[Phase 1] HIP #49 Change of Arbitrator

DAO
20

The Kleros distribution graph says nothing. It’s just a graph of the wallets that have voted in snapshot, but like Kleros is not one wallet one vote, you cant know if in the small squares a user controls multiple wallets (like in farming / puppeting in poh). Even more, as I said you in twitter, Ive more than two accounts with PNKs and I’ve no voted with all of them in the snapshot, so the input data it’s biased. In addition to that, the snapshot votes in Kleros DAO are more about Court parameters (most of them, if not all, were accepted), not related to other topics for now.

Regarding this proposal, there are no explanations of several technical issues that can arise in the implementation, so for the phase 2, if this proposal it’s approved in this phase, I would like to see a whole implementation of the arbitrator in a testnet. And for phase 3 would be nice to have the arbitrator with all the associated infrastructure deployed on mainnet (with jurors staked) so we can battle test it before doing the change in arbitrator, for the sake of security of this project. There are other technical issues that came to my mind, but I will expect more clarification about how will be this implemented to say it, because depends on that implementation if it’s a problem or not.

From my perspective, this HIP it’s in the wrong direction because this change doesn’t attack the “concentration” or whale problem that it’s intended to attack, if it’s just switching token to UBI as is sayd in the proposal. Maybe if the UBIVote token it’s used (I don’t know how will solve the SEVERAL technicalities that will arise if that change it’s made in a kleros fork court) you will have a 1human-1vote in the court, but you will be doxxing the jurors which in my opinion leads to a worst scenario. I would like to remember to this community that in the early days of PoH several challengers that uses their PoH wallet were attacked in the telegram groups because they made a challenge (even when the challenge was right, the profile doesn’t met the listing criteria). So, to me, doxxing the jurors it’s not the best approach. Remember that in a dispute, always will be a side that will lose, then that part will be hurt and can take revenge publicly only because the juror vote against him, even if the vote is coherent with the policy requirement. And with revenge I mean doxxing more data of the juror that is not in PoH.

There is another major issue, that the biggest problem of PoH IT’S THE POLICY. And no matter which arbitrator you use, if your policy has grey zones, there are some vector attacks that you can not solve. So, as I’ve said so many times in the last year. LET SOLVE THE POLICY, but I know that it’s easy to attack the arbitrator instead of seeing our own problems.

8 Likes
21

I think we should remember that at the base of this DAO is a system that works because of sybil resistance. The projects built on top of PoH, which serve as incentive to sign up (UBI, governance etc.) have value because the sybil-proof mechanism works.

I am not aware of any Kleros fork that was successful. That doesn’t mean that forking doesn’t work as much as it means that an arbitration system is a very hard thing to do right, or better than Kleros is currently doing it. Changing the arbitrator to an experimental one is nothing less than an existential risk. Is the DAO aware of all the factors involved to take such a risk?

Many of the recent debates in the PoH have been about greyspots around the policy and how Kleros dealt with those. We can accept both the policy and Kleros as non-perfect.

The incentive for Kleros is to retain the reputation as the best arbitrator. Rulings around greyspots don’t show fragility in the system that much and don’t hurt the reputation. Thus a rational solution here is to constructively help improve the policy so to make it more clear and not leave room for rulings from Kleros that go “against the DAO” (since the policy should reflect what the DAO requires for people to be on PoH). I believe this kind of separation of powers is best to make the DAO coherent and cohesive for both insiders and outsiders.

We can also accept Kleros as a non-perfect system and try to fix it. As mentioned, Kleros is working on a new version allowing drawing of jurors based on for example PoH registration status. Why not let Kleros experiment with such a new mechanism and, if successful, have PoH adopt it?

Even discussion around the centralization of Kleros can be more constructive and understanding. I can vouch that members of Kleros support decentralization and are currently working on (and interested in) solutions to bring working decentralization outside but also inside the organization (also for PoH). I believe recent discussions have shown the need for such solutions and will foster the way for Kleros and others to develop them.

I believe these points cover most of the expressed cause for the momentum built around such a proposal.

9 Likes
22

this proof of humanity HIP aims to vote court parameters with the proof of humanity governor and not get those settings from the kleros coop which is heavily centralized and operates with a corporate hierarchy (has a CEO, a CTO, etc).

