Hypercerts - retroactive impact funding - workflow of evaluation - incentives

There is an overlap among Web3, Regen, ReFi, Gitcoin, Supermodular, Hypercerts, Metacrisis.

I would like to cross-pollinate an idea, initially posted on Hypercerts Github:

Previously I was working on a new definition of value that accounts for externalities.

Accounting for externalities and accounting for impact = this is pretty much the same.

Based on my research and experience I am suggesting two improvements:

:one: Workflow of evaluation

Noone knows the project better than the founders, participants, people on the ground. They have a much better understanding of the impact they are creating.

Then use Kleros jurors who will have a simple job:

  • :white_check_mark: Yes, correct, the overall impact assessment is "reasonable"
  • :no_entry: No, wrong, there are some issues

I know that Gitcoin was using Kleros in the past to evaluate if a project is a public good.

Using already experienced jurors solves the problem of finding evaluators of the impact - which is totally not scalable - see how Verra is doing it: verra.org/validation-verification

Related project: The Price of Impact Index: Benchmark Your Outcomes (that particular page went offline, currently only web archive version)

As of March 2023, the AI got to the point where the owner of the project can describe their activities in plain text - then autocompletion and autosuggestion for relevant quantified categories.

:two: Negative impact

It’s about social pressure and providing incentives to buy positive impact certificates.

I can totally see the future when by law it will be required to compensate for historical damage.

Media attention and social pressure can force those who did the damage to compensate for it by buying positive impact certificates on the open market.

Providing demand for impact certificates = pretty big deal :muscle:


You’ve set yourself an interesting and valuable challenge their @mars. My only comment at this juncture is that I’m not wholly convinced of the applicability of Kleros courts.

A Kleros juror has personal gain rather than justice as their foremost motivation, by design. This has a fundamental implication — they will vote the way they think the majority will (by which they profit) rather than by their determination of ‘right’ or ‘wrong’. i.e. the latter is secondary when it should be primary.

1 Like

Yes, and…

Yes, but…

I’m not disagreeing with you. In fact I believe you are mentioned something truly valuable, only a few people can grasp.

Level 1: Not that many people know Web3
Level 2: Not that many people know Kleros
Level 3: Not that many people know the incentives of the jurors
Level 4: Not that many people know the incentives of the jurors and when it can go astray

Welcome to Level 5

Leonardo Di Caprio Inception GIF - Leonardo Di Caprio Inception GIFs

We need to go deeper…

Jurors vote for the majority.

  • you get what you incentivize
  • show me the incentive and I will tell you the outcome

The incentive is to behave coherently, therefore they will vote for the majority, full stop.

I tried to come up with an example where majority is not right = … … … … … … … (please come up with ideas here)

There is an incentive to be the “unpopular juror”:

  1. Vote as a juror against the majority
  2. Case goes to the next round
  3. Your decision is upheld
  4. Financial incentive :moneybag::moneybag::moneybag:

NOTE: If you are that “unpopular juror” - you should publish your thinking process, upload it as evidence, convince others.

(Kleros has no KYC, same human can be a juror, plaintiff, defendant, using multiple ETH addresses = feature not a bug)

So far, there was not a single case that was obviously wrong. There were some cases:

  • technicalities
  • nuances
  • wording
  • loopholes
  • multiple possible interpretations
  • astroturfing and attempts to spread misinformation

(and Kleros remained resilient at all times)

When I look at FTX SBF legal fees, I think Kleros is 100x more cost-efficient and 10x faster.

(gathering evidence, voting, funding appeal = it takes a few weeks)

(interesting thread about one of the biggest, most controversial cases)


Another controversial case: #82 GRID PLUS Badge – Challenge | Heliast

It was about 3rd party audit. Project form ConsenSys ecosystem and ConsenSys audit wasn’t considered 3rd party.

More thoughts:

My forum post about definition of “fairness”:

I want jurors to consider what is fair, not just coherent.

Director of resaerch:

More Discussion

Telegram link: Telegram: Contact @kleros

There are people would love more Level 5 discussions :brain:


By contrast, I have tried to come up with an example of any approach to justice that actually invokes the majority, and found none in my (undoubtedly limited) understanding of the topic.

