Recently, Uniswap Delegate Reward Initiative - Cycle 2 has passed and top 15 delegates have been selected.
There are many thoughts and changes including:
We want to open the floor for different ideas , suggestions, as well as concerns. Specifically, any case studies or theories or data would be quite helpful to the discussions.
Thanks Doo. I’ll start my list here of things it’d be nice to discuss as we’re thinking about what Cycle 3’s structure looks like. Will add to this comment as ideas come to mind.
Hi everyone, We’d like to introduce an idea for cycle 3. This idea isn’t perfect, but it could be helpful to try. At least by sharing it here, someone might have an idea to improve upon it.
First, let’s recap the problem we’re trying to solve. As of cycle 2, we have 16 eligible delegates who will receive pay so long as they vote on most proposals and receive an additional bonus if they provide their rationale. While this has increased the number of active delegates and made it easier for the DAO to reach quorum, it doesn’t necessarily encourage delegates to perform.
Incentiving delegates to “perform” (provide value to the DAO) is difficult because “value” is subjective. That is why companies traditionally leave compensation up to HR, a compensation committee, or your boss.
If we want delegates to be incentivized to perform, we must introduce a subjective component into our compensation structure.
How this would work is we would have a pool of eligible delegates as we had with cycles 1 and 2, but rather than having fixed compensation, delegates would rank the performance of all eligible delegates each cycle, and the cumulative scores would determine how each delegate was ranked and, thereby, the compensation they would receive.
Of course, ranking delegates can be uncomfortable, so we propose a system that protects voters from their ballot submission. Here is how it would work.
Because the ballots are submitted publicly, delegates can verify that their ballot was counted. Still, because the voter’s name and wallet aren’t associated with the ballot, the voter’s identity is protected. This leaves two trusted elements: the application’s responsibility is to ensure the same voter doesn’t vote twice, and the second is not to store any information connecting voters and their ballots.
To incentivize delegates to perform, the top delegates should be paid significantly more than even the 5th and 10th best-performing delegates. However, to provide some base compensation, all delegates below 10th will be compensated the same. As for why some ranks pay the same amount, as the rank decreases, we suspect delegates may feel the amount contributed between two delegates might be very similar.
| Rank | Pay | Rank | Pay |
|---|---|---|---|
| 1 | $30,000 | 9 | $15,000 |
| 2 | $27,500 | 10 | $12,500 |
| 3 | $22,500 | 11 | $7,500 |
| 4 | $20,000 | 12 | $7,500 |
| 5 | $20,000 | 13 | $7,500 |
| 6 | $17,500 | 14 | $7,500 |
| 7 | $17,500 | 15 | $7,500 |
| 8 | $15,000 | 16 | $7,500 |
As for implementation, the idea is purposely simple so that if delegates choose to experiment with this in cycle 3, there is enough time to build the MVP.
Here is a quick look at how a basic interface (delegate names entered randomly) for an MVP could look (thank you v0). Delegates would click and drag on each row until they were ready to submit.
This is interesting but have a few reservations:
The overall idea of having variable compensation is something being experimented in several daos, but always with a cap. This idea is intriguing, but I would tie it more to the type of outcome you want in governance: if the goal of the delegate reward incentive program is to have more delegates, more involved, with an higher democratization of governace, usually a compensantion cap serves the purpose. Instead an higher cap with a bonus is somehow a structure more suited for service providers.
Doesn’t mean we shouldn’t plug a measurement for quality here tho. Worth iterating upon.
Agree that this is an interesting proposition but also echoing @_JoJo’s concerns.
I think the easiest and most non-biased way is to have a set of metrics you evaluate against in a more straightforward and objective way. Delegates voting on each other seems ripe for non-meritocratic back scratching imo. If you nonetheless wanted to experiment with delegate voting in the compensation scheme, anonymous voting could potentially lead to more honest assessments by removing fear of backlash. I think the voting pool could probably also be expanded to include the wider community as I don’t see any downside to this (in reality why shouldn’t they have a right to express their opinion on this?), if anything to dilute any potential impact from the risks of biases/collusion.
