Interesting idea! This is essentially the same thing from a protocol perspective as setting a TVL limit for ETH-only pools, except it requires every node to have at least some RPL, which might provide more of a decentralizing effect in total. A much simpler alternative with (I think) the same effects would be just reducing the RPL collateral requirements as the number of validators increases, then capping the number of ETH-only minipools per node. Realistically, since this would be implemented with megapools, I think actual implementation would look like a cap to the amount of ETH staked in a megapool based on the amount of RPL staked.
There is only only one permissionless ETH-only option available currently: StakeWise v3, but this doesnāt give NOs increased rewards over solo staking (just some liquidity via osETH). Even after Lido CSM comes out, they wonāt be allowing it to grow to more than 10% of their TVL based on the same safety concerns we cite in our report. Any protocol that does permissionless, ETH-only bonds with no limit will almost certainly become centralized very quickly.
3 Likes
I feel like the authors and many others have responded (or will respond) to specifics, so Iāll avoid that. I want to communicate my general impression because I spent the time to read it and came away frustrated. Thank you for posting this regardless, I learned stuff.
I feel that the report is an exercise in rationalization, where the conclusion was written before any arguments were considered. The tokenomics changes destroy Constellation as it was envisioned. Per the report Constellation in its current form represents two years of investment from NodeSet, some amount of which becomes useless if this is adopted. That feels like the reason this report exists. I was hoping for a more objective review of the proposals, but that may be a faulty expectation on my part. Thereās value in a motivated argument, but the wording itself and the comments here seem to try to imply the report as objective, and I did not feel this was the case.
I think many of the included arguments are specious and poorly supported. The first citations I checked did not support the sentence it was linked to (RPIP-30 impacts). Justā¦ yeah. I feel like this could have been a third of the length and 3x more useful! Better would be to just outright assert at the start that NodeSet better solves some of the problems that the tokenomics proposals address, and that you would like to ensure your better solution will still be viable when youāre ready to deliver it. Some of the rest of the content reflects more supportable concerns you have around centralization, but its so mixed in with the chaff that its a slog to engage with.
To end on a supportive note, I think:
- Replacing the burn with an opt-in value accruing token should be explored and contrasted with burn. Completely passive investors should not get rent for nothing.
- As others have said, Sub-linear bonding leading to centralization is hard to refute. Iām unsure if this can be addressed adequately.
- Destroying NodeSetās investment in Rocket Pool is dumb, and should be avoided if at all possible. Screwing an entity after two years of investment is a good way to prevent future investment.
10 Likes
This is important point. But the landscape has changed in these past two years and we must account for it. I think we should start implementing the new proposed system anyway, and by the time the first parts of it are ready weāll be able to assess NodeSet impact already. If not yet, we can set parameters of the new system in a way that makes it virtually mimic current system. For example 3% to NOs, 6% to voters, and 5% to buy+burn, maybe turn off some parts. This way NodeSet has time to adapt to new changes and RP gets potential to unleash if necessary.
Regarding that feared CA, I think if we change gradually (and NodeSet is good enough to significantly ramp up RP TVL) itās unlikely that we will get this CA at the beginning. And if we can reach near-maturity, it is the same unlikely to get CA at that point - itās just too late and RP have no capacity for CA due to self-limit.
In theory yes. But in practice, large entities that can become CAs are legal entities and legal entities are transparent due to business demands.