I agree with this. the highly technical aspects of this proposal should require on each phase advancements in the development of the proposed alternative solution.

i don’t think things boil down to the policy. during the weekend we saw behavior very much against the community in the telegram channels. it’s only logical the community wants to test how effectively decentralized the protocols are.

23

This is cherry picking on a highly consensual and low impact (thus very low turnout) proposal about adding a court in the experimental Kleros on Gnosis Chain.

Most Kleros gov proposals are pretty uncontroversial - adapting deposits or fees to market conditions. People are free to submit parameters, and if a vote was particularly controversial, you would probably have a voter pool more like the people actively participating as jurors. (William George)

A better comparison in the context of Proof Of Humanity would be the repartitions of voting power in the Humanity Court which looks like this:
image

The settings are not determined by the Kleros Coop (which membership is open to any direct or indirect user of Kleros by the way) but by PNK holders.
Kleros has never voted to change settings against the opinion of projects using it (and the vote you are referring to is actually a proposal from a project wanting to use Kleros).

5 Likes
24

This court is mannaged by 10 wallets, it is very higly concentrated. This is one of the main reason we want to chage the arbitrator parameters and include UBI instead of PNK. Give to the POH community the poewr to mange his own registry.

A better comparison in the context of Proof Of Humanity would be the repartitions of voting power in the Humanity Court which looks like this:
image

3 Likes
25

GM santisiri, I have questions regarding conflict of interests.

You are the founder of UBI and a significant holder. Currently you publicly hold 295k UBI, you are the biggest active voter on the UBI DAO and you also have the biggest amount of delegations on POH. Do you have any significant amount of UBI in anon accounts? If yes, would you reveal how much for the sake of transparency?

I also remember when an anon UBI whale who was an early democracy earth investor dumped UBIs last year. Are there still democracy earth “investors” holding significant amounts of UBI around? If yes, how much and what is your relationship with them?

I was trying to check what the UBI DAO vote power distribution is and surprisingly I might have found something concerning (notice that you should be comparing PNK vs UBI, not PNK vs POH, which is dishonest). If you look at the last Snapshot poll, it is quite evident that there is something rigged. Those values are not square roots of UBI, not even with delegation, and you have more than 50% of the voting power. I don’t understand what is exactly wrong there. Maybe UBIVOTE is broken or the UBI snapshot space that you manage is corrupted. Not surprisingly, many supporters of this proposal also voted with a corrupted amount of UBIVOTE. Please clarify what is happening ASAP.

On the other hand, during the Mission Board elections, which you won by a big margin, you said that you didn’t vote in order to let the community speak louder. Taking into account that there are some conflict of interest concerns, would you also abstain from voting HIP 49 to let humans speak louder?

Lastly, let me say that this proposal seems like a trap to me. It copies the bad of Kleros and makes it worse. Even with quadratic, human votes, UBI Kleros courts will still be a stake system. If you are interested in a fully democratic system, maybe we should start heading in the direction of more democracy instead of insisting in staking systems: [Phase 1] HIP-47: Democratic submissions and removals.

4 Likes
26

I can argue that the people who develop and study 24/7 through several years are in better position to suggest the court parameters rather than a lot of people who don’t understand how the arbitrable works. But again, i would like to see how this experiment goes. Maybe in that way all this people that think that make an arbitrator reliable it’s easy will find out that it’s very very hard.

How decentralized it’s the UBI DAO?, because if the new arbitrator will rely on UBI, we should compare UBI against PNK instead of PNK - PoH. There are corporate with CTOs, CEOs holding UBI? How will this impact in the arbitator?

3 Likes
27

I don’t hold any other UBI than the one you can see on my public account santi.eth

To my knowledge there are no significant investors from the UBI regulated pre-sale that we did in 2018.

Yes, we realized that the delegator plugin was pointing to UBI instead of UBIVOTE. This was an honest mistake and it was fixed right away. Fortunately the only UIP affected by it was one that had 99% of support. Shouldn’t happen again. Apologies for any confusion regarding this.

4 Likes
28

I have a right to represent my delegates and humans already spoke very loud during the mission board election, so expect my vote to be cast with conviction.

