Name of Bounty
Kill the Premium
What is the nature of the proposed bounty?
The Rocket Pool triangle is composed of rETH, NOs, and RPL. RPL value is linked to NO demand by the minimum collateral requirements. NO oversupply is linked to rETH by the use of most of the minipool queue to generate rETH yield. Staking ETH oversupply is more tenuously linked to NOs by the ability to arbitrage a premium from rETH secondary markets (RocketArb et al).
This premium arbitrage is very useful when it stimulates new rETH supply- that is, when it provides financial incentives to spin up minipools that are not otherwise viable, such as bond reducing in high gas environments or with long beaconchain queues, or new node operators who need multiple transactions to start validating or may be concerned with purchasing RPL)
However, this premium arbitrage can become maladaptive when adequate financial incentives exist to make new minipools, but node operators wait with the suspicion that higher incentives will exist later because of arbitrage. As node operator bonded ETH requirements go down, these incentives become more maladaptive:
An 8E minipool that expects a premium of 1% could wait 5 months and still turn a marginal profit over spinning up a minipool now; however, that 5 months wait would result in 0.45E capital efficiency loss for the other 24E waiting in the DP.
A NO could wait 8.5 months for a 4E minipool, and the 28 E the the deposit pool would lose 0.9ETH.
A single 16 ETH minipool could wait an astounding 1.85 years to bond reduce in order to still make a profit, effectively losing 2.2E in rETH rewards during that timeframe.
The premium arbitrage also suffers from the additional constraint of becoming inaccessible when the deposit pool becomes overfull after withdrawals; in this case, even a very high premium cannot stimulate NO demand without some NO or NOs forgoing arbitrage to spin up minipools (read: a good samaritan). So even among those waiting for a given premium, it is possible that few will be able to actually benefit from that premium.
Lastly, there are several extremely maladaptive ways to game the system
One would be exiting a large number of minipools to create an extremely overfull DP, and waiting for a premium to build, and then rejoining once other NOs had formed minipools to bring the DP closer to 18k.
An even more insidious way would be to have large NO and rETH positions; exit multiple minipools to drive DP to overfull- wait for a premium - sell your rETH position at a premium - re-enter your minipools and catch whatever arb is remaining on the way down- rebuy rETH and repeat.
There are several plans to increase the financial incentives for creating rETH supply on the horizon (megapools with a bond curve, LEB4s, and RPL staking rework). Hopefully there will be a proposal for universal variable commission. However, the timeframe for these is close to one year- optimistically. A full DP for one year loses about 666 ETH or 1.2 million USD in capital inefficiency, in addition to driving down our rETH APR which impacts induced demand (thus that 666 ETH is likely a significant underestimate)
My thesis is that there are some or many NOs who have the RPL capacity and desire to spin up 8E minipools, but are waiting for a premium; despite the huge damage to rETH value, this is a rational, not a malicious action. Thus, now that gas and queue are low, the arbitrage system has become maladaptive. So how do we change the incentive structure?
We do this by killing the expectation of a substantial future premium by paying NOs to create enough minipools to bring the DP <18k, where it can be effectively insta-arbed to 0% premium. In real terms, we tell NOs “I know you’ve been waiting for for a premium, and we would love you to earn cash, but RP needs you now. It’s not personal, but we will never allow that premium to rise, and we are willing to burn money to ensure that. So you might as well take the gains now, because you are never going to make a killing.” This will act as a bridge until we get more permanent structural changes to the incentive structure.
There will be two prongs of rewards for NOs completing this bounty:
1) New minipools formed:
A. Form a new minipool (maximum of three minipools per node-see risk: sybil thoughts)
B. DP is fuller than 100E below limit (currently this would be 17900)
C. Have at least one validating minipool at the start of the reward period, and have no minipools exited from the node during reward period (october 10/18/2023-10/18/2024).
