Minimum Collateral: 10% of protocol ETH (as previously discussed)
Maximum Collateral*: the greater of (A) 150% of NO ETH** or (B) 20% of protocol ETH
*Note that max collateral didn’t have a strong consensus for LEB8s. See my comment below.
**This should count NO credits as well
Examples
Formula for max effective RPL stake
Minipool16
LEB8
Hypothetical LEB4
Hypothetical LEB2
Current RPIP-8
1.5*Protocol_ETH
1.6-24
2.4-36
2.8-42
3-45
This proposal
The larger of: (1.5*Node_Operator_ETH, .2* Protocol_ETH)
1.6-24
2.4-12
2.8-6
3-6
Rationale
Maximum collateral currently serves as
a cap on “privileged” (inflation protected) speculation
friction on rebalancing
a cap on voting power per ETH
All of these are best served by scaling with NO ETH. At hypothetical extremely low LEB sizes, this could start damaging the “secondary” collateral purpose, which is why we have Part B for the maximum.
This change is small enough to have a shot at making it to Atlas if we have strong consensus.
Rationale with a lot more words
The Maximum Collateral didn’t get much discussion during LEB8 discussion (in part my fault as I was trying to push something through with consensus on the most critical bits).
As collateral threads have popped up recently, I’ve been thinking about what our current design does at 16 ETH. With 1.6-24 ETH worth of RPL being effective. Note, I’m not talking about what’s “best”, nor what was “intended” – I’m talking about what it currently is. I think it’s very important not to change the rules of engagement on folks after they’ve opted in based on a particular understanding, which is why I’m putting this emphasis on what is. There can be exceptions, ofc, if inaction would be devastating to protocol health etc, but for the most part the rules should be stable.
So here’s what I’ve (eventually) come to:
The minimum end is dictated by (A) RPL’s role as secondary collateral and (B) the modeling done for RPIP-8 that showed that in a rETH-demand-limited world scaling this with protocol ETH best kept RPL value unchanged
For the collateral role, it behooves us to reward additional collateral to a point so that price volatility isn’t as impactful to the protocol. This is where part B of the maximum in my proposal comes from.
The maximum end, I struggled with for a while. I decided it was serving as (A) “privileged speculation”, (B) friction for selling RPL, and (C) a voting power cap.
A: Naked speculation (outside of a node) is subject to loss from inflation. Inside a node it’s protected from inflation to some degree (and yield can even outstrip inflation, as it does now). For the speculative role, it makes sense to scale with NO ETH. The creation of LEBs doesn’t really mean we need to privilege more speculation.
B: For the friction role (which disincentivizes casual exits to rebalance RPL vs ETH), it makes sense to scale with NO ETH, since that’s what you’re rebalancing against.
C: For the voting power cap, we want voting power to be more expensive than just RPL by adding ETH to it. Thus it makes sense to scale with NO ETH.
This section is where part A of the maximum in my proposal comes from.
Note that no additional RPL will be able to leave during or after swapping to LEB8s - this proposal doesn’t cause immediate sell pressure.
A few reasons I’m against. I can’t rationalize this into being net-positive given the points below. Also, the sentiment in the previous thread leans opposed AFAICT. Is there a reason to think that the tweak suggested here (dynamic max vs. period-reduced max) would be more favorable?
This adds unnecessary (IMO) complexity to an already complex collateral system. We should err on the side of simplicity wherever we can.
Since any change involves change management risk and dev time opportunity cost, the benefits need to be significantly work it — we should continue to keep a high bar for changes we introduce. I simply don’t see a strong enough benefit here (or really any benefit). Basically I’m echoing this sentiment.
It limits the efficacy of UEBs, if we decide to introduce that in the future. There are ways that the protocol can make productive use of this “excess” RPL. I’d rather see us take the perspective of “Let’s find ways to make better use of excess RPL” instead of “Excess RPL is useless so we should lower limits and give some back”.
In general I’m hesitant to bundle other tokenomics changes with LEBs since that could potentially multiply risk. I don’t think we fully know what a post-LEB world will look like.
Being honest, this limits RPLs value capture, which will have secondary impacts on things like pDAO budgets.
From a growth and marketing perspective, reducing the cap will directly limit maximum NO yield, which makes RP potentially less appealing for new node operators.
20% seems somewhat arbitrarily chosen. What’s the reason for this specific magic number? Why not 30%? 15%? Can we justify this with security or research?
