In order to keep the application threads clear of discussions (to make it easier for committee members to read and score them), please use this thread for any and all questions and discussions of round 6 period of grant, bounty, and retrospective award applications.
@peteris could you comment on if you’d be interested in reviving and maintaining your attestation feature? If so, cost?
Assuming both paths above are blocked, then I think this is a good spend - not too expensive, and maintenance is truly minimal cost.
Like the idea.
I’d like to see a commitment to a path, or a clear guideline to if you deliver X, you’ll get $A; if you also deliver Y, you’ll get $B. Right now I read this bounty as just the $500 cuz as a bounty-hunter I don’t have a good path to count on more payment.
I like this. A couple of things that were discussed that I want to note:
- We currently have no payment for writing good RPIPs or vote text, so it’s a little weird that we have “catch an error in it”. That said, the value is very very clear.
- Not making the same genre of mistake is important. This doesn’t capture that. The checklists govalpha worked on try to start to get at this. The awards here are small enough that it seems it would be a significant burden to contributors if we say “not only do you have to catch an error, but you need to make a PR to a checklist and …” – so I think it’s the right call to leave it as presented.
I was actually used as a sounding block for price point on this based on essentially the title and being familiar. I responded “$5k? I haven’t thought much about it”.
If anything, I would lean high. GovAlpha have been awesome. They are the best thing to happen to RP governance in a while imo. Encouraging them is valuable. Encouraging them to use retros over grants/bounties is also nice for us (read: zero risk).
Depending on how urgent this is, it would make sense to fund @ramana’s work. We plan to cover his ideas in our upcoming rework of the explorer/dashboard, which is scheduled for mid/late 2024.
Either way, the grant should not depend on what we will do
As is probably obvious there’s no harm in having multiple redundant views of attestation performance. Diversity ftw.
General feeling is that this is very cool, and I support longer-term funding going to members of the community that have shown they can be trusted to achieve results.
I like the fact that there are no strict requirements, milestones or deadlines. I strongly believe that when motivated and honest individuals are supported unconditionally you get the best and most exciting outputs.
However, my strong recommendation here is that this should be written up in the form of an RPIP rather than a grant. Here’s my reasoning:
- The honorariums are voted by the pDAO. You require pDAO buy-in for this to work, and its worth getting this upfront. In a world where the pDAO disagrees with this concept, the outcome is every honorarium being denied, and wasting everyone’s time and cohesion.
- The pDAO’s attention is limited, its arguably a bad precedent to set that committees can generate regular votes (ie attention work for the pDAO) absent prior agreement of the pDAO through an RPIP. Admittedly a yearly vote isn’t much, but still, it would be a recurring cost that was imposed without input.
- There are lots of process and rule details written up in the grant application that should be recorded in a more durable, binding and accessible fashion (ie, an RPIP.)
- Distributing funding can still be handled by the GMC, but potentially this should be called out as a separate line-item for clarity in pDAO expenses.
- Given these are fairly significant amounts, especially if RPL price appreciates significantly, it makes some sense to give the budgetary power explicitly to the pDAO. If this grant is approved, the power to modify the scope implicitly lies with the GMC, which has the potential to cause drama if adjusted in the future… And if you’re having a vote to avoid drama, it should be defined as an RPIP.
- Defining these sorts of votes is basically what the VOP framework is designed to solve, but that only really works if they are present in RPIP form.
I really want to reiterate that I think this is a cool idea though. I’m happy to help write up or review an RPIP if you do decide to go down that route.
Other random feedback:
- Titles as a concept are cool. We should come up with something more fun than the professor / university theme though, imo. Who wants to be a ‘Full Professor’ when you can be ‘Grand Poobah of all Such and Such.’
It would be very easy to setup a static site similar to RPIPs, Bounties, etc to host this. You could apply custom styling easily that way (compared to docs.rocketpool.net) and it would be open-source (as opposed to rocketpool.net.) The downside to this is that it would not be high-traffic unless referenced by the official sites.
