An argument in favor of protocol ETH

[Draft: publishing early due to time sensitivity. Under active revision.]

tl;dr

My attempt at a tl;dr:

  1. Everyone agrees that the best idea is the one that is minimally disruptive to tokenomics. There is significant disagreement about which choice represents that, and below I argue that adopting protocol ETH is the most minimally disruptive option available.

  2. RPL’s primary purpose is to collateralize rETH, referred to here as protocol ETH. If we accept that increasing the minimum is appropriate as we’re now collateralizing more protocol ETH, it is logically consistent to also increase maximum collateral so as to keep the same 10-150% collateral backing per unit of rETH.

  3. If there are concerns about the efficacy of RPL as collateral due to liquidity challenges then governance should address liquidity rather than changing tokenomics.

  4. Lowering maximum collateralization per protocol ETH sets the present that RPL should not benefit (in terms of structural demand) as Rocket Pool improves its capital efficiency. My position is that it’s reasonable and appropriate for RPL to benefit from our growth, to at least some degree.

  5. Voting power concentration is not a valid concern because of quadratic penalties and a known attack vector (sybil) that is much more capital efficient in gathering voting power.

  6. Everyone seems to agree that whatever change we implement will possibly be replaced in the near future after longer, more considered research and discussion. In the meantime this option is the minimally destructive to the tokenomics and is appropriate to adopt in order to “do no harm”.

Framing of RPL’s role

This is a principled argument, so bear with me.

The stated benefit to the protocol of staked RPL is to act as a form of collateral. Therefore, it is logically consistent to express the allowed collateral range of RPL in units of protocol ETH (rETH, effectively), and specifically not units of minipools or units of node operator ETH.

One could equivalently describe the 3 options here in this alternative way, taking the perspective of rETH collateral backing:

  • Node Operator ETH: 10%-50% protocol ETH (+exception made for existing 16pools)
  • Fixed ETH: 10%-100% protocol ETH
  • Protocol ETH: 10%-150% protocol ETH

Choosing to denominate the collateral in units of minipools or node operator ETH is just a reflection of our community bias towards node operators, which make up the vast majority of our community and have 100% of the voting power.

This perspective is consistent with the consensus we reached on RPL minimum collateral – we are allowing node operators to mint 3x more rETH, and asking them to stake 3x more RPL (again, per unit of protocol ETH). By doing so, we are maintaining a 10% minimum collateral backing of RPL. By the very same logic, we should apply that same range to the maximum and accept 150% of protocol ETH as the established precedent.

Tokenomics

The established maximum until now has been 150%, and I would argue that the only way to leave the RPL tokenomic value intact is to preserve 150% protocol ETH as the maximum.

Here’s an example to illustrate the point.

Let’s say we take the position that RPL collateral should range in units of node operator ETH. If we could magically convert our entire rETH supply from 16pools to LEB4s, then the RPL collateral would be limited to a maximum of 21% (source). Currently average collateralization is 80%.

By reducing the room for “privileged speculation” we are destroying any demand for RPL and not letting RPL benefit from any protocol growth achieved through LEBs. In fact it’s a 4x relative reduction in demand.

Capacity goes up, absolute RPL demand stays fixed, relative RPL demand goes down.

My position is that it’s reasonable and appropriate for RPL to benefit from our growth, to at least some degree.

My understanding is that our goal with LEBs is to introduce as little tokenomics changes as possible. I agree, and would argue per the point above that that means attaching the limit to protocol ETH.

Arguments against and justification in favor

High collateral RPL is not as useful to the protocol. There are diminishing returns on the value of extra collateral. Why should we pay for this?

I agree. There are diminishing returns on excess collateral. I think we should research this and have an opinion what the utility vs. collateral graph looks like. Since the protocol should ultimately pay for utility, we should design the rewards structure around that. I expect it would look like something quadratic and not linear as it is today. I support this outcome.

Also, there are other orthogonal questions here that would affect RPL’s efficacy as collateral, such as:

  1. How can we increase the effectiveness of perceived “excess” RPL in the protocol? UEBs?
  2. How can we better position RPL as useful collateral? Can we guarantee it can be used to cover losses?
  3. How can we improve RPL liquidity, which will bolster RPL’s role as collateral and able to be more effectively used in liquidations?

This is a separate topic entirely from choosing the maximum collateral setting.

How do we justify allowing 3x more RPL to be staked (and earning 3x more), when it’s likely not adding 3x more value to the protocol?

One could argue that 3x more RPL has 3x more responsibility, insofar as it is collateralizing 3x more protocol ETH. Still, this is likely overpaying for incremental collateral value and we should revisit that as per the quadratic rewards point above. But since many agree that this topic will require further research, the goal in this vote should be to do minimal harm in the interim. There is disagreement about what that means, but I believe that from a tokenomics position, a 150% protocol ETH maximum is the least disruptive.

