Pre-Proposal: Decentralizing Network Stake via Parameter Revision(s)

Yes, I came in with an attitude that was not suited to this format. I was going meta, and unnecessarily preempting a sentiment I’ve seen expressed by many indexers on this forum. It was not appropriate productive- but you are correct, the language is nowhere to be found in your posting. I was reprimanded, I’ve apologized and I’ve attempted to contribute. Still, the language of the irrational delegator is found within some of the responses here, so the sentiment does exist. I also tend to think your original post makes some implied assumptions about delegator ability to understand things like cut vs real cut percentage that you just cannot possibly have the data to support.

Perhaps I need to become a bit more concise to best contribute to the discussion. But as you point out- I have said an awful lot here, so I would hope you could respond to at least some of that more directly.

I agree. Myself and others have made the argument that placing new caps on stake or delegation capacity would be mostly ineffectual to the stated goal of decentralization, largely unequitable, and quite possibly disruptive. We’ve proposed softer measures that may be more effective at accomplishing the goals of decentralization. I’ve yet to see much meat on the bone of any counter argument.

Bad faith? This is a very charged term that I don’t think should be used casually, especially by a council member. Hasn’t any indexer who’s been running on any stake/delegation ratio since mainnet launch (with the exception of those running negative real % cuts) earned a lot of GRT on pretty minimal overhead? And can’t any indexer choose to unstake at any time, more or less? Can anyone claim to be blindsided by 100k being on the edge of feasibility, in the context of 10b tokens? And exactly what claim to hardship would a min-staked indexer have if they simply walk away in a month or two with significantly more GRT than when they started? Perhaps a revised “Recommended Minimum” in addition to the protocol minimum could be added to any literature for prospective new indexers? But the more you make the argument that min-stake can’t be viable, the more it sounds like any Parameter Revisions to change that would need to be quite extreme.

Once again, this begs the question- are there soft measures or infrastructure initiatives that might prove to more elegantly address the problem of over-centralization, such that it even exists?

Since this was flagged as spam, and I have to resubmit this as an edit, expecting to be flagged as spam again in which case it will be up to the staff whether or not to approve the publication of this comment- let me point out that much of the content of this preposal was already addressed here, two weeks ago, and the consensus was generally that the ideas don’t have legs.

The proposal was written in response to the threat of stake centralization. The problem of stake centralization is that it overly benefits and empowers a few vs. all of us in equal ways. It’s not as simple as responding to that threat by simply suggesting to smaller indexers that they should become delegators instead. If we don’t attempt to create a structure that balances stakes in a decentralized philosophy – and doing so early before the train leaves the station – we are at risk of becoming as centralized as the world has become around us.

Decentralization extends to the thought of true freedom, equality, seizing control back from large organizations and politicians. Millions of people identify with that that ideal, especially next generations. The Graph is looking to lead the way on this path and if we ourselves can’t figure out how to prevent indexer centralization from being allowed to happen, then The Graph may fail as a whole and the rest of the crypto world could turn their back on us in the long run.

However, I also feel there should be no free lunch for indexers. I’m a delegator and involved in telegram groups. The main chat I’m part of has over 1,500 members and we’re a hodler group that has been advertised within The Graph community many times. Yet, our group only includes a total of about 10 (!) indexers.

Moreover, less than a handful of them are actively interacting with us delegators (and they are GREAT btw). I personally am delegating to all of those that are active, helpful and show integrity. I know that other delegators factor in these criteria as well and are not just looking exclusively on APY. However, there are dozens of other small indexers that I neither see in our group nor much active elsewhere, such as discord. I would personally not be too concerned if those indexers that are not active in the community at all don’t make it in the end.

I like to propose an alternative idea to the cap limit. I am not a technical person, so feel free to poke holes into the suggestion if there are feasibility restrictions. It’s a two part suggestion:

1. Solidarity Rewards Share
Large indexers with total stakes exceeding a certain % of the total coins staked on The Graph (say 5%) will have to allocate that same percentage to a solidarity pool vs. splitting it up across indexers and their delegators. Example: Indexer 1 has a 7% share of total coins delegated on The Graph protocol. They earn 100,000 in indexing and query rewards. Indexer 1 thus allocates 7,000 GRT to the solidarity pool and can only allocate 93,000 to themselves and their delegators.