As Rawls noted (1971), “the rights secured by justice are not subject to political bargaining or to the calculus of social interests.”

You also I think place far too much faith in Kleros jurors’ readiness to invest their time in reading. They are not sat in a courtroom stripped of their mobile devices. There’s no dopamine hit here. The behavioural axiom here will be simply to invest as little time as possible.

As for evidence that it works. If there is some objective adjudication of the adjudications(?), and if such analysis concludes “it works”, then a critical caveat is omitted. Such a conclusion might only be stated rightly as “it appears to work at its current tiny scale.”

This is correct but incomplete.

Jury votes in accordance with what they think the majority will vote for. The key here is: ‘what they think the majority will vote for’.

For a moment, assume that what they think others will vote for is 100% in accordance with justice. Then you would say Kleros produces 100% just outcomes, right?

Yes, so now the problem limited to how to make people to think that others will vote in accordance to justice. Here comes in the Schelling point principle. Kleros utilizes this principle with court policies, which say in a nutshell, votes should adhere to the contract, the evidence presented and the common sense of justice.

In addition, there is always a threat of an appeal, so being coherent with the current jury is not enough, you need to be coherent with the potential future jury. So colluding is almost impossible unless you hold more than 20% of all PNK in circulation.

To begin with, I would like to express my gratitude towards your scepticism and greatly valuable insights.

It is a clear signal that Kleros should improve the communication / awareness / education aspect.

Of course you can read their book, whitepaper, research - but ultimately it will come from Kleros-aligned network - instead we need more RED TEAMING (concept borrowed from Metacrisis) - researchers actively trying to find holes.

In England and Wales a majority of at least 10 votes out of 12 is needed for a verdict. If fewer jurors remain, majorities allowed are 11–0, 10–1, 10–0, 9–1 and 9–0. Failure to reach this may lead to a retrial

Have you been to real-life court recently?

2017-2019 I was in court 10+ times in relation to my divorce and roughly the same started engaging with the Kleros community.

(never officially employed, just hanging out)

I stake 100k PNK (around $3k) in the court and recently was a juror.

There are not that many cases. I put the effort in because I didn’t want to lose the money. Fear of losing money was bigger than winning money.

Source: https://klerosboard.com/

I think Kleros can safely handle cases in region of $1m at stake.

$10m stake would be a breakthrough, that would surely generate interest, media attention, more publicity.

I think that’s the term for that?

1 Like

Yes, exactly, thanks; what I meant is a Keynesian beauty contest.

Here’s a helpful article,

Kleros: Gaming in justice – Kleros response

published on the Laboratorie de Cyberjustice, a University of Montreal research lab addressing some Kleros sceptisism. Here’s an excerpt,

The most common objections to ADR in consumer disputes include time, complexity, and costs concerns as well as concerns of bias. Since processes like Kleros have been developed primarily to address the access to justice gap left by courts and ADR, the only convincing objection that opponents of decentralized justice can truly cite is if said processes were in some way consistently producing results that were patently more “unjust” than those produced by centralized processes.

However, most critics have not so far managed to demonstrate the unjustness of Schelling Point-induced decisions via empirically sound evidence, preferring to engage instead in hand-wringing about the evils of money couched in ornate academic language. Ultimately, what we are left with is pages of what is essentially an appeal to personal incredulity.

At the time of writing, the Kleros protocol has resolved 1365 disputes by 779 jurors, and more are being added each day. It should thus not be difficult to identify potential design flaws in a more practical manner.

And regarding you point,

Yes. Small is beautiful : ) — however Kleros is built to scale.

Empirically, Kleros is working well for it’s users. I would be happy to discuss practical examples of Kleros in use.


I’m not sure about this exact number.

Should it be 51%?

I’m familiar with 51% attack, if you have more than everyone else obviously you can outvote them at a cost of a fork and worthless PNK tokens.

I’m familiar with the concept that if a niche court (with low amount of PNK staked) runs out of jurors (because so many rounds of appeal) then it another court in the hierarchy is used (“General Court”) and effectively all PNK will be voting.