I will check HIP 47, thanks for sharing.

29

To my knowledge, there are some CEOs holding UBI like Marcos Galperin, but it’s not a significant amount. He accrues like any other ordinary human

The biggest holder of UBI to date is Vitalik Buterin and you can check this on Etherscan, with 1.8M UBI

2 Likes
30

Actually this isn’t the case (but it’s probably an honest mistake as simply looking on etherscan doesn’t show tokens in pools, I advise using debank for this purpose).
The largest UBI wallet belongs to the cooperative with 2.1M UBI (we removed those from the pool for transparency).
Note that the Cooperative didn’t get any UBI allocation, it bought those to provide liquidity and in order to finance the DAO. We bought for 150k$ of UBI (worth 7k$ today) from the DAO and the remaining passively as a liquidity provider.

I think it could be interesting for voters to know:

  • How was the initial supply of UBI allocated?
  • How much money was raised from the presale?
  • How the money was spent?
  • How much money is there left?
  • How much UBI belongs to Democracy Earth?
8 Likes
31

Could you further explain what this means?

32

  • The initial supply consisted of 10M UBI of which 4M where allocated to the DAO, 4M to the liquidity farming contract and 2M to the pre-sale investors and community.

  • $1.2M in 2018 valuating each UBI at $1.25

  • 100% - We are a 501 (c) 3 not for profit corporation, we don’t do equity or dividends, everything is invested.

  • From the raise, Democracy Earth has invested it all in the R&D that led to the launch of UBI between 2018 and 2021 which includes projects with the Ethereum ecosystem (eg. MolochDAO smart contract), implementation of pilots (eg. Quadratic Voting in Colorado in collaboration with Glen Weyl et al.), participation in conferences (eg. Devcon in Prague and Osaka where we met to brainstorm Proof of Hymanity) and published papers (eg. “Who Watches The Watchmen”)

  • 330K UBI, never sold.

It should be clarified that Kleros bought UBI to the Proof of Humanity DAO, not to Democracy Earth. The ETH from that transaction is in the DAO treasury and you can vote on how to allocate it.

5 Likes
33

The allocation of UBI is not a deal since there are some ideas that involve only DRIPPED $UBI to play the role that $PNK plays today.
(I still do not know if that is technically possible, but i’m aware of the discussion)

Additionally, jurors may be asked to be verified humans. Another possible upgrade to the current model.
So is it really pertinent the $UBI distribution?

1 Like
34

Antes de exponer mi idea deseo aclarar por lo declarado por el Sr. Clement es que no comparto y no me parece adecuado mezclar en este hilo la HIP48 y Hip49 dado que las dos representan propósitos diferentes y a su vez están realizadas por autores diferentes.
Como parte de la comunidad hispanohablante expondré mi idea en castellano (usemos traductor)
Sito lo expuesto por Santi y que comparto “La complejidad de la implementación no debería impedirnos expresar lo que queremos representar como DAO… “ los autores de esta HIP hemos sido críticos del arbitrador en todos los debates que se han llevado a cabo en los grupos de Telegram y más en el de POH en Español, algunos lo han expuesto de manera más pasional que otros pero con la visión en común.
Debo asumir que mi conocimiento sobre lo técnico se los dejo a los especialistas, soy analfabeta en ese campo, sin embargo eso no me inhabilita para reconocer que los objetivos que persigue la Comunidad de POH no está alineada a los intereses del arbitrador, es relevante recalcar que el governor de nuestro arbitrador lo maneja una empresa privada . Y frente a ese escenario nada nos impide tomar el control de nuestro destino, a la descentralización. Seguramente el camino no será fácil pero tampoco imposible. Reconozco en algunos comentarios un alto grado de amenaza y una forma de instalar miedo, herramienta utilizada muchas veces como arma de dominación política y control social.
Seguramente este es un momento histórico como DAO para ver si las decisiones que no se alinean a los intereses de Kleros son capaces de prosperar dentro de POH, reconocer que tan descentralizados estamos y parafraseando a Santi reconocer“Si somos un proyecto propiedad de kleros o un proyecto libre”

5 Likes
35

This proposal is about using UBI, so that is definitely pertinent.

And if you want jurors to be humans that could be done in the future via a custom Kleros V2 module.

