In this note, we will collect various sources on Kleros’ upcoming upgrade, known as Kleros 2.0. Of special interest to us are matters about the modularity of Kleros’ architecture, which may facilitate the integration and customization of our proposed solution.
Here, we highlight some points from the yellowpaper that stood out:
Note that for applications of Schelling points in blockchain systems it is often impossible to guarantee that agents will not communicate as they tend to be pseudo-anonymous. While this is the case of Kleros, we will see that Kleros has an appeal system that incentivizes participants to agree with how potentially not-yet-determined agents in some future appeal round would decide, recovering a partial impossibility of communication. Furthermore, we are undertaking research on how to incentivize participants to not trust any communication they might have between each other, building off of work in [28] that argues that Schelling points also arise in such situations, see Section 4.9.
[28] David Friedman. A positive account of property rights. Social Philosophy and Policy, 11, 1994.
(Note that this is an argument in favor of the $x_{n+1} = 2 x_n + 1$ rule of juror increases. It allows new jurors to always outnumber the old ones.)
"Lone voice of reason" and appeals:
Remark 1. Above we saw that the redistribution of arbitration fees and lost deposits is handled by round. Note that if a given voter then knows or suspects that other voters in her round have voted “incorrectly”, this gives her more of an incentive to vote honestly. In the extreme, a single juror that agrees with the final outcome in a round where every other juror disagreed would receive all the arbitration fees and lost deposits for that round. We call this phenomenon the “lone voice of reason” effect. We will note further implications of this effect below.
The number of jurors increases exponentially as one appeals; hence arbitration fees also rise exponentially with the number of appeals. This means that, in most cases, parties won’t appeal, or will only appeal a moderate amount of times. Hence, via the appeal mechanism, Kleros manages to avoid the unnecessary duplication of effort and high costs that would be required by having a very large number of jurors consider every case while nonetheless the possibility of appealing a high number of times provides a defense against an attacker bribing the jurors. See Section 4.11 for a further discussion of this point.
Two models of appeals:
(...) Hence, when designing an arbitrable contract, one should consider combining crowdfunding with pre-ruling appeal fee insurance (...)