I don’t think it ever happened?

Happy to understand more and have more Level 5 discussions.

Apologies for the delayed response; I’ve been preoccupied.

The 20% figure I mentioned earlier was an arbitrary threshold I devised in order to keep my message succinct. If it led to confusion, allow me to clarify my rationale.

In a situation where an actor possesses over 50% of all tokens in circulation, the system would be effectively compromised. Although a forking mechanism could potentially salvage the system, I will assume, for the sake of this discussion, that no such mechanism exists. Note that the current version of Kleros lacks this feature, though the next version will include it.

A successful attack may still occur at a threshold lower than 50% for several reasons. First, the voting participation rate comes into play, as not all token holders are required to take part. Consequently, holding less than half of the tokens could still grant an attacker practical majority control over the Court.

Secondly, the division of tokens among subcourts allows for the possibility that a whale may achieve majority control in a specific sub-court with a relatively smaller overall stake.

Another factor to consider is intimidation. In Kleros, the jury’s objective is coherence, which relies on the Schelling Point mechanism. Ordinarily, jurors would believe that adhering to the rules offers the highest chance of achieving coherence. However, the presence of a whale can disrupt this assumption, leading jurors to perceive a greater risk in following the rules due to the whale’s potential to resist compliance using their significant voting power.

As a result, jurors may reevaluate their strategy in an attempt to maintain coherence in this new context. If they become intimidated by the whale and opt to follow its lead instead, the whale effectively gains even more voting power. This psychological effect can perpetuate a cycle wherein non-attacker jurors are swayed to join the attacker through intimidation.

So, why choose 20% as the threshold? It is derived from my personal experience and intuition. There is no definitive cut-off point, as greater token concentration inherently poses greater risk. However, I believe that a concentration of 20% should serve as a warning signal.

Please note that this vulnerability analysis is not a formal proof but rather a reflection of my own observations and insights. Additionally, these views are solely my own and do not represent those of Cooperative Kleros.

It would be “Layer 0” social coordination, underneath each system there are humans running it.

Didn’t think about this one.

Or maybe it would trigger: Rally 'round the flag effect - Wikipedia

All the jurors would on high alert and motivated to vote truthfully?

PNK distribution: Pinakion Token Contract and Distribution Chart

Of course, 1 attacked can have multiple wallets.

Overall: I see this mostly as a “theoretical debate”…

Over time the safety limit will increase.

Indeed, it is technically possible for individuals to recreate Kleros from a snapshot excluding the attacker’s wallets. However, in practice, this approach is deemed impractical and therefore unlikely to occur. Recognizing this, the jury opts to rally around the attacker instead of attempting to defend the system. Fortunately, the upcoming version of Kleros aims to address this issue, offering a robust forking mechanism as a very dangerous threat to attackers: losing all the capital deployed for the attack by getting excluded.

In this scenario, the primary incentive (the flag) is the economic benefit, not the ethical ideal, which effectively rallies individuals around it. It’s important to recognize that this system not only rewards but also penalizes participants. Consequently, whenever a juror votes against the attacker, they risk losing both their potential rewards and their principal investment. As the attack progresses through each appeal round, resisting it becomes increasingly difficult. Smaller stakeholders may eventually succumb to despair and capitulate, further strengthening the attacker’s influence with each successive surrender.

I must also highlight a point that was not addressed in my previous responses: the cost of each appeal round increases exponentially. Although crowdfunding exists as a potential mitigating factor, the astronomical expenses incurred after several rounds significantly reduce the likelihood of further appeals. The security of the coherence game is intrinsically dependent on the possibility of an appeal round. If jurors believe that an appeal round is unlikely to occur, they may feel compelled to surrender.

To summarize, in a system that relies on the Schelling Point, the behavior of actors is heavily influenced by their perceptions. If an attacker can successfully manipulate these perceptions, they can effectively disrupt and compromise the system.

Meanwhile in Hypercerts Telegram:

Once $1m cases are regular, there will be more robust ecosystem of jurors, more robust incentives. I still believe that the suggested workflow is sensible / reasonable / practical / decent / acceptable / probably one of the best option available as of now:

1 Like