Pero, ¿podrían encontrar un caso en el que Kleros no dictaminara de acuerdo con las reglas?
En realidad, en su mayoría han estado criticando las reglas, pero debido a que las reglas las decide el DAO, no pueden criticarse a sí mismos y trataron de usar a Kleros como chivo expiatorio.

But could they find one case where Kleros didn’t rule according to the rules?
Actually they have mostly been criticizing the rules, but because the rules are decided by the DAO, they can’t criticize themselves and tried to get Kleros as a scapegoat.

Esto es falso, el gobernador de Kleros es el DAO de Kleros, no la Cooperativa (y la cooperativa es, como su nombre lo indica, una cooperativa a la que cualquier usuario, incluidos los usuarios de PoH, puede unirse).

This is false, the governor of Kleros is the Kleros DAO, not the Cooperative (and the cooperative is as its name indicates a cooperative that any user, including PoH users can join).

5 Likes
36

There is a long list of cases where there was an inconsistency between the written rule and what was actually happening.

  • The practice of precedent chaining of edge cases is just one example,
  • the 360p quality and not 360 pixels minimum interpretation is another.
  • The invention of a thing called “effective size” in rotated and padded videos is another (it was not defined in acceptance criteria and yet they were challenged and removed).
  • Calling “filter” something that is not (black and white pictures are not a threat to sybil resistance and it is also definitely not a filter, but yet they were challenged and removed).
  • Using a small cropping on top or the bottom of the face and calling that “no facial features were visible” when actually any facial detection filters available where able to identify the face corresponding to the user.

There is constant overshooting and abuse of the written law, that is beyond an underspecification of the metaevidence.

2 Likes
37

To begin with the pie charts checks all the boxes of what is considered a really “ugly visualization” Pie Charts in Data Visualization- Good, Bad or Ugly? Them being:

  1. Limited display of Diversity: Pie charts are not very effective at displaying categorical data where the number of categories exceeds 3 or 5. The more slices in the pie, the harder it is to read at a glance.
  2. Limited Precision: Since Pie Charts are area-based representations of data and do not have an axis, conveying information with higher precision is difficult without the use of data labels. The inclusion of these data labels would conversely limit real estate on a dashboard and detract from the ability to consume information at a glance.
  3. Ambiguity in Size & Orientation: Pie charts lack specific ordering with respect to their slices and are open to differences in interpretation. Moreover, additional accents such as 3D or tilting can cause misrepresentation in the size of slices.

The best representation of the juror staking distibution is one that you have a clear distinction of how many wallets belong to a specific range of stake, and lumping them together as a way to summarize the available information.

This is the plot that actually reveals some information (and funny enough, it was repeatedly flagged as “misleading”). It shows the amount of stakers per level of staking, and the number in blue represent the amount of stakers within each category.

There is not to much to say that there are currently only 36 wallets occupying more than 3/4s of the humanity court, while 164 take the rest of the seats.

image
image680×543 41.8 KB

Moreover, it is the view of Kleros’ CTO says publicly that this level of concentration is a good thing to have, because supposedly he considers a high concentration at courts (like him occupying more than 50% of the stake of Technical court) a way of the “good jurors” set the “good criteria” (whatever good or bad is in this case, and in the context of game theory).

Also, Clement himself recognizes that “bank attacks” could happen in a Kleros court, in which regular registered Humans would be helpless to defend themselves, because even with the possibility of crowdfound an appeal, no one has enough time or resources to gather that amount of funds, especially when the perception of concentration and bias in the court is incredibly high.

image

The “whataboutist” comparison with ethereum mining is first, not relevant since we are not talking about mining or not even governance, and second very tangent to what we are talking about which is access to the right of participation in the justice process of our own DAO.

Let me add also what the criteria of what the high standards of one of the Kleros DAO leader has over what he considers what a “good court” is:
image

38

Could you give links to those? Maybe you disagree on some edge cases, but the rules were always applied. If you believe the way to handle some edgecase is bad, you can make a proposal to modify the rules to specify the edge case and this would not be an edgecase anymore.

This pie chart is highly confusing and you know it. I showed this chart and asked people how many people were required to get the majority and they all said 3 (which is wrong). This pie chart is grouping all wallets with similar staking size together while the chart I gave is objective row data.