I’m happy to:
- Setup a static frontend as described.
- Apply some limited styling if given direction, banner images, etc.
- Review the content.
I can likely provide some input on the content as well, but I would be much more comfortable if someone else did the lions share of the generation work there.
[Placeholder – I think this is still in a draft stage??]
This is paying out about the same as a premium of 0.417%. In chat, I’ve seen folks saying they’re waiting for 1% or similar a number of times. That was with it being instantaneous. With the “wait for a year” and “be locked in for a year” components, I’m guessing this just isn’t enough to get anyone to act. This is my big point, and the rest are much smaller. I simply don’t think this will move the needle.
The rETH value paragraph assumes the DP is non-empty. If it’s empty for 11 of 12 months, it goes from 1 in 9 needed to break even to 12 in 9 needed to break even (aka, guaranteed losing even without any sock puppets; I do think the need to have extant nodes is a strong disincentive vs sock puppets, even if imperfect and not repeatable).
This feels a lot like coinbase earn stuff. I’ve clicked through those and can’t even tell you the names of the things I looked at. This has more since it has an onchain quest, but nonetheless. I would point to the rabbithole quest that saw our number of unique minters spike up, but did absolutely nothing to rETH supply (see “rETH Supply vs Time” and “Unique Minters vs Time” on 2022-08-31 at https://dune.com/drworm/rocketpool).
I also have questions on the lesson like: why does the merge matter anymore, do people need to know validator details to hold an LST (to me one of the benefits is that you get to keep higher level), what is a green payment network, what is a 33% attack (I’m only aware of a self-defeating one that causes temporary delays), etc. The rETH component seems quite small – maybe the 33% bit (clearly aimed at Lido) and our little section at the end (maybe 10-20% of the lesson?).
Anyhow - I see little value for $25k. If we want more bankless audience eyes on rETH, I’d suggest normal ad space over this.
I am extremely dubious of the claims here. In the end, everything has to go through logical checks to determine whether to revert. The idea that a high-level offchain ML thing can better judge what these checks are and execute them in realtime, on any contract, on the EVM, for acceptable gas costs… I don’t buy it at all. I am also guessing this needs admin level privileges, so I might actually end up thinking we’re at more risk after than before. Anyhow. Would need a lot more reason to trust this than my 5 minute browse got me before I’d trust billions of dollars of protocol to it.
As I said in their initial forum post, I don’t think they’ve got the appropriate foundation yet. The conclusions they’ve noted as learnings are things that a strong fraction of the discord server has known for months/years.
Their response back to me on the forums shows a belief that we need to incentivize adding more RPL than the minimum. That doesn’t help the protocol grow, it’s pure pumponomics. It looks like the community is prioritizing growth (see Snapshot).
This is secondhand, but a summary of their twitter spaces chat claimed
their recent analysis suggests that nodes require an attractive rpl Apr to maintain the 10% exposure. It does not show that. In the same response linked above, they note that this is for going beyond the 10%.
I can’t recommend this spend.
Cool idea at a high level. I’d like to see this get more discussion and work.
It’s not clear to me if this should be through the GMC, though that’s the closest thing we have right now. It’s not clear to me that we wouldn’t get in trouble via popularity contest. It’s unclear to me that we have many that would really take a sabbatical.
I’m also a bit worried about overhead and competing with ourselves here. Eg, I could make a retro for RPIP work, or I could not do that and try to fund via sabbatical (and apply for a retro if not chosen). If the “CV of unreimbursed activities” is similar to retro applications, what are we earning by doing it this way? A major thing that could be cool here is if was totally open ended – no CV, no nothing – just trust engendered by past contributions. There’s worries about double dipping etc there too ofc.
Part of me goes “let’s do the appreciation titles (via vote) and leave the rest to grants/bounties/retros”.
Here are me thoughts on this slate of applicants.
RocketPerf - I like having more data, but I’m personally not sure how this is unique from rated.network’s break down of performance.