Doesn’t RPL still win even if I accept your position that it weakens the tokenomics?

Yes and no. RPL price is a function of protocol growth x tokenomic value. Ultimately I view reductions in collateral maximum as sacrificing more value than will be gained through growth, and that not being a tradeoff that needs to be made.

Doesn’t this centralize RPL voting power by creating a higher ceiling?

This is not a real vulnerability from a governance perspective, because of our quadratic voting. An example:

If somebody is conducting a governance attack, they could either a) stake 3x as much RPL and get +73% more voting power (quadratically penalized), or b) divide their RPL share across 10 nodes and get +216% voting power (quadratically boosted). Since b) is cheaper than a) and gives 3x more voting yield, it is the obvious path for governance attacks.

Why is RPL value preservation so important? This isn’t an important principle that we should try to protect. Should we not seek to limit its role as collateral in favor of ETH, which has much greater liquidity?

Our treasury is denominated in RPL. Our ability to do marketing, pay for the oDAO, do liquidity incentives, etc. is all denominated in RPL. Our rewards structure is denominated in RPL. Our community has skin in the game in the form of RPL.

If we’re not going to have “RPL value preservation” be a principle then we should just drop the token entirely. RPL success and appreciation is critical to our success as a project and we should embrace it as such.

Additionally, making changes to reduce RPL’s value proposition from tokenomics is doing a disservice to people who made choices about staking and asset allocations under previous assumptions. One could argue we have at least a mild responsibility to not adopt changes that undermine RPL’s value.


Thanks for reading.

9 Likes

In case anyone else wonders about this–the rETH framing does not account for additional collateral that a NO stakes in the form of ETH, but as LEBs get smaller and require less ETH, that factor is minimized. There is an inverse relationship between the rETH provided by a minipool and the ETH collateral backing it.

First off, I disagree with your re-writing of the 3 options. The flat fixed ETH option was designed specifically not to scale with either side of the pETH or nETH debate. This is important as the equation gets more extreme as we reduce ETH collateral.

  • Fixed ETH: 10%-100% protocol ETH

is very different from up to 24 ETH of RPL per minipool long term even though they align for now. The flat cap recognizes that there may be an asymptotically approaching 0 ETH bond which would create an unbounded RPL stake cap. As such, it abandons the charade altogether. The reason why the min and max were assigned what they were was because of calculations based on different values.

I disagree here. It is logically consistent if the reason we limited it to 150% was related to security the way the 10% floor was set (more protocol ETH to secure → more bond required). This as we know is not true. The floor was set by security concerns and the max was set by incentive concerns. There could have been an unlimited amount of RPL stake allowed but 150% set the limit so that gov power and inflation did not centralize. As such, the logically consistent thing would be to maintain the same analyses going forward – security for the floor, incentives for the max.

Prima facie, the change that carries the least tokenomic change would not multiply the RPL value 3x. It would maintain value if all effective RPL remains effective and none becomes or leaves effectiveness even with full migration from 16 to 8.

This framing is misleading. We pay for the responsibility in the floor 10% inflation. Individuals who choose to add more collateral are not increasing their responsibility. The decision on if the responsibility is worth it or not is decided at the floor, not the max speculative point.

It was irrational in my view for someone to try and predict what the future of RPL tokenomics will be when ETH bonds drop to 2/4 based on what they were at 16 ETH. That was a user risk based on imprecise language since we can’t agree on what the collateral limit basis even is. People trying to predict the future isn’t the responsibility of the DAO. Getting the best tokenomics is.

I want to add that I like the concern you bring up with the long tail for nETH scaling. We don’t want to limit our collateral to 10-21%, however, I argue this is bad because it’s too restrictive on user choice. Security wise this would be fine. What I believe is important in this vote is that we acknowledge the floor is set by security concerns, the max is set by speculation concerns. and that a simple scaling equation will not work long term.

5 Likes

TL; DR section

1 - :ballot_box_with_check: I agree we all want least disruption. I believe NO ETH scaling is the path that achieves that best.
2 - I don’t agree with this. While we use that word, I don’t think more than the minimum actually serves as collateral. If we need a larger pledge in the event of default, that requires a greater minimum. I’ll allow there’s some wiggle room here to allow the price to drop and stay above our preferred level, but I don’t think 15x is remotely in the ballpark.
3 - The tokenomics as they have existed include the context as it has existed. I don’t believe “liquidity challenges” are realistically addressable (see liquidity section below).
4 - I don’t agree. I believe RPL value benefits from greater rETH TVL. I do not think we’re trading off any RPL value at all for that TVL.
5 - :ballot_box_with_check: I mostly agree, though “this isn’t the worst thing therefore it’s not bad” is a shaky argument.
6 - :ballot_box_with_check: We may or may not replace this setup with further researched options. As you said in (1), we are trying to minimize change; I believe NO ETH scaling is the path that achieves that best.

