KIP-66: Long-Term Juror Incentive Program

Date: 2023-12-22

Author: Kleros Cooperative

Context

Maintaining the economic security of the court is vital for Kleros protocol. “Kleros economic security” refers to the system’s resistance against attacks and manipulations by malicious actors. A higher staking percentage enhances resistance against attacks and reinforces trust in the protocol.

On Feb 3, 2021, the community voted KIP-37 in order to distribute rewards in PNK to all users that staked their tokens. The program was renewed by KIP-46 on Dec 22, 2021, and by KIP-58 on Dec 29, 2022.

As of December 19, 2023, data from Klerosboard shows that approximately 213 million PNK are staked in Kleros courts across both Ethereum and Gnosis Chain. This represents 28% of the total PNK supply. Meanwhile, around 120 million PNK, or 15% of the supply, are held by the Cooperative (either committed for future compensation for team members or used for liquidity on platforms like Uniswap and Bitfinex).

Since the implementation of KIP-37, there has been more than 70% increase in PNK staked in Kleros (from 123M in March 2021 to 213M in Dec 2023).

The Juror Rewards Program was successful in increasing the proportion of staked PNK and bolstering the resistance of the protocol against 51% attacks.

Proposal

The goal of this proposal is to further increase the percentage of tokens staked in order to secure the protocol over the long term.

For this, it seeks to establish a long-term incentive program where users who stake their PNK tokens receive monthly rewards in PNK. The rewards to be distributed will depend on the duration and amount staked each month, calculated with the following formula.

The total reward for the next month (i+1) is given by:

R(i+1) = R(i) * (1 + t - s(i)), where:

  • s(i): Percentage of total PNK staked in monthi.
  • R(i): Reward in month i (amount of PNK distributed in month i).
  • t: Target percentage of PNK.

The target will ramp up by 1% each month starting from 28% (current staking level) until it reaches 50%.

In Dec 2023, 28% of PNK are staked, so the target for Jan 2024 will be 29%, the target for Feb 2024 will be 30% and so on. Every month there will be a 1% increase up to a final target of 50% in Oct 2025. Since that moment, the target will remain at 50% every month.

The target of 50% was chosen because with 50% staked it is extremely hard for a malicious player to do a 51% attack with a market buy.

Below are examples for January 2024 and how February 2024 rewards could behave:

Reward(Jan24)=Reward(Dec23)*(1+Target(Jan24)-Staked PNK %(Dec23))

Reward(Jan24)=1 M*(1+29%-28%)=1.01 M PNK.

Example 1: Increase in Staked PNK % in January

If, hypothetically, the staking percentage increases to 32% in January, the reward for February 2024 would then be:

Reward(Feb24)=Reward(Jan24)*(1+Target(Feb24)-Staked PNK %(Jan24))

Reward(Feb24)=1.01 M*(1+30%-32%)=0.9898 M PNK

Example 2: Decrease in Staked PNK % January

If, hypothetically, the staking percentage decreases to 24% in January, the reward for February 2024 would then be:

Reward(Feb24)=Reward(Jan24)*(1+Target(Feb24)-Staked PNK %(Jan24))

Reward(Feb24)=1.01 M*(1+30%-24%)=1.0706 M PNK

With these two hypothetical examples we can clearly see:

  • If the Staked PNK % increases over the target the rewards will decrease.
  • If the Staked PNK % decreases below the target the rewards will increase.

Why this Formula?

The formula aims at continuing the increase of the percentage of staked PNK as well as to allow for a long term stability and predictability of the Juror Incentive Program.

Unlike the previous rewards model implemented by KIP-37, KIP-46 and KIP-58 which used a fixed total reward and that had to be renewed every year, this proposal adjusts the rewards efficiently, avoiding excess allocation when goals are met, and gives long term predictability to the program.

This formula also encourages decentralisation and behaviour alignment by rewarding users who contribute to securing the protocol through staking and adhere to Kleros’ long-term vision. The fact that the adjustment is done gradually gives market participants enough time to position themselves accordingly.

Allocation between Ethereum Mainnet and GnosisChain:

The proposed allocation is 90% on Ethereum Mainnet and 10% on Gnosis Chain. This allocation mirrors the previous year, focusing on having the maximum economic security on Ethereum Mainnet.