Compensation caps seems reasonable to maintain fairness, as does having a base compensation plus tiered bonuses based on performance metrics for extra incentives to those that go above and beyond. Another idea if you insist on having delegates voting is also to tie in the performance to weighted voting to give more say to delegates who consistently perform well.
As with any effort seeking to provide structure to a given process, I think agreeing on the goals/type of behaviour we want to see from delegates first is a good starting point. I know there’s some work being done to outline a set of guiding principles etc for Uni delegates, perhaps that could be relevant here.
While I agree to consider better rewarding delegates who contribute the most, I agree with @_JoJo’ concern that this mechanism would be easily gameable and could encourage delegate pacts to vote each other up in the rankings, which would be undesirable and would conspire against best governance practices. I support the objective pursued but I do not agree with the proposed way of trying to achieve it.
Current delegates are expected to participate in calls, vote on proposals, stay updated on delegate news, and align their interests with UNI holders. My primary question regarding this program is: what is a delegate supposed to be?
Developers, liquidity providers, and traders are the most important value drivers for the Uniswap protocol. These stakeholders are closely aligned with UNI holders; however, the majority of delegates who receive rewards and vote on proposals are neither builders nor liquidity providers, and many do not have a personal stake in UNI. Most delegates derive their voting power from one or two large venture capital firms, many of which have vested interests in competing protocols. For instance, Coinbase Ventures holds stakes in both Aerodrome and Uniswap. Coinbase has actively promoted Aerodrome for cbBTC on Base, marketed Velodrome, and has not engaged in Uniswap governance. Interestingly, many VCs either do not delegate their voting power or refrain from participating in governance voting, possibly due to legal concerns. This leaves most decisions in the hands of just one or two VCs, despite their delegation to aligned groups.
The perspective and trajectory of Uniswap depend on a very small number of individuals. If anything, payments to these delegates should come from the VC-delegated voting power. In this way, the influence of behind-the-scenes players could gradually diminish, allowing professional delegates to operate based on their own UNI holdings.
While we may glamorize the importance of delegates who are paid to vote, this is inconsequential to the main risk facing the protocol: UNI holder apathy towards voting, engagement, and delegating to value-aligned representatives. Until the most active delegates include Uniswap application developers, traders, and liquidity providers, I believe this program in its current form should not exist. Moreover, the pace at which delegate compensation is growing is concerning. If this trajectory continues without significant changes, the program may become entrenched and resistant to improvement.
Suggestions for the Current Structure:
Implement Value-Aligned Points for Eligibility:
Expand the Number of Delegates Eligible for Funding to account for the above criteria inclusion:
Additionally, @eek637 suggestion for a vesting structure is a necessary improvement for value alignment of current cohort. I agree with @GFX that it may be premature to implement tiered voting at this stage, it will be needed in the future and experimentation around this should be started now. We need more delegates with diverse backgrounds, and delegators to hedge against collusion in tiered voting. @_JoJo 's concerns about collusion among the current cohort of delegates highlight why this program is moving too quickly in the wrong direction.
Right now, delegates are blanket compensated without any guide for being “good” or “bad.” If we want to encourage delegates to go above and beyond, then we’re going to have to introduce a subjective element to compensation. We like to think that Uniswap delegates are robust and honest actors, and many delegates have been active for years, but you are correct that there is a possibility delegates could collude to enrich themselves.
The beauty of this program is that we can try it for one cycle, judge its performance, and modify it thereafter.
Each delegate will have their view on what outcome they want. The nice part about a simple program like this is that delegates are free to develop their own opinion on the behavior they wish to encourage.
As for the compensation model, we’re open to suggestions, but we would like to highlight that the base compensation is currently $7,500 per cycle with a total program cost of $242.5K, which is $45.5K less than the current program.
The proposed solution is anonymous voting.
That is a fair approach. If you have something in mind, please propose it. In our experience, creating objective metrics to judge delegate performance is very hard.
Our solution could work with token voting if governance prefers it, but for the initial pilot, it would probably be easier to keep the program for delegates.