I re-read the previous thread to catch up and also to collate sentiment. The mechanics are different in this proposal (dynamically adjusted vs. planned reductions per-period) but the justification of being for or against should transfer cleanly.
Apologies for any miscategorizations – feel free to clarify if I misread.
[Edit: Deleted image as this thread has gotten enough discussion to stand on it’s own]
I was opposed to the change the last time a post about this was brought up in the forum, and I’m against this one too. @Marceau did a really great job of summarizing some of the reasons why he’s opposed to it, and I would say his points 1, 4, and 5 make the most sense to me, but I also agree with the rest of his post.
For the reasons above, I am against this proposal.
This thread is dramatically different. Importantly - while the thread you point to changes minipool16 max collateral from 24 ETH to 12 ETH, this thread leaves it entirely unchanged.
@sam, @epineph, @Hodja, @sheen and myself explicitly made cases about not changing the current workings in a thread that suggested changing them. This proposal doesn’t do that and putting a list together from something that different is disingenuous.
There are also several that responded about “not changing” as we move to LEB8s. That is exactly what this proposal is trying to do, as I see it after a fair amount of thought.
@waqwaqattack please read my response to 4 in particular.
That doesn’t seem materially different when 16 ETH minipools will clearly trend to 0 over a long enough period of time. The two ideas are economically similar but differ mostly in terms of mechanics. We should design for the ideal future state rather than cater to a specific 16=>8 upgrade that will glidepath to 0 over time.
In either case, the core objection was around the strategy of tweaking collateral max as I understood it. But obv the people tagged should form their own opinion and vote accordingly.
Disingenuous is a bit harsh. One might argue that taking a materially similar idea to a vote is attempting to re-roll the vote when we have strong signal on community sentiment?
Maybe, although there’s significant overhead with reducing collateral max only to maybe increase it again? Practically speaking this would be closing the door to a future valuable improvement based on current conditions that can just as easily be changed. Why limit future optionality?
Disagree. This proposal would reduce the average RPL collateral backing each unit of rETH. How is that not a broad tokenomics change and risk multiplier? Actually in fairness this might only be true in certain circumstances. Needs more thought / research.
But you’re suggesting we directly amend an existing RPIP with a number that was chosen arbitrarily. What is our process for discovering that right number, since you’re proposing we deviate from the current 150%?
I have no confidence that 20% is the right number and not 13%, 58% or 150%.
Thank you for this. I’m generally in support of this for the reasons you mentioned, specifically that it helps the protocol even if there are some overcollateralized folks like ourselves who benefit less. The following two points:
I believe we need some research on how we use our collateral (particularly whether we have enough liquidity to auction RPL for even small correlated slashings). Otherwise the 20% protocol ETH is a bit of a shot in the dark. Ken’s analysis of LEBs is largely focused on MEV, where the current penalty system is unlikely to touch RPL in 8E LEBs- but maybe RPL should be touched?
I think the RPIP system needs to somehow deal with contentious issues directly. Everyone supports LEBs in some form, so to me a RPIP is actually not necessary for yes/no on that. I think the RPIP should ask the community at least some of the hard questions that don’t have consensus: how low an ETH bond, min collateral/max RPL collateral, rETH commission, etc. And there needs to be a reasonable alternative in the vote, rather than ‘support all these very specific variables or no LEB for you’. Deciding on these questions before the snapshot gives a great deal of power to unelected (and intelligent/selfless/good hearted) activists and takes governance away from the community. This is a narrow line to tread and I wouldn’t relish writing an RPIP , but god bless those who take it upon themselves.
reply to valdorff
@Valdorff yes you are, and I appreciate the rereading of the discussion (I did the same yesterday and it’s a bitch to find common ground in all those voices)! But even if this forum poll wins 70/30 (to make this change i support), to me that is not enough of a consensus to place in an RPIP. The sample size is too low and there is obviously bias on who is voting (ie people who frequent the governance forum or people tagged!). To me that is still a contentious issue, and the community as a whole should have input. My two very poor cents. And my apologies, I’m working on a post for the min collateral side of things
I think I’m attempting #2 right now. The snapshot has to be pretty simple, but discussions before that can hit individual parameters more easily.
I should note this is also just a part of voting - we can’t have every parameter be voted on separately, because they tend to make design sense in groups, not fully independently.
I think tokenomics is one of the most important parameters to a protocol’s market value and getting it wrong can break the whole thing.
I have done zero research but have an opinion, which is the most armchair kind of opinon. But this is the internet so thoughts will be shared.