Specific Allocation for Kleros 2.0:

The Cooperative is currently working on Kleros 2.0 (more details here). At some point in the future, during a testing phase, Kleros 2.0 will be live in parallel to Kleros 1.0.

When Kleros 2.0 is deemed usable and secure enough, Kleros 1.0 will be deprecated and all the court activity will be moved to Kleros 2.0. This event will be called “The Kleros Merge” in analogy to the Ethereum Merge.

When Kleros 2.0 becomes live, before the Kleros Merge, 10% of the incentives distributed by this program will be directed to Kleros 2.0 jurors (leading to a 80% in Ethereum mainnet, 10% in Gnosis and 10% in Kleros 2.0). After the Kleros Merge, all the rewards will be distributed in Kleros 2.0.

Details

The tokens will be distributed using a Merkle Drop contract. The distribution will be done once a month and the computation of rewards may use snapshots.

Means

Initially, the reward distribution will be handled by the Kleros Cooperative. After Kleros 2.0 is launched, a proposal could be made to implement the distribution at the protocol level (and potentially auto-compound them).

At the beginning of each year, the Kleros DAO will mint for the Kleros Cooperative Safe (eth:0xE979438B331b28D3246f8444b74caB0f874b40e8) an amount of PNK corresponding to the current yearly distribution rate (rate of the last month of the year X12, so 12M PNK for the first year). At the end of each year:

  • If there were more PNK distributed, the DAO will mint an amount corresponding to the difference between the initial minting and the amount distributed (ex: if 14M PNK were distributed in the first year, the DAO will mint 2M PNK) to reimburse the cooperative.
  • If there were less PNK distributed, the Cooperative will reimburse the difference to the DAO (ex: if 10M PNK were distributed in the first year, the Cooperative will reimburse 2M PNK to the DAO).

Voting Options

  • Yes, implement the program.
  • No, rework proposal.
9 Likes

The proposal has been put to vote.

If the staking rewards formula starting from Jan 2024 uses the total amount of PNK staked during Dec 2023 as the baseline starting point (28%), and if the rewards for staking PNK decrease as the total amount of staked PNK increases monthly, doesn’t that mean that if the target for staked PNK is reached every month from now until Oct 2025, ie every month the total amount of PNK staked increases 1%, then the current reward rate of 5.7% ish for staking PNK is only going to decrease from now until 50% of the PNK supply is staked in Oct 2025. With that being the case, couldn’t this proposal actively incentivise current PNK stakers from dissuading current and potential future PNK holders whom haven’t yet staked their PNK to stake their PNK, since the current PNK holders whom are staking 28% of the PNK supply would gain more staking rewards for themselves (at the expense of total network security) as the disparity between the target for total amount of PNK staked each month and the actual amount of PNK staked each month becomes greater and greater ? What checks and balances do you envision if such a scenario were to play out ?

Yes, for stakers it’s better if other people do not stake.
Note that it is already true with the fixed reward model which has been used so far (more stakers, means that rewards are diluted).
What is new is this proposal is that it is made to be self-correcting:

  • If not enough (compared to the target) people stake (and the case of stakers dissuading others from staking falls into this category): More rewards are given, thus more incentives to stake.
  • If more than enough (compared to the target) people stake, the rewards decreases, preventing an overspending of PNK while there are enough people staking.
2 Likes

Thank you for the clarification, I really respect and admire the work the core dev team do to develop the Kleros protocol.

1 Like

Hi,

pedro here! I’m a huge fan and supporter of Kleros.
It’s fantastic to have proposals that consider the protocol in the long term, so congratulations! And sorry for the delay in feedback :slight_smile:

It would be great to have a breakdown of these numbers and larger reasoning on why this is the most important metric that you guys are measuring for considering the rewards program as “successful”.

According to KIP-37, KIP-46, and KIP-58, >36M PNK have been distributed so far (+ a retroactive airdrop). (I apologize for not sharing the links to the proposals, but my new user status doesn’t allow me to use more than two links per post)

The amount of staked PNK went from approximately 100m to 140m in the program’s first year. That is indeed a good metric. But then we can argue that a huge part of the following year’s increase in staked PNK comes from the incentives themselves being staked by the same people who are receiving this incentive.