We are on the same page here: gunning for quality and not just activity should be one of the most important main goal. And totally agree on the fact that current delegates are good, robust actors.
But, assuming the program will expand in future, is just too easy to have a collusion situation, especially if someone just spin up 3-4 different “delegate entities”. Likely this would create the byproduct of also having a strong set of delegates, well respected because they get more money (yeah i know), not aligned with the overall goal of the dao.
There are a bunch of ways I guess to steer behaviour in a more qualitative than quantitative way, and I guess it partially comes down to what do you want this delegate reward initiative to be. Depending on the fact you want either more new delegates, or current delegates more involved, or reactivate dorment delegates etc, there are different things that could be done. In general tho, if we democratise this evaluation to the people who are also receiving compensation we will always have -ve outcome. Personally i think that either this evaluation becomes centralized, or is outsourced to a set of large people not part of the program.
As others have commented, voting on each others performance has its own problems. The way I see it, for the approach to work as intended, almost all of the voters must be unwilling to collude, not even if “innocent” ways like preferring to vote for their friends.
Traditionally, crypto protocols have had different requirements for the number/power of honest actors for the system to function properly:
Introducing mechanisms that assume N-1 out of N honest actors seems like a regression, and almost against the crypto’s ethos. I think when similar mechanisms have been tried, it hasn’t ended well. We have seen protocols like EOS with voting-based delegated PoS become vulnerable to cartelization.
Moreover, even if we roll with the assumption that almost all of Uniswap’s top delegates are unwilling to collude (which it may well be at the moment!):
An unrelated point is that designing such an anonymous voting mechanism seems like a non-trivial technical challenge on its own, especially if you’d want the usual properties like auditability and trust minimization in the operators.
In short, here are some of my own thoughts / recent position:
In sum - add more delegates that are closely aligned with the interests of the protocol, and don’t shift the focus on compensation for activity on its own. (Other than voting activity, to be clear.)
Exciting to see many ideas and discussions. I do feel like it’s one of topics where there are valid points for counter points for each one. For example, if the goal is to have diverse delegates and hedge against collusion, then raising the bar with strict requirements or directions will reduce the new delegates and increase the chance of collusion.
Governance and Governance Security is a good example of public good. If the governance came to a halt or decline, it’s going face malicious proposals passing similar to what happened in other DAOs recently or at least it will become more difficult for beneficial proposals to pass. Having healthy governance also contribute more positively to resilience as well as other potential risks. Uniswap token, when launched, could have technically gave all of the treasury to Uniswap Labs or later to Uniswap Foundation and removed DAO governance. But of course they didn’t because they also knew the benefits of having such DAO governance.
However, the nature of public good would also mean that it will likely to be underfunded compared to their benefits. But of course, it can’t be priceless (Sorry but Mastercard commercials lied to you).
There can be several back-of-the-envelope calculations. But is it worth $1,152,000 per year? and clearly this depends on how much is governance security worth in Uniswap DAO. An often discussed model is if the malicious proposal is able to hold Uniswap’s $1billion+ treasury in hostage, but even we believe that’s an extreme case. More reasonable case could be for example, what happens crucial voting decisions such as funding Uniswap Foundation is not meeting the quorum and has to be delayed for several months till it passes (which does happen in other DAOs). If we believe beneficial proposals can bring benefits bigger than their budgets + $1.1 mil then it can be considered as funding that’s worthy.
Another way to think about is UniswapDAO’s previous funded public goods. For example, DEF received over 15 million USD total from the DAO. So is DEF worth 15 years of Delegate Reward?
But also to be fair, Ironically, DAO treasury itself also can be considered public good. So maybe this should be bigger conversation on how we should we think about DAO treasury and spending.
Think there are easy frameworks for this.
Is not really about important proposals, re: funding uni lab. Is more about tail risk.
IF (and is a big if) we collectively think the reward initiative is an effective stop of malicious governance attack on the uniswap 1B treasury, then a 1M spending per year or 0.1% is an extremely good spending of capital. And can likely increase ten folds (ok not ten folds) and still be a good investment.
