Redesigning NO Rewards

About 87% of node operator rewards (or 60% of total inflation) is currently paid for collateral levels above 10%.This represents the biggest expense of inflation by a good margin. At the same time, there seems to be no good understanding why we are distributing rewards this way. The goal of this post is to examine some of the justifications put forward, exploring an alternative and hopefully sparking a productive discussion.

Bootstrapping the Protocol

People have argued in the past that node operator rewards are necessary to bootstrap the protocol. But does the current system actually incentivize desired behavior? I would argue that people are making a portfolio allocation choice between ETH and RPL based on their expectations about future prices. And a choice about how much RPL to stake and how many minipools to run with that portfolio based on liquidity considerations.

If someone is naturally not interested in holding RPL, NO rewards provide some incentive to buy a bit of RPL anyway in order to run minipools.

If someone is already looking to hold RPL, there is some incentive to create minipools to the point that RPL can be effectively staked. There is some disincentive to create minipools that put your collateralization below 150%, because it hurts your ability to unstake and thus liquidity of assets. Importantly there is no incentive from these rewards to create minipools that put you below 150%. One minipool at 150% gets the same rpl rewards as 15 minipools at 10%.

Overall, current node operator rewards do a poor job of bootstrapping minipools.

Security: Protection against MEV Stealing

So far RPL is not used like this. Maybe this will happen in the upgrade after Atlas. Assuming that’s the case, we should be selecting the ETH bond and minimum RPL requirement in a way that is safe against MEV stealing. If we do so, additional RPL beyond the minimum does not contribute to security because malicious actors would only provide the minimum.

Security: Protection against Black Swan Events

RPL collateral is only used after the node operator’s ETH collateral is exhausted. For the effective balance to drop below 16 ETH (or even 24 ETH with LEB8) and RPL to come into play, it would take a massive black swan event of mass slashings or validators being offline for a long period of time. If RPL is ever slashed, it will be slashed on many minipools at the same time. In such a scenario, the limiting factor on ETH recovered will be the available exit liquidity for RPL, not the amount of RPL staked. Currently the effective staked is ~7.2 million RPL. The available exit liquidity is less than 2000 ETH/120k RPL. During a black swan, we should expect that liquidity to quickly disappear in anticipation of massive amounts of RPL being dumped on the market through the auction process.

There is no scenario in which RPL staked at high ratios is adding security to the protocol.

Privileged Speculation

Valdorff has put forward the idea of maximum collateral being a limit on privileged speculation. I don’t understand why this is something the potocol wants and should pay for.

Alternative: RPL Rewards per Minipool

Rewards are split according to number of minipools eligible node operators have. There still is a minimum collateral requirement that needs to be achieved to create new minipools. You also need to be above the minimum at time of reward snapshot to be eligible for rewards. You can freely choose to overcollateralize to hedge against price volatility and dropping below the minimum.

This creates a direct link between desired behavior (more minipools) and rewards. Minimum collateral level, protocol security and implications for token value remain unchanged.

Impact on Minipool APR

With the current ~9k minipools we have, this would translate to ~73 [18859621*.035/9065] RPL per minipool per year. Combined ETH and RPL APR for a minimum collateral minipool would be 11.65%.

Currently combined APR for a minimum collateral minipool is 5.6%

Token Value

Some people might be wondering about the implications on token value. I believe that long term collateral is trending towards minimum and this proposal does not impact the long term collateral ratio. Fundamental token value is driven by the amount of capital the protocol is able to attract. This proposal makes better use of protocol spending and increases incentives for protocol growth. I am convinced that this would be very beneficial for token value.

I don’t think that many people would be dumping their RPL in response, because I don’t think that holding is about APR. Rather, it’s conviction about future value appreciation, which should only be strenghted by better tokenomics.

14 Likes

Interesting proposal i would support. The RPL stake has to be unlocked to allow existing stakers to switch to the new reward system. There should be an additional incentive for NOs to stake more RPL than necessary, e.g. higher RPL rewards or higher commission.

I agree that this approach would strengthen the tokenomics and fundamental value proposition of RPL and the RP protocol. That said, when you take away the “privileged speculation” premium (real or perceived), even as people simply rebalance by spinning up new minipools for the best yield, I think a lot of people would dump RPL all at once, rapidly pushing the price lower toward the fundamentals-supported price.