D. Not be highly suspicious of malicious sybil activity (eg, exiting from one node and forming minipools with another)
2) Bond reduction/stake with credit:
A. Successful bond reduction OR deposit with credit from a 16E minipool present at the start of the reward period.
B. DP is fuller than 100E below limit (currently this would be 17900)
C. Have at least one validating minipool at the start of the reward period, and have no exited minipools on the node during reward period (october 10/18/2023-10/18/2024).
D. Not be highly suspicious of malicious sybil activity (ie, exiting from one node and forming minipools with another)
The amount of RPL rewards will be calculated by 0.1E / [RPL ratio] x [365 - days elapsed from 10/18/2023] / 365 per minipool formed as above. This will be awarded on or just after the reward period ends (10/18/2024) and will be given to all eligible nodes.
The maximum amount of ETH equivalent will be enough to start 1000 minipools- that is, a maximum of 100 ETH; based on the formula, that 100 eth will go toward many more minipools if they are later in the game. Once 100 ETH has been designated, no further minipool creation will be rewarded.
The goal of this bounty to contribute to a positive feedback rally in NO demand- that is, some NOs will form minipools for the bounty, but once it becomes obvious that there won’t be a premium to wait for more NOs will bond reduce for yield, regardless of the bounty. Thus, if the proposal is successful it will self limit the expenses of the pDAO; similarly, if NO demand increases for other reasons (for example, RPL staking proposal) then further expenses will stop.
Thoughts on sybil risk:
Directly rewarding new minipool creation can be fairly easily exploited by sybil actors: Node 1 exits minipool, transfers ETH to node 2, starts new minipool: profit. This is why there is a limit (3) on payments for starting new minipools since we have no great sybil resistance. For small sybil nodes, detection would be nearly impossible and any rewards paid to these actors would essentially be wasted; however, even this loss is not worthless to the protocol in that:
A) starting to develop ways of tracking sybil activity is important, and only n-1 of an exploitive individual’s nodes can participate- so it eliminates a bit of sybil activity
B) a sybil actor may wish to earn the bounty on multiple nodes without exits - that’s just as good from the protocol as different NOs
C) ultimately sybil actors are still part of the pDAO, so a transfer of money from the pDAO to a sybil actor is, strictly speaking, staying within the protocol and is net neutral (excepting the small gas losses)
D) This bounty doesn’t pay anything unless the entire node doesn’t exit minipools between now and October 2024. That friction is a huge benefit, even if the above sybil vectors are used.
Directly rewarding bond reduction is much harder to be exploited by sybil actors. While not impossible, the RPL requirement is more challenging- either node 1 can be entirely exited and node 2 staked more, or minipools can be exited from node 1, converted to RPL and transferred to node 2. Neither of these give as much benefit to the sybil actor, and are unlikely to be much used. And even if this were the case, a minipool exited and converted to RPL can create 5 minipools through bond reduction, or with 8 ETH then 80 rETH can be created.
For this reason there is no limit on bond reductions.
Thoughts on benefit-per-cost:
This is a huge expense (100 ETH ~ 8,000 RPL currently ~ 2.5 months of GMC rewards currently), so the following points should be considered:
- this is spread out over a year, payable at the end of the year, to likely hundreds of individuals
- The rewards are going to validating NOs (ie, from the pDAO to the pDAO)- from one point of view, this is more of a reallocation of funds than an expense.
- This bounty will not pay out unless NOs have no exits until next year- this is a significant friction benefit and makes it harder to game.
- If this is successful, the full 100 ETH won’t be used because the DP will empty as people realize that 0.1E is about as good or better than they are ever able to get from premium arbitrage. When the DP drops below 17900, the 0.1 ETH may not even be worth waiting for.
- If other ways of stimulating NO demand occur before a year, then the cost will also be minimized
- If a NO exits prematurely for personal reasons, both rETH and that NO have still benefited from the interval yield without the GMC paying a thing.
rETH value: 0.1E paid for 1 year of rETH (~0.9 ETH staking gains). To break even, this needs one in nine (1/9) minipools created that wouldn’t have been created without the bounty. This is purely to account for the benefit to rETH holders, and does not account for the benefit to new rETH buyers, the NOs who benefit, additional RPL staked for bond reduction, the mild sybil benefits, and the ‘free’ use of nodes that exit the bounty early.