But only if the program is an effective countermeasure, and there could be concerns. Foundation is: how much is the collective weight of votings of delegates enrolled in the program + “good” aligned parties vs how much is the cost of a governance attack?
As it is now, without looking too much into numbers, it feels like we might both need more aligned delegates but also a bigger voting weights for the current delegates. There was a few months ago an initiative to increase the voting power of certain delegates, most if not all of which are now in this program. There is a scenario in which this could be replicated and expanded more.
I am quoting this, which is something a bit outside of scope of the current discussion, because as it is now uniswap would see a merit in having both initiatives (increase delegation power of active delegates + reward initiative) run in tandem, in coordination, to leverage positive results one against each other.
But maybe I am just ranting here.
We believe this is a good idea as it encourages delegates to perform at their best in order to secure a good ranking. While we recognize that some level of collusion may occur, with delegates who share similar political views ranking each other higher, we don’t think it will significantly impact the overall outcome. Even if one delegate ranks their friends higher, the remaining non-affiliated delegates would still be ranked based on the value they provide. This ensures that delegates who genuinely contribute value will have a chance to remain in the program, even if they don’t secure top compensation. Well-known delegates will likely receive the highest compensation, but that’s fair given their contributions to the DAO. Overall, we see more pros than cons in this system.
The one suggestion we would make is to increase the number of slots significantly, to something like 50, similar to Arbitrum’s delegate program. Delegates ranked 17 to 50 could receive a smaller amount, such as $1,000 worth of UNI tokens. These delegates wouldn’t be selected by vote, but instead based on meeting basic criteria, like voting and providing rationale for their decisions.
This adjustment would give new delegates a real opportunity to compete, as the paid delegates would be more incentivized to contribute and retain their slots. Without this, the same 16 delegates would likely dominate the top positions. By increasing the budget by around $34k per month, we could significantly boost competition and delegate participation. This smaller compensation would help cover costs and motivate newer delegates to learn the protocol and add value, giving them a chance to move up into the top 16 in the future. We believe this would be a great idea to try, and if it doesn’t work, we can easily cancel it. Even if only 10% of these 34 new slots end up developing into strong delegates, it would still be worth it in our view.
I genuinely don’t think Uniswap Governance should increase more delegate compensation for Cycle 3. Also that’s equivalent to around 30% increase.
Yes, there was delegation to underrepresented delegates, which in various metrics, showed great success. The issue, of course is if we ask governance for additional delegation, how much is ideal? Also, how do we prevent potentially delegates growing too powerful that are detached from delegators . Many questions to consider
It’s definitely a significant percentage increase, and we also don’t like the idea of increasing the DAO’s expenses. We’re suggesting this because we see it more as an investment than an expense. This could attract a significant number of new participants, some of whom could end up creating a lot of value for the DAO. It’s also worth noting that a 30% budget increase would result in a 200% increase in the number of delegates.
That said, we understand the concern about raising the budget. What do you think about reallocating the budget for positions 15 and 16? Instead of paying $12k for two delegates, we could create 10 additional slots and pay $1k each. This way, we could test the idea without increasing the budget.
It’s a balance but I really wouldn’t want delegate compensation to be like bounty or airdrop equivalent. While I disagree with @Userisky 's view often, do agree with his suggestion that delegate compensation should focus on quality or alignment
you are totally right and I am on the same page on your question.
Mine was more a qualitative idea, in the sense: if part of the reward initiative program goals are to shield the DAO from governance attacks, working in tandem to a second round of redelegation could definitely solidify this stance, because you will have not only delegates that are compensated, but that also have a weight in their voting.
Again, more about the goals. And if that is the case, it bears as you said the question about how much in total, how much each, what is the way to split, etc. No single answer to this but top down approach probably is the best approach.
Want to also take a stab to this. How do you define the alignment of a delegate? is a tricky question as well, because technically any vote as long as comes from a distributed set of delegates should be, well, the alignment of the dao. Kinda makes me go back to the main overarching goal of this program. Really think I need to do some digging here.