For the community to accept a change like this, I think there might need to be some kind of transition function where RPL stakers would continue to get rewards that gradually ramp down over a period of say, 1-2 years. There are a lot of ways to do it, probably none of them perfectly fair, but if there is a chance to meaningfully improve the fundamentals of Rocket Pool, I think it deserves consideration even if comes with growing pains.

4 Likes

This definitely makes sense, but I’d ask about the two things that are today related to the amount of RPL you have staked. One is governance votes. It’s unstated, but can I assume the proposal would also tie number of votes directly to number of minipools?

Second is the level at which one can withdraw RPL. It’s unstated, but what makes sense to me in this alternative is any amount over the minimum can be unstaked and withdrawn, is that what you’re proposing?

Assuming these two items are affected similar to what I wrote, I’d support this.

1 Like

I consider myself a countercurrent outside-the-box thinker, but I tip my hat to you. This would essentially take all the pain of price discovery and compress it into several weeks as overcollateralized NOs sell RPL to get minipools, dropping the ratio, and undercollateralized NOs have to buy RPL to stay at backing. The RPL floor then becomes rock solid, as nearly all NOs would have heavy incentives to buy as ratio drops. I think this is a reshuffling likely has to happen at some point, so earlier is fine in theory; in fact, i think this is how the tokenomics should have been like this from the start.
Where I disagree is whether token value will remain constant. Under current protocol NOs 30-150% collateralized, switching RPL-> minipool gives (approximately) +6% ETH and -9% RPL rewards (essentially a wash); under the new regimen switching RPL->minipool gives +6%ETH and +7.5%RPL or so. So instead of a bet on whether RPL/ETH ratio will increase, it becomes a bet on whether the ratio will double every 6 years, which even RPL bulls will have much lower conviction in.
Overall, my largest problem is that this RPL price drop seems so easy to front-run by anyone with liquid RPL, since all the NOs that it affects already have their RPL presumably locked until the moment the protocol changes.
I think I could only support this if NOs had something to do with all that excess RPL except sell (SAAS, UEB, other collateral, eigenlayer, etc). Could you spell out in a little more detail why you think RPL price will not be heavily affected in the short term, or if my numbers seem incorrect?
EDIT:my numbers were incorrect, and now i think they are correct.

5 Likes

Both good questions. I think that per minipool makes sense for rewards, because rewards should be about incentivising behavior.

I think you could argue that voting power should be about how invested someone is in the protocol and someone with more RPL should have more say, so I dont think it is clear cut for voting power.

Some people really seem to care about the locking mechanic as something that adds token value. I disagree and think this is ultimately inconsequential. I think fixing rewards will matter a lot more, so from my point of view whatever gets majority support is worth doing.

As I said, I dont expext RPL holders to sell based on this change.

It’s an interesting question regarding the nature of RPL (or what it should be going forward) beyond access to commission and providing collateral. It also goes to the question of what invested in the protocol actually means. Zooming out, I’d say that the number of minipools you operate is how invested you are in the protocol, basically the same reason we’d tie rewards to the minipool. I’m not sure why we’d tie rewards to one metric and governance to another, but will be interesting to see how it shakes out.

Agree on the lock. It’s a fairly temporary issue, which means if it provides any “value”, that “value” will last until Shanghai, IMO.

2 Likes

Yeah, it’s true that we are already not exactly using RPL as governance token like many other protocols do. RPL in the hands of non node operators gets no vote. Linking voting power to number of minipools would further separate governance power from token ownership. I’m not sure if that’s something we want honestly.

I would like to see rewards tied to overall validator/node performance. We have some large node operators performing quite badly, which brings down the overall RP return for rETH holders. At present they are rewarded regardless of node performance. We have the technology to measure such as it is already being applied to smoothing pool distribution.

3 Likes
reply to knoshua's comment above, so i don't jack the whole thread

If we are at an equilibrium now, why wouldn’t giving 12% yield incentive encourage many people to convert? Particularly as ~7.5% is in RPL so they maintain exposure. Isn’t that the point of incentivizing per minipool, to encourage people to run more minipools instead of holding RPL?

reply to knoshua's comment below, so i don't jack the whole thread
  1. If one has even a 30% conviction that ratio will 10x in the next few years, the rational decision would be to hold RPL and not run minipools. NOs are people who are on the intermediate spectrum between ETH maxi and RPL maxi; thus I believe they (like me) will migrate to the incentives, particularly as the incentivized behavior is good for the protocol and Ethereum. I suppose neither of us can argue with a belief.
  2. I actually am in favor of popping this speculative bubble so that we can link RPL value to its actual use. However, i think this proposal has no chance of proceeding at this point because most holders are not as hawkish as me, and to me it seems disingenuous to argue that this would not affect RPL price. Thus, I would wait until we have other use cases for RPL.