I understand that the total PNK supply is 1bn. So 213M would be 21,3% of the total supply? I guess the % you are referring to is the circulating supply. This new incentive program proposal states that it will be funded by PNK minting rather than the cooperative reserves (as in previous incentive programs). This being the case, and if the trend remains the same as in the last two years, the increase in staked PNK will be the same as that of the supply, meaning that the percentage of staked circulating PNK will not change significantly, and the program’s objective will not be met. Why would it be different now?

Also, I am not sure if this kind of incentive continues to be attractive for new jurors to join the program. I understand that the goal of the proposal is “to increase the percentage of tokens staked to secure the protocol”. But I also understand (correct me if I’m wrong please) that decentralization on PNK holdings is desirable and amount of active jurors is a metric I believe should also be considered. Has the number of active jurors increased enough to justify this spending?

On the other hand, there has not been any time in the last 3 years where Kleros didn’t distribute any incentive at all. This means, there is no evidence if the people will stick to the protocol once the incentives are gone. Renewing without this data is a race until you run out of money to incentivize the people to join or stay.

Have you considered incentivizing coherence votes? Kind of extra reward for voting coherently. I’m not sure if this would work, is just an idea. It would also incentivize dispute creation and the program would run for a longer time since the distribution depends on the amount of new disputes and not only for being there.

I am on old guy, and remember a large discussion the last time minting new PNK was proposed. I believe no more PNK has been minted since then, hasn’t it?

How much of the 300M PNK allocated for jurors incentive programs has been spent so far?
And for how much time are you estimating that this program is affordable?

If the program does not work (example 2 scenarios, or staking increases but at a lower rate than 1% monthly), have you planned a specific timeframe to conduct a new analysis of the program’s effectiveness in achieving its goal? And if it is determined that the goal is not being met, will you discontinue the program, or will you use the funds until they are exhausted?

Generally speaking, I am not a supporter of this kind of incentive program because it attracts holders who are not aligned with the protocol’s objectives, but rather with the expectation of profit, and who will abandon it once the incentives end.

Thank you very much and happy new year to everyone!

1 Like

Hi Pedro,

Thank you for your constructive feedbacks and Happy New Year to you as well ! Below our answers to your questions:

The focus is on the overall staking percentage of PNK, as it is vital for Kleros long-term success. Indeed, with a high percentage of PNK staked, Kleros becomes more secure and even more resistant to attacks while being able to secure higher value use cases. This is also increasing court decentralisation and bringing trust to any protocols or projects that want to integrate with Kleros.

Indeed, the current total supply is 764M so we were around 28% in Dec23. However, it’s worth noting that the 1bn you mentioned was the maximum amount the Kleros team could mint when PNK was controlled by the Cooperative, now it’s controlled by the DAO, it’s up to the DAO to choose to mint more or not.

This would be different now because the formula is adjusting the rewards based on a monthly target. So, if the “old stakers” are the only ones who stake, the staking percentage will remain below the target. This will “push” the rewards up for the next month. So, the non-stakers will have a higher incentive to stake, by not staking they will basically get slowly diluted. It’s a self adjusting system that is incentivising rational market participants to stake. The long term vision is that this could be handled at protocol level when it becomes technically possible.

The “active juror” metric can be difficult to define and track over time, for instance, we can’t know if a single person controls many different addresses. So, we have chosen to focus on the amount of PNK staked because that’s what provides resistance against attacks and economic security.

Decentralisation of the court is highly desirable. The proposal aims to continue incentivizing broader participation and staking. This will enhance decentralisation as more people staking means “old stakers” would have less percentage of the votes in the courts.

The main challenge Kleros has is getting more cases. This is why we conduct very intense business development activities. As more cases start to flow into the platform, we expect the intrinsic incentives of making money by resolving cases to be the main motivation for people to keep their PNK staked.

Our long term vision is having Kleros behaving in a similar way than Ethereum. At its beginning Ethereum was mostly funded via emissions, with time and high demand for its block space, transaction fees have taken a bigger share.