Framing of RPL’s role

This section relies on the use of the word “collateral” instead of looking at how RPL actually functions today. The rest is built upon that, so when the initial premise fails, it all fails.

Tokenomics cares about reality, not intent.

Tokenomics

Honestly, I’m unsure I get the point that’s being attempted here…
It seems like max RPL is being seen as a demand engine? It is not.

RPL for ETH commission and RPL yield is more valuable by definition, so buying for just RPL yield is irrational… unless you’re speculating that price will go up. If you believe price will go up with respect to ETH, then you don’t need another demand engine.

Arguments against and justification in favor

(A) High collateral RPL is not as useful to the protocol…
I agree, although I’d say it has near-zero value. It can’t be used for MEV theft, because thieves won’t provide extra collateral. It can’t be used for a black swan (see liquidity section below).

The argument presented says “we have little current value”, “maybe we can figure out a way to get value”, “figuring out if we can get value is a separate topic”.
I strongly disagree. If we can’t get value today, we shouldn’t implement something like this today. If someday we do get value (eg, UEBs), we can alter our stance.

(B) How do we justify allowing 3x more RPL to be staked…
The argument here has 2 parts:

  • Suggest maybe there is 3x more collateralization happening
  • Assertion that 150% protocol ETH is least disruptive

I agree on the first bullet, but that refers to the MINIMUM. The maximum doesn’t serve as collateral in practice.
The second bullet is just an assertion.

(C) Doesn’t RPL still win…
This is a strawman. I don’t accept the position, so further logical steps from that premise don’t make sense.

(D) Doesn’t this centralize RPL voting power…
I mostly agree here. We certainly have a large vulnerability in node splitting. Still, “this isn’t the most harmful” isn’t an argument for at all, just not an argument against.

(E) Why is RPL value preservation so important?..
I fully agree here. I believe we need to preserve RPL value, and believe scaling the max with NO ETH is the most effective way to do so.

Bonus section: Liquidity, or why “extra collateral” has no value for black swans

Per https://discord.com/channels/405159462932971535/774497904559783947/1034451062435889193, we have liquidity for only ~2000 ETH of RPL. Liquidating half the current minipools at the minimum level would need over 7000 ETH. We’d need to 3.5x liquidity to fully use our minimum collateral in a black swan event like this (which is where amounts above the minimum might play). To use the average collateral, we’d need to ~28x liquidity. Improving liquidity this much would be a massive undertaking.

All of that is assuming that all liquidity is out there waiting for us to use it and nobody else uses it before our auction. This is highly improbable. Much more likely is that the market is far more nimble than our auction, so we’d currently find near-zero liquidity available, as RPL speculators would have used it before we could. Fixing this would be nearly-intractable. We’d need to, eg, get folks to bond liquidity with us and be able to turn off external access to that liquidity in the case of a black swan.

5 Likes

Very well put - and true.

I 100% agree. And have seen no evidence for a difference in RPL value.

I’ve posted this several times in discord nowň

I’m looking for a reason to add flat to my vote. The bar I’ve put up is an argument for pETH/flat that isn’t refuted by “liquidity” or “people can exit and re-enter”.

The arguments for increasing value tend to either fall to the latter, or assume irrational speculators (self-refuting).

Thank you @Marceau for writing this up. I think that there are valuable insights and valid concerns here. IMO, the strongest arguments for this proposal are

  1. RPL price stability
  2. RPL value preservation

RPL price stability

I agree that this change to tokenomics will incur short-term volatility as we are removing friction for selling NO staked RPL. There is short-term uncertainty and I think that this is a bullet we will need to bite eventually, for the average collat. to reach market equilibrium.

RPL value preservation

In the long run, I think that RPL value will be primarily driven by the 10% min collat. on protocol ETH, and NO “friendliness” or user experience. After growth tapers off, we can expect desired collateralization to fall due to lesser RPL speculation anyway. Lowering the effective average cap allows more effective discovery of price, and imo this can be a plus for NOs looking to come in, if they believe that RPL price is efficient and not being artificially held up.

Improving NO experience by providing flexible risk management options (RPL collat) is critical for onboarding new NOs, especially in the face of competition from other pools, once withdrawal risk is reduced.

It is important for min. collat ratio to scale with protocol ETH. But I will be voting to scale max. collat ratio with NO ETH / flat as I believe it’s the best thing for our product, protocol growth and, overall RPL value preservation in the long term.

[I] have seen no evidence for a difference in RPL value.

What would this evidence look like?

Would it look like a RPL/ETH graph which begins a consistent drop around the time when this discussion started?

As you can see from my highly scientific drawing, the RPL/ETH ratio has decreased consistently since this topic was brought up. Perhaps it’s noise, or perhaps we need more data to draw a truly definitive conclusion (the outcome of the vote will be informative), but the reasoning presented here for fundamental valuation changes seems sound, and the evidence so far shows that RPL/ETH market value is being affected just by the discussion itself.

2 Likes