That isn’t true, I said that initially to bootstrap courts good functioning it was good for large stakeholders to stake, but that overtime courts should decentralize and the Humanity court was a good example of that.

Again, taking my messages out of context. This is an answer to the idea of allowing an unlimited amount of appeals. So yeah, in a hypothetical version of Kleros with an unlimited number of appeals, you could have bank attacks, that’s why there is a limited number of appeal of in Kleros, thus Kleros is protected from bank attacks.

It is not “whataboutism”, I’m a huge supporter of Ethereum. The comparison is only to show what people consider decentralized.

I don’t see anything wrong with that, do you mean you want stupid people to rule on cases?
Here I’m pointing the importance of initial conditions for Schelling games. Since jurors are incentivized to rule according to the final ruling, if they know that the court is mainly composed of clever people, they should do it in a clever way, but if it were composed of stupid people, even clever jurors would have to rule in a stupid way to be coherent with them.

5 Likes
39

These are just your subjective opinions. In PoH you will find 16k slightly different opinions on how to rule disputes with any given policy. For example, in my case, I disagree with 4 out of the 5 points you mentioned, but I still recognize that there are reasonable arguments on the other side of the discussion. As it has been said and done many times, the good thing is that PoH can adjust the rules if there is democratic consensus that a case should have been ruled differently. So far, some of those adjustments were aligned with your believes (360p quality), some others not (black & white pics).

With your rationale (thinking that there is only one right way of interpreting subjectivities in the policy), there is no arbitrator that will ever fit your wishes. Not even the Kleros-UBI fork. I recommend you waiting for Kleros v2 and setting up a “Ludoviko Module Subcourt” where you can be the single, qualified juror and rule according to your believes.

Talking about stake distributions… It may be counter-intuitive, but even a Ludoviko Subcourt would be decentralized if it can be immediately appealed to a Humanity Subcourt. Decentralization is not a picture.

Let me ask you some questions:

  1. If we continued using Kleros, what would you consider a secure stake distribution? For example, would 30 wallets having the current 50% of the court stake be enough? 50 wallets? 100? 1000? How many and what is your criteria?
  2. In a Kleros fork that uses UBI, how will people know if a stake distribution is secure enough according to your standards? 30 wallets having the 50% of the court stake? 100? 1000?
  3. If a court with stakes that you consider distributed enough rules against your believes, what will your reaction be?
  4. Let’s say that UBI Courts are implemented and governed by PoH. Can you give me a few simple examples of HIPs that could modify court parameters in a way that prevents/punishes juror behaviors that you consider unacceptable?

Don’t get me wrong, I don’t think Kleros courts are perfect (no court will be perfect). Here are some of the improvements that I’d like to see (and none of them is about the distribution of stakes of a given court at a given time):

  • Better alignment of incentives between appeal funders and jurors. Today there are some edge cases in which a juror can be “coherent” even if the final ruling (according to PoH) is opposite. This will not happen in Kleros v2.
  • Lower fees. Cheaper arbitration means that first rounds can include more jurors without getting prohibetly expensive and/or it gets also cheaper to take the risk of appealing. This will improve both if Gnosis courts are used or if Kleros v2 on a rollup is used.
  • Easier forking mechanisms integrated into the court itself. Again, this will improve in Kleros v2.
  • Bigger market cap and bigger total stakes. Currently there are 350k USD staked in the Humanity subcourt and 4M USD in total. This is nice but it would be better if it increases. Demand for Kleros services is increasing and with competitors such as Aragon failing it seems pretty obvious to me that PNK stakes and market cap will increase in the future. (How does that look for UBI courts??)
  • More diversity of subcourts with activity. I think we will see a lot of increase there, in particular with Curate, Realitio and Insurance claims.
  • Maybe secret vote, but I’m not convinced that it will improve the quality of the courts. In any case, PoH could ask Kleros to create a Humanity subcourt with commit & reveal scheme.
  • Better ways of determining when an appeal should jump to a parent court. I believe there is some work in this regard in Kleros v2, but I don’t know for sure.

In my opinion, UBI courts will be just another low quality fork of Kleros.

10 Likes