the note Design of a dedicated court for Lido. For details on how a typical case may look like, see White-labeling evidence types.
As a reference value, we can consider the fees currently used by Kleros’ blockchain technical court: with 3 jurors, $C = 3 \times 0.079 \text{ ETH} = 0.237 \text{ ETH}$.
</aside>
$n$: number of validators controlled by a given node operator.
$m$: number of validators involved in a dispute started by an accuser.
$B(n)$: bond that must be posted by node operators to be a part of the permissionless protocol. $B$ can be a function of $n$.
$A(m)$: bond that must be posted by accusers when starting a case. $A$ can be a function of $m$.
$R(m)$: reward to pay an accuser upon winning a case. $R$ can be a function of $m$.
Our objective is to find the relations between these variables to guarantee the correct functioning of the dispute resolution mechanism, while still choosing a $B$ as small as possible, for capital efficiency purposes.