If something isn’t broken what are we fixing? I’m wondering what protocol risk are we mitigating here by adjusting the maximum collateral rules for the LEB8 (which I expect will form the majority of minipools in the long term future).
Rocketpool growth depends strongly on the growth of the node operator pool, as they provide the service for rETH holders.
Node operators also put up a lot of capital, have the most illiquid position, have the majority of the RPL volatility exposure and thus carry the a lot of the protocol related risks.
I believe the majority of node operators are the yield seeking type that just want to sit on their capital and sell to cover their lifestyle. Kind of like a property investor who rents their house. You are not going to set up a node with transaction fees and illiquidity with protocol risks for measly extra 4-5% p.a. yield in a short term trade.
RPL yield is trending down and may actually fall below ETH yield in future. Say if RPL/ETH doesn’t change, such a node operator will prefer minimal RPL collateralisation, preferring to just stake more ETH.
Therefore I believe the future overcollateralised node operator is one who believes RPL/ETH is uponly and hence will be of the type that believes in and is most interested in the success of the Rocketpool protocol.
In terms of supply and demand I’m thinking most of the RPL demand will come from these types of overcollateralised RPL/ETH-bullish node operators in the future. This demand has to beat the 5% supply inflation. If that happens all node operators win.
Question: Is lowering LEB8 maximum collateral counterproductive by lowering demand from those future NOs? Those NOs who stake most RPL will also be most likely aligned with the success of the protocol therefore, is reducing their voting power in favour of smaller operators favourable?
(Currently voting against, but this is more of an “inertial vote” - i.e. do what we’re already doing until it becomes clear this is very beneficial)
Unfortunately, we are not currently doing anything yet, as LEB-8s don’t exist and have never been subject to any votes (RPIP-8 predates effective governance, so maybe that’s not so wild, but it never went to Final from Draft). You can still go with “inertia” (I did for a while before I took the time to form an opinion), but it’ll be the inertia of a number that was explicitly not a consensus number, which the author of the RPIP (me) put in.
I am not sure I am grasping all the details here, but I tend to be in favor.
One thing I think is important is to not change the conditions mid-game for people at least until withdrawals are activated, hence the reaction against touching the maximum collateral for current 16 eth minipools.
Now, for LEB8s there are no rules set yet. And in that case I believe it is easier to go with less collateral and later if we see it doesn’t work increase the allowed collateral (so no restrictions to people already committed), than to go with the same rules as for 16 eth minipools and then realise it made sense to have lower collateral.
The rules of engagement are that a minipool requires 16 + 1.6 to 24. Converting to LEBs is optional. I think focus for LEBs should very much be on what is “best” and not what is most consistent. I can see why you don’t want to change rules for existing minipools after the fact, so I’m supportive of keeping that part as is.
Putting it in terms of collateral range for the examples given:
16 ETH Minipools: 10-150% (no change)
LEB8s: 10-50%
LEB4s: 10-21.4%
LEB2s: 10-20%
This effectively removes all RPL upside for LEBs, and forces the average rETH collateral backing of RPL down over time. It ties together capital efficiency improvements with reducing collateral, multiplying protocol risk. It’s also more severe in terms of enforcing collateral limits than the 10-75% range in the other thread. I don’t think we can reasonably justify any of this.
In terms of protocol ETH collateral range.
It has not been established that this is the correct metric.
The average has never mattered for rETH, because it can only be used per node. It’s not like if my node with 5% can’t insure something then it takes from your node at 500%. Our answer to the question “how much is safe” is our minimum. The only rETH upside is insofar as we stay safe even given volatility.
What else would collateral be collateralizing if not each unit of protocol ETH (rETH)? Do we disagree about the fundamental role of collateral in the ecosystem?
Yes, I disagree that 24 ETH serves a role as collateral in the current state of things. As I explained in my original post’s full “Rationale” section. I don’t believe the maximum end is at all serving a collateral role (despite word choice) - as I said there, I believe the role of the max is about controlling:
As an aside, I like stake/effective stake better as nomenclature, but that’s not really the point.
If true, wouldn’t that suggest that we should change collateral max for 16 ETH minipools as well? Why shoehorn this into LEBs if the justification applies more broadly? I think the answer here is because it provides an easier upgrade path (16 => 8 without increasing collateral limit) but that seems very temporary and the wrong way to go about designing economic principles.
I still don’t see the desire to limit “privileged speculation” to the degree of capital efficiency we can obtain. To me these are 2 distinct proposals that should be evaluated separately to keep things simple.
, but god bless those who take it upon themselves.