If you believe that RPL will 10x compared to eth, you wouldnt want to sell now just for marginally higher yield. As I said, I believe people are making a choice about how much eth and rpl to hold first and then decide what to do with their eth after. Per minipool rewards is about maximizing the eth that the protocol can attract. So instead of running the minimum amount of minipools necessary to make all of their rpl effective and doing something else with the rest of their eth, people have a reason to put all their eth into the protocol.

I would be in favor of such a proposal.
Just one quick question/thought :

Considering the majorly boosted APR, would we stick to a minimum collateral of 10% x pETH ?
LEBs would both benefit from the higher APR on Ethereum AND the higher APR on RPL if that system was to be adopted.

Seems like we could request a higher minimum collateral. One backed by solid research as to what it would take to provide enough exit liquidity for RPL in the case of a Black Swan event.

There would be a few options to consider if minipools are too attractive at that point:

  • Commission can be used to make reth more competitive with other LSDs and lower minipool APR
  • inflation changes could lower minipool apr and redistribute to other uses like grants and bounties
  • Collateral could be used to improve token value
1 Like

First of all, thank you for reaching out further into the design space with this analysis + suggestion. We have a lot to work with in terms of reward designs, so I’m sure we can find something better than the current system with its known issues.

If we can tie RPL rewards and minipool count together like this, it would be a huge fundamental upgrade for the tokenomics and the protocol as a whole. However…

I would sell, and I have to imagine a lot of others would, too. I don’t know how many, of course, but that’s at least one person. Obviously, this means I’d have to vote against it as-is.

But like I said, I do really like the suggestion, so if we can find a transition plan or some other way to work around perceived RPL devaluation, then it could be a really big upgrade. Maybe the protocol could buy out high-collateral RPL holders somehow? It would have to be at market price, of course.

3 Likes

@knoshua

I think the basis of your idea is great. Would you consider amending additional rewards for collateral over 10%, but at a much lower rate? We should expect everyone to have a reasonable price cushion there, and hopefully it grows naturally over time!!!

A potential set of amendments:

So inflation rewards for NOs are currently 3.5% of total supply. Let’s say 50% of the rewards go to the first 60% collateralization ratio (of protocol ETH of course). This is what you’ve suggested so far, but for only part of the available awards. I think it would be okay to incentivize a small buffer zone. So the next 30% of rewards could go to RPL collateralization range of 10-20%. The last 10% can cover the remaining RPL collateral up to max of (what’s looking to be) 150% NO-supplied ETH.

Reasons I’m interested in the amendments:

  • Covers us through LEB4s (4x150%=6eth) > (28x20%=5.6eth)
  • maintains rewards at all levels we’re currently receiving and WILL BE receiving post-atlas
  • incentivizes reasonable buffers but not hoarding (I’m a hoarder myself)

Thoughts?

What do you mean by that? Are you saying there is a benefit to have more than 10%? What is that and how do we value it?

Practically, people are going to collateralize at minimum + small buffer to ensure rewards. Everyone hope RPL go up. This slightly positive pressure will mean most people truly have a collateral that approaches 20% (estimation, could be anything) despite only trying to have, let’s say, 12%. Leaving this locked up is good for a multitude of reasons. I don’t think you want people dumping pools just to lower their collateral ALL THE TIME.

Naturally more people will allow it to grow and use the gains to hopefully run additional pools when it suffices. Also, if 10% minimum collateral is desired, and we theoretically rely on RPL collateral more as time advances, then someone desiring a safety margin slightly higher than that is probably a good thing.

As for how we value it, that’s up to a discussion. I’m simply suggesting a significantly tiered system

While I agree with your analysis I think it is definitely too soon for such a change. Long term collateral will trend towards the minimum but that is far in the future.

Your reasoning is very idealistic. People will definitely sell and RPL value will be destroyed. That is how the real world works unfortunately.
Changing the tokenomics often will also deter solo stakers (many of whom are already reluctant to buy RPL) from using Rocketpool.