So to me the major question is, will >90% of the participants be sybil? I don’t see a great way in which they would be.
Changing staking environment:
But what if the cost of staking increases (ie, either the queue lengthens or the gas price rises):
Once the cost is > 0.1E, I don’t expect this bounty to stimulate demand in isolation; however, in this case the premium can still build and from that standpoint the pressure release to increase NO demand (RocketArb) will still be present.
Last important point: this proposal was considered for a few months and publically discussed 10/18/2023 when the premium was essentially zero. It is NOT a reaction to the particular premium today. I just haven’t been able to get my shit together.
Must the results of this project be entirely open source:
Yes, that’s fine. I’ll likely use an excel spreadsheet and I’ll ask someone to help me create an open source bot.
Benefits - enter N/A where appropriate
|Potential rETH holders
||Killing the premium helps new entrants looking to stake ETH by eliminating the one time tax (effectively, a premium of 1% means 4 months of lost rewards). The drag on the APR also causes us to compare poorly to, for example, Lido.
||A full DP is a drag on the APR for all rETH holders, in the realm of 0.96 ETH per year per minipool not created.
||This specifically does not help potential NOs, except to simplify the decision of when it is best to stake a minipool. That time is now.
||More staked minipools means more utilized RPL, which indirectly helps boost RPL price which allows more minipools to be created for NOs. This also provides direct payments to NOs who are growing the protocol by creating more rETH
||Indirectly, more locked RPL stabilizes price for RPL holders, and more minipools increases the value for whale mariage/loans etc
What other non-RPL protocols, DAOs, projects, or individuals, would stand to benefit from the bounty being successfully completed?
Will the results of the completed bounty be open source?
Stop asking me that. Yes.
Work and Verification
What steps would be entailed in completing the bounty? Do successful examples of such work exist elsewhere?
The bounty will be performed by a node operator who fulfills the guidelines above, namely creates a minipool and doesn’t exit any minipools until the end of the period.
I think the best way for this to work is something along the lines of a bot which notifies in a particular discord thread each time a minipool meeting criteria is created; a running tally through the year could then be easily performed by me, which could then be cross checked at random (since everything should be on-chain verifiable) before disbursing funds; funds could be sent en masse with disperse.app or similar.
Have toyed with requiring participating nodes to send a non-zero amount of ETH, and get a POAP- just to increase the chance that NOs are consciously deciding to participate in the incentive program.
How long is the proposed bounty available for? Is it awarded to the first team to successfully claim it, or is it in some way divided among all such successful claims in the proposed availability period?
It is awarded to all NOs who meet the criteria between 10/18/2023 and 10/18/2024, first come/first serve until the requested amount is expended or the period ends. Any excess funds at the end of the year will stay with GMC. The GMC reserves the right to cease funding at any time so that only previously created minipools are eligibile.
Who will test any products submitted for claiming the bounty?
The GMC, or anyone else, can review the list of recipients.
What is the acceptance criteria for awarding of the bounty?
The node operator fulfills the requirements laid out above
How much USD $ is the applicant requesting for successful completion of the bounty?
0.1 ETH (in RPL) per minipool created; maximum of 100 ETH pending the number of participants.
Conflict of Interest
Does the person or persons proposing the bounty have any conflicts of interest to disclose? (Please disclose here if you are a member of the GMC or if any member of the GMC would benefit directly financially from the successful completion of the bounty).
I have one staking node that would fulfill criteria, and it has no 16E minipools. So my maximum reimbursement would be 0.3 ETH. I am running a dark horse campaign for GMC.
Will the applicant, or any protocol or project in which the applicant has a vested interest (other than Rocket Pool), benefit financially if the bounty is successfully completed?
All node operators who fulfill criteria will benefit. Including me.