Proposal for Limiting Juror Repeated Selection for the Same Case Except in the General Court

Research
#1

Hello Kleros community,

I would like to propose a modification that, in my perspective, might increase the overall fairness and effectiveness of our arbitration system.

Presently, a juror who has staked a substantial amount of PNK can be selected multiple times for the same dispute in different appeal rounds. This could potentially affect the impartiality of the dispute resolution process for two main reasons:

  1. Economic Incentives: A juror who has voted a certain way in a previous round might feel compelled to vote the same way in subsequent rounds to avoid the potential loss associated with their initial staked PNK. This pressure might influence their decision-making, regardless of the new evidence presented.

  2. Confirmation Bias: Alongside economic incentives, a psychological phenomenon known as ‘confirmation bias’ might play a role. A juror might unconsciously favor evidence that supports their initial decision while dismissing contrary evidence. This bias can potentially skew the juror’s perspective and thus affect the impartiality of their decision.

By considering both these factors, it becomes clear that repeated selection of the same juror for a specific dispute could introduce certain biases into the arbitration process.

Moreover, the proposed system could actually encourage broader participation by reducing the dominance of ‘whales’, i.e., those with significant PNK stakes, in the juror selection process. While a high-stake juror might still split their tokens to be selected multiple times, this would introduce additional friction and could serve as a limiting factor.

Of course, in the final appeal round, i.e., the General Court, we should maintain the current system where all jurors, including those who have previously voted on the case, can be selected. This is necessary to ensure a sufficient number of jurors for the final decision.

This is just a proposal for discussion, and I would love to hear your thoughts, critiques, and ideas. Your feedback is invaluable in shaping a more robust and fair arbitration system for Kleros.

Me and ChatGPT :robot:

#2

Hey Nicolas,

Your post is based off of a false assumption of how Kleros as a system works. The evidence period is before any jurors rule on the case, and once the evidence period is over, new evidence is deemed invalid. If this were not the case, someone could withhold evidence maliciously and release it last minute, thus baiting people into voting incorrectly.

Let me know if you have any other comments or questions.

#3

The juror’s selection proposal is independent of the submission of evidence.

The confirmation bias here relates to the fact that if we voted A in a first instance, we’ll have a greater or lesser psychological bias to vote A again to rationalize the first vote, whatever the evidences revealed or not.

#4

Some of these concerns are addressed by proper setting of the jurors required for court jump parameter.

For example, in the humanity court, the court jump is set at 31 jurors, once a case has been appealed several times in the same court, it should jump to a larger jury pool since the diversity of opinions in the current court are likely exhausted.

Due to the lack of sybil resistance, such a proposal to limit juror redraws in appeal rounds can be circumvented since jurors can unstake any PNK beyond the locked amount and send it to another address to restake and become eligible for jury duty.

In Kleros v2, we do have a sybil resistant dispute kit which will support low level courts (low in terms of the court tree). Jurors can be drawn in appeals, but in any individual round, unique jurors are to be drawn no more than once.

In any case, Kleros v2 provides us with a more flexible, modular court architecture, so we can support a wider diversity of juror selection strategies (SBTs), but the highest appeal courts (general court) will typically be the most inclusive (all PNK holders are drawn without restriction). There is a bias, variance trade-off here which is an active research topic for juror selection.

1 Like