[Phase 1] HIP 17: Reduce Number of Kleros Jurors Drawn in First Rounds of POH Cases to One

HIP: 17
title: Reduce Number of Kleros Jurors Drawn in First Rounds of POH Cases to One
author: William George
status: Draft Phase 1
created: 2021-06-14

Simple Summary

This proposal would adjust the parameter that controls how many Kleros jurors are drawn in the first round of disputes concerning Proof of Humanity registrations/removals.


I propose to reduce the registration deposit by specifying that the initial round of a dispute in case of profile registration (or removal request) challenge only draws one juror instead of three.

This would be done by updating the following Proof of Humanity parameter.

Use changeArbitrator (specifically changing the _arbitratorExtraData) so that the initial round of a dispute only draws 1 juror.


The goal here is to reduce the value required for deposits, to the degree that this is securely possible. This idea was already evoked here: ETH amount for proof - #12 by william However, I’ll repeat the pros/cons discussed there for the sake of this post being self-contained.

Currently, in the first round of Kleros disputes for Proof of Humanity cases, three jurors are drawn to rule on the case. This number of jurors was initially selected to ensure greater stability in rulings and a lower number of appeals while the registry was in its infancy. The deposit that new submitters to the list have to make needs to include enough ETH so that if the submission is ultimately challenged and rejected, the arbitration fees paid to those jurors can be taken from that deposit. Hence, if one starts with initial panels of only a single juror to judge Proof of Humanity cases, this reduces the size of required deposits to submit to the list.

Recall that Kleros is a Schelling point based system and jurors are incentivized based on whether they are coherent with the final juror vote. Hence even when starting with only one juror, there is an incentive for that single juror to vote seriously because there is a potential for appeal. So if people think the juror voted incorrectly, there is likely to be an appeal and the first round juror would be rewarded or penalized based on what the appeal jurors decide. So far Kleros has used first round panels of only a single juror for Linguo (https://linguo.kleros.io/)

The negatives are that some borderline cases might not be appealed, so there is an added level of variability in terms of whether the first round juror represents how a larger panel would have voted, and that in (the minority of) cases where there is an appeal and you wind up back with a three juror panel anyway, there is an additional delay and some extra gas to have gone through the one juror round.


Would it be possible to decrease the voting period for the first round? Maybe make it only one day, or two days, to speed up the process.


I’ve been reading some of the klero system is very good. BUT.
The most important thing that I notice is that a different or lesser system is used for PoH in which the votes are revealed, which leads to predictive majorities either for belonging or for not losing money.
I think the solution would be to set up an ADMINISTRATIVE process within PoH and in the case of NOT arriving at a solution then just move on to kleros; but when it is passed to kleros if the ruling issued by the administrative instance of PoH does not have funds to be submitted to kleros it should be assumed by PoH. And defend it as an own position by some group of defenders armed for this purpose, a legitimate defense outside of an economic decision,
. Some of the points to take into account is that in this case it would not be the victim who would be subjected to the decision of kleros but rather the administrative decision of PoH, which would greatly change the parameter of the caranchos
You could put together something like an oracle with judges chosen at random using something similar to kleros with deposit as well and incentives but with secret votes until the resolution and the substantiation report. This would be something like the foul judges (the MUNICIPAL JUDGES)

Estuve leyendo algo de el sistema de kleros es muy bueno.PERO .
Lo más importante que noto es que se usa para PoH un sistema distinto o de menor cuantía en el cual los votos son revelados lo que conlleva a mayorías predictivas ya sea por pertenecer o por no perder plata .
Creo que la solución sería armar un proceso ADMINISTRATIVO dentro de PoH y en el caso de NO arribar a una solución luego recién pasar a kleros; pero cuando se pasa a kleros si el fallo emitido por la instancia administrativa de PoH no tiene fondos para ser sometido a kleros este debería ser asumido por PoH. Y defenderlo como una postura propia por algún grupo de defensores armado a tal fin , una defensa legítima fuera de una decisión económica,
. Algunos de los puntos a tener en cuenta es que en este caso no seria la victima quien sería la sometida a la decisión de kleros sino la decisión administrativa de PoH lo que cambiaria mucho el parámetro de los caranchos
Se podría armar algo como un oráculo con jueces elegidos al azar usando algo similar a kleros con deposito tambien e incentivos pero de voto secretos hasta la resolución y el aporte de fundamentación de los mismos . esto seria algo asi como los jueces de falta ( los JUECES MUNICIPALES )

1 Like

For cases due to minor errors we don’t need a long voting period but is warranted for highly contentious ones (like the one where someone is suspected of paying people in the street to make the video to be able to use their account).

1 Like

I like this proposal.

I believe that it rewards sending more-than-perfect submissions and highly curated submissions (looking at you The Rolling Funds), and reduces the loss for minor mistakes that are not worth appealing - and I would say most of the cases are like this. I believe that this is the direction we need to take.

1 Like