Bankless Academy - I’m a big fan of educational content and this seems great for that, however, the price tag is a bit high for something that’s not a standalone feature. I enjoyed their Optimism quest and minted the pope at the end. I think it was well-made.
SphereX-Protect - not my area but sounds sus.
Nethermind - I like the idea. I like having a more professional analysis for tokenomics. I like the structure it could provide. That said, this is a complicated undertaking and there’s a chance the end product ends up going unused. I support the proposal if it includes the 2 reports - from the proposal it is unclear if the cost is being covered or not.
Rocket Pool Sabbatical - I like the thought but I’m not sure this is an effective approach. Maybe with some fine tuning.
Hildobby: Big yes - amazing content
GovAlpha - their contributions are incredible and we need more of it. I support paying above 5k to encourage more contributions at their quality level. There are very few people in the community who have the time to formalize our wish washy governance structure. As longforwisdom once put it, our bus factor for governance falling apart is one (Val). The more polished we can make the system, the better.
Ethereumstakingguide - I am a big fan of their spreadsheet and have used it / referred it to others. Further, i support the increase in payment. I believe, though, that the project would benefit greatly by moving to a dedicated site instead of a google doc.
Drworm dune - Hell yes. This dune board has been critical to me on twitter and for many who keep track of rETH. When we were in the process of getting a Chainlink oracle, the rETH/ETH volumes on DEXs chart on the dune board was crucial for our success.
Val dune - another slam dunk IMO. The peg and APR tracker on this dune board is the best place to track performance right now for rETH. When the queue was killing us, this board helped us visualize the damage.
RPIP Review - fantastic idea, honestly the rewards should be bumped up a bit given it’s almost like a bug bounty.
Kill the Premium - seems liable to farming and increasing minipool churn. Cool idea but needs workshop.
Scorecard - strong support. The lido scorecard is wonderful and I hope to see the same kind of transparency on our part. This would be great for marketing and is generally a good thing to do.
RocketPerf would be much more informative than what is currently available on rated.network as far as I understand: you would be able to see your performance down to the granularity of each individual attestation duty (plus the other validator duties).
These would of course be aggregated over the desired time period for visualisation, but the user could drill down to arbitrarily fine detail.
I hope to present this high resolution data in a way that satisfies the most exacting node operators (see exactly what the beacon chain thinks of me) while not becoming overwhelming. I believe this can be done with clever use of user-controllable computed aggregate views and good defaults.
This is significantly a cost-down of a previous proposal. Where the previous proposal suggested creating a new iteration of an AI monitoring solution AND making it easily accessible for RP NOs, this one focuses on making an existing AI monitoring solution accessible for RP NOs. I think the much lower sticker price does make this significantly more interesting from my PoV.
That said, I continue to have serious questions about (a) how common these issues are and (b) about what steps can be taken to mitigate issues if seen.
For (a): my understanding from the previous discussion thread is that there have been 2 anomalies (that may or may not have been attacks) observed across 2 user-months of testing.
For (b): if the only realistic solution to such an attack is to use something like vouch to have a separate listener and submitter nodes (as discussed here), then would we be better off funding work to make that setup easier rather than work that can only tell us we’re in trouble, but not get us out of it? Are there other effective solutions @Davidlerma? I’d like to understand what an average RP NO could realistically do if they did detect an issue?
SphereX-Protect is not about high level ML that can better judge what checks should every transaction go through. It is actually about reducing the security risk by hardening the protocol to eliminate potential attack vectors, and minimize the system’s attack surface.
Given a history of transactions and user interactions for a given protocol, SphereX-Protect removes the ability to send abnormal transactions, thus preventing hackers from exploiting bugs which almost always represent abnormal edge case and unpredicted behavior, never-seen-before.
It does not necessarily require admin level privileges, and can be customized to comply with any governance structure, such as DAO. In any case, SphereX has no control over your protocol.
Anyways, we propose to start with a proof of concept (see the project’s milestones), in which we install and deploy our capabilities on either private blockchain or testnet, so you (or other Rocket Pool relevant stakeholders) can experiment and “play” with the system.