Removing maximum effective cap on RPL vote power

Governance
1

There is currently a discussion ongoing to clarify minimum effective RPL for voting (whether the pDAO wants NOs with RPL collateral less than 10% borrowed ETH to be able to vote). I wanted to raise the question about whether we should examine the other end of the spectrum (aka, maximum effective RPL at 150% of NO bonded ETH) and consider removing that cap at the same time.

Explanation:
I am proposing allowing NOs to vote/delegate with unlimited RPL; this would divorce RPL reward calculations from effective voting power. While this does increase the relative vote strength of overcollateralized NOs, it does so fairly minimally because vote power is still based on square root of staked RPL (see below). A very conscious decision was made in RPIP-4 to include only NOs and not RPL holders; however, IIRC the major reason for this was because 1) there was no good way to use quadratic voting while preventing massive sybil attacks by RPL dust holders, and 2) folks actually running nodes were felt to have the best protocol alignment. I think this proposal actually enhances both those goals. Additionally, after reading Kane’s notes on proposed on-chain voting and understanding at least 20%, I think this could be a way to simplify the process with fairly minimal downside.

Examples:

  1. Sybil action vs removing cap: A NO has a 16 ETH minipool and 150% RPL (24ETH worth). If they want to double their vote power in the current system, they can create another identical node for 40 ETH equivalent. By removing the maximum effective RPL cap, they could double their vote power by adding 72 ETH equivalent of collateral to their node (ie ~90% more expensive).
  2. Megawhale vs common man/woman: A NO has 10,000,000 RPL (more than half the supply) and decides to stake it on one node with a 16ETH minipool, giving a whopping 843,750% of NO bond. They have an effective vote power of 1581. A minimum collateral 8E minipool (2.4 ETH of RPL) has 6.67 power. So 237 minimum collateral NOs (<10% of our NOs) could outvote that non-existent megawhale. It would be much, much more effective for the megawhale to sybil, which is allowed in current state.

Risks:

  1. Increases the vote power of overcollateralized NOs (counterpoints are these NOs are still highly Rocket Pool aligned; they can get even more vote power currently by sybiling; and the vote power of overcollateralized NOs is already nerfed by square root voting).
  2. NOs could stake large amounts of RPL for specific votes, and then withdraw after the 28 day wait (the snapshot is taken at the start of voting period, so I’m not actually sure if there is a risk here, but i’ll add as possible risk for completeness)
  3. Any changes to something as fundamental as voting carry the risk of creating uncertainty and potentially alienating members, so status quo should be the default. Also, increasing the surface area of discussions could risk delaying other critical decisions. (edited)

Benefits:

  1. If the pDAO decides NOs <10% are indeed eligible for voting, this would take RPL(price) out of the calculations for vote power, and thus seems to significantly simplify the on-chain voting model proposed by Kane.
  2. Simplifies explaining who and how much RPL is allowed to vote.
  3. Provides a non-monetary (aka pure governance) use for staked RPL.
  4. Is a concession to overcollateralized NOs, just as <10% is a concession to undercollateralized NOs.
  5. If tokenomic changes happen (aka valdorff’s proposal or others), the current maximum voting power rules would seem antiquated anyhow since they wouldn’t be linked to rewards or withdrawable amount.
  6. Would moderately discourage sybiling

Anyhow, posting to see if there are any additional risks that I’m missing, and see if there is community appetite for this change.

2

For me, the votes about 10% are an indicator if we want voting to be tied to effective collateral or not.
So, if the outcome is that <10% are not allowed to vote, then the logical conclusion (for me) is, that voting power is tied directly to effective collateral and thus is (currently) at max 150%.

Which means, I’m in favor of your proposal if the other vote passes, otherwise I’m against it.

1 Like
3

A single minipool with a large amount of effectively liquid RPL makes vote power very gameable, for instance borrowing RPL from AAVE just before a snapshot.

Sybil nodes while not ideal do provide a little more benefit for the protocol in additional capacity and more commitment/friction in the required ETH & gas fees.

Aside from that, from a practical stance it is worth being cognisant of how too much further tweaking of voting will delay other potentially important decisions.

1 Like
4

First - just wanna be clear that while this might build off the necessary clarification, this is a fully separate proposal. Importantly, that means discussion here isn’t blocking (other votes) and there’s no time pressure.

If <10% is not accepted, I’m against this proposal.

In other cases, I’m unsure. I agree quadratic limits intense versions of this, but would probably want to do modeling work with more realistic numbers (eg your hypothetical whale and 100 minipools or something). I do also share the gaming worries uisce has. The 28-day cooldown means that very real price risk is being taken on, but unbounded still feels scary (and note that while we don’t know when votes are posted exactly, following the governance channel is probably good to within a day).

In short – I’ll come back and think more if <10% comes out on top.

1 Like
5
Response to Uisce

Thank you for weighing in on this and helping me think through this!