This is a different approach that could be interesting but we have identified some of the following issues:

Our goal is to have a high rate of staking on mainnet because higher value cases are there, so it is particularly important to have the highest possible level of economic security there. However, currently most of the disputes are on Gnosis Chain, so you would want the per dispute reward to be higher on mainnet.

Also, if the reward became large relative to the dispute fees there could be some strange effects - e.g. people creating disputes that would not necessarily have been created in an attempt to manipulate the reward program. Possibly inventing things for courts that have been inactive to give themselves the best odds of being drawn etc. In the end, this could become detrimental to the protocol, so we have chosen not to go that way and to proceed with this Long-term Juror Incentive proposal.

Yes you are right this is the last minting of PNK related to the Cooperative allocation and there wasn’t any more minting since then. We have already used around 37M PNK mostly for the juror incentives program and the tag registry incentives program. We think this program could be sustained forever and even included at the protocol level. The tokens created by this program are given to stakers so as long as they stake they never get diluted. The scenario that we expect is, as Kleros gets more cases, this program will progressively become less relevant as people would be staked anyway to make money by resolving cases.

We have come to this proposal after an exhaustive internal discussion. After evaluating different distribution models, we believe this one has a high likelihood of fulfilling the goal. We will monitor this closely and if we find it’s not the case, we would reevaluate. Our goal is to strengthen the protocol.

The aim of this program is to focus on having the current holders that aren’t staking to stake. By “staking”, we don’t necessarily want them to be day to day jurors, as the “juror side” was never an issue in our marketplace model. The focus is on being staked to provide economic security, that’s why if you just stake in the general court you can qualify for the rewards of the program. As explained previously, a high staking percentage increases the overall security of the protocol and court decentralisation.

Moreover, this would help our Business development efforts. Indeed if we get to 50% staking ratio, then we are extremely more likely to have important integrations than at 28% staking rate. More percentage of staked tokens means more economic security which leads to big names of the industry becoming more likely to use and trust Kleros for high value use cases.

When we get more integrations, the incentives to stake will come more from the intrinsic value of making money by resolving cases than by the reward program. So it would be ok if speculators leave when the rewards from the program become very low. At that point, we expect to have many more cases. Speculators would sell their tokens to users who are actually willing to use them to solve disputes.

If the program stopped and people unstaked their tokens, this would reduce the security of the protocol and would reduce the likelihood of being adopted.

3 Likes

thank you very much for all your answers!
and congratulations on the proposal being approved. I hope the goals are met!

2 Likes

February Update on KIP-66: Staking Rate Progress

We are committed to monitoring and reporting on our progress towards increasing the staking rate as outlined in KIP-66. These updates will be provided monthly in this forum to ensure transparency and community involvement.

Staking Rate Overview

  • December 2023 Staking Rate: 27.5%
  • January 2024 Target: 29%
  • January 2024 Actual Staking Rate: 29.18% (Note: This rate is considered “artificially high” due to a minting error in January, with an adjustment expected in February.)

Minting and Rewards Distribution

  • The planned 12M PNK minting for January was delayed due to an issue with the Governor, resolved and executed in February.
  • In January, 0.998M PNK were distributed as rewards from the Coop Treasury, slightly below the intended 1.01M due to a calculation error (see the details below). The discrepancy of approximately 16k PNK (around $300) will be corrected in February’s distribution.

Correction and Adjustments

  • Correction: The accurate formula for January’s rewards should have been r(Jan) = r(Dec)(1+t-s(Dec)), not r(Jan) = r(Dec)(1+t-s(Jan)).
  • February Follow-Up: We will use the corrected formula with a target (t) of 30% and the staking rate (s(i)) accounting for the 12M PNK minted, expected to lower the staking rate from January’s 29.18%.
  • March Planning: The calculation will include a target (t) of 31%, focusing on native rewards from February, excluding the additional 16k PNK distributed for January’s gap.

Noteworthy Achievements

  • Between December 2023 and January 2024, there was a significant increase of over 13M PNK staked on the mainnet.
  • The adjustment in the rewards formula is part of a long-term strategy, with monthly tracking and updates to be shared in this forum.

Your insights and questions are invaluable to us. Feel free to ask questions or suggest improvements to this report.

2 Likes