“A single minipool with a large amount of effectively liquid RPL makes vote power very gameable, for instance borrowing RPL from AAVE just before a snapshot.”
This I disagree with, mostly, i think. Borrowing from AAVE for this purpose would be a costly and risky endeavor. From what I can tell, there’s a bit under 300,000 of borrowable RPL on AAVE- Minimum LTV is ~80.5%, so you’d need 5031 ETH (8 million USD). You’d be liquidated if RPL rose ~3.1% during the time you borrowed it (minimum 28 days for staking). In exchange you’d gain sqrt (300,000)/2 = 273 votes, the equivalent of ~41 minimum collateral LEB8s.
You could gain the same vote share by investing 426 ETH into 41 LEB8 sybil nodes currently; >90% less capital required with no liquidation risk, borrow fees, not to mention the capital inefficiencies of those millions of USD on the sidelines. While definitely more complex, with current queue you can create and be out of your sybil nodes before the 28 day wait period is over; and honestly with that extra 4500 ETH, you could swap for rETH and earn enough in your 28 days to pay for 41 proteuses (sorry joe); plus the additional rewards from your 41 nodes and RPL.
And as way of comparison, 273 votes would be about 3% of the last snapshot, less than half of the delegated vote power of any of valdorff, joe, or waq. And theoretically if it was perceived as actually malicious, that RPL is penalizable. And this leaves out the additional RPL voting now from the dozens or hundreds of honest overcollateralized nodes, which will likely out further dilute any benefit. So big risks for smallish reward?

“Sybil nodes while not ideal do provide a little more benefit for the protocol in additional capacity and more commitment/friction in the required ETH & gas fees.”
I don’t think that it is clear that the protocol benefits more from additional sybil nodes than locking up additional RPL. To me it’s kinda a wash, and the governance aspect definitely tilts it towards locking up RPL as more beneficial.

“Aside from that, from a practical stance it is worth being cognisant of how too much further tweaking of voting will delay other potentially important decisions.”
Yes, I definitely left this out of risks- thank you, will add. In decisions as fundamental as voting we should definitely err on the side of the status quo and have a high bar for changes. And we can’t afford to delay a decision on RPIP-4 in rewards to the <10% nodes- that’s required to move forward with anything. However, we are changing RPIP-4 regardless, so while we are focused on it I’d like to make sure there aren’t other tweaks that would benefit the protocol. I’d prefer all changes in proximity than drips through the year, and honestly I think talking about this change now may soften some opposition to giving voting rights to undercollateralized nodes.

response to valdorff

Agreed that that first we need to clarify who votes currently. Then we can talk about who should vote.

But I think that if there were community interest in subsequently allowing overcollateralized nodes proportional voting power, it may make allowing undercollateralized nodes voting rights less contentious (aka to me its a reasonable compromise that somewhat favors undercollateralized but has something for both sides).

The biggest reason I like this change is humanistic, and is the same reason I think NOs <10% should vote:
People at 200% collateral are the same people that they were at 150% collateral, but we value their vote less based solely on RPL price action.

For example, two node operators have different effective RPL amounts- as RPL price goes up 50% they still have the same relative voting power, as it goes down 20% their relative voting power is the same. But as soon as one either drops below 10% borrowed or about 150% bonded ETH, their relative voting weight abruptly goes all to hell. And that discontinuity does not make sense to me, because they are the same people.

2 Likes
6

Agree that after looking at numbers it is less of a risk than I thought, with the caveat that if you are being malicious, borrowed RPL increases your return. That isn’t really an issue with proposals on snapshot but may well be when we move to onchain governance. That being said, you would probably want to sybil for max vote power in that case unless you were actively harming node operators.

7

I’ve always felt that while the base protocol itself should remain permissionless, DAO participation is a privilege and not a right of protocol participants. This precedent is already set by not giving voting rights to rETH holders or unstaked RPL.

Sybil attacks will always be a problem with permissionless governance, so the pDAO should focus on implementing non-doxxing anti-sybil measures such as Gitcoin passport for any kind of DAO participation – especially committee membership but also voting.

With the current permissionless governance, there is no way to prevent a highly capitalized negative actor from vote buying with an RPL-neutral strategy via borrowing. The pDAO needs to focus on the real problem (permissionlessness governance = attacks) rather than tweaking parameters in an attempt to minimize the effects of such attacks.

There are many examples of permissionless governance failing to prevent bad actors from capturing the system: Rook DAO and Nouns DAO are the most recent examples.

4 Likes
8
response to Wander

So I agree with many of your points, but perhaps not the conclusion as i understand it. Preventing attacks with something like Gitcoin passport will ward off the great majority of scammers, but in isolation I worry will increase the damage for attacks that get through by decreasing the number of honest actors; to reduce damage we need much more participation in governance, not less.

For example, I have two gitcoin passports; that was rather trivial. Managing a third would be challenging for me. Getting 10 likely not possible for me. However, for someone who wrote code for a living 10 would be trivial; for someone who owned a company where low-wage workers relied on him/her for their livelihood, getting 500 passports would be trivial and in no way illegal under ‘code is law’. God only knows the arms race when AI progresses.

From that standpoint, i think attack mitigation is equally or more important than avoidance.

Moreover, Gitcoin passport relies on death by a thousand papercuts- no single barrier is sufficient, but the totality makes it very challenging for most scammers to get through. We can do the same thing to mitigate damage by decreasing the benefit or increasing the cost of sybil attacks a little bit here and a little bit there… tweaks… until the totality means something substantial.

So from that standpoint, I think that allowing overcollateralized nodes equivalent vote power makes sense to increase the denominator of honest Rocket Pool aligned actors, and does not increase the risk profile.

Perhaps ironically, I wrote this post with yourself and other similar stakeholders in mind; but I get the sense from your post that you feel effective RPL votes should remain capped at 150% bonded ETH?