Thoughts:

  • The goal is not to eliminate large indexers, I too believe that we need them. The goal is to find a % threshold that both incentivizes large indexers to join as well as not letting them become too big
  • The bigger an indexer becomes, the larger the Solidarity Rewards Share, the less is being allocated to their delegators, the lower that indexer APY becomes, the higher the incentive for delegators to find a new indexer. It would happen organically without need for additional enforcement mechanisms

2. Snapshot Vote
Since we’ve just all seen our new voting tool, why not put it to good use! Once a month, the money collected in the Solidarity Rewards Share pool gets put up for a vote. Eligible to be voted for are indexers in the bottom tier by total stake delegated (say the smallest 25%). Those indexers then receive that % share that was cast for them in the vote from that month’s Solidarity Rewards Share pool which they then include in their next allocation process.

Thoughts:

  • The idea is not to simply reward the very smallest indexers. The list of eligible indexers should be large enough to actually provide meaningful choices
  • It incentivizes smaller indexers to become more active in the community, especially with other indexers and delegators. The thought is that people tend to vote for those they have built rapport with and who have earned one’s vote via positive engagement.

I am aware the above idea would need some vetting with solid definition on some details around it. Again, since I’m not a technical person, I didn’t want to propose anything too detailed yet since there may be show stoppers here that I might not be aware of. This proposal is trying to bridge
thoughts expressed in this chat thus far. My hope is that it also represents a softer, yet effective, measure for balancing stake distributions on the network that allows large and small indexers to successfully co-exist.

Hi All,

Very good discussion!
We generally agree with the concern of centralisation as well as the operational cots on ethereum mainnet.
Regarding the hard cap, in our view the hard cap for combined own+delegation stake would make sense, which would mean e.g. indexer with 200/300/500M of own stake would not be able to get any delegation.
Regarding the concern of profitability for smaller indexers versus lager and as @indexer_payne mentioned. Allocation towards all the subgraphs with curation is is not necessary the best strategy, but it is seems more appropriate for the bigger indexers, due to the size of their stakes. During the testnet for example, we were able to maximise indexing rewards by allocating to less than 10% of all subgraph with curation.
Regarding the centralisation - query market is what the concern is, not the APY centralisation :slight_smile:. Since the distribution of the query gateways, the biggest centralisation would come from those who has multiple medium indexers distributed globally rather than the single large indexers.
We see concerns now are not necessary what future problems may be. What is definitely worth exploring is an L2 alternatives to lessen the ethereum gas burden, which would be highly appreciated by indexers of any size.

Regards,
grassets-tech team

Mate, I literally just wrote about that case and I said that in a healthy market these inefficiencies will get covered just like Indexers airdropping GRT from their own rewards get covered by delegation quite fast.
Would you please care to read everything before commenting for the sake of participating?
Expand this reply and read:

I would note my previous comment that the cost of allocating among subgraphs seems to affect smaller indexers regardless of whether big indexers also exist or delegators mainly choose big indexers to delegate to. If smaller indexers were able to get a higher portion of indexing rewards and some more delegations, the bar is lifted some, but there’s still going to be a very substantial cost that the smallest indexers will not be able to cover.

In other words, it doesn’t seem like the operational costs of a small indexer has all that much to do with whether the big indexers exist in the network, even if that theoretically makes the issue worse in a sense. An indexer with 100k staked probably will have issues turning a profit with or without framework, figment, and p2p in the system.

I appreciate this. The thawing period has probably been the single biggest concern of delegators (as a whole) so far, and shows also why it can be difficult to discuss some of these proposals (or pre-proposal, in this case) without looping into discussion about other proposals, because they’re all intertwined in the network. If we can implement a solution that allows delegators to move more freely across the network as long as they stay “in protocol,” then we should see active delegators move around between indexers more.

As it is, delegators are by and large stuck with their chosen indexers, and it takes some serious thought (and usually some bad acting by the indexer) for a delegator to choose to undelegate. If I wanted to move the portion of GRT I delegated to p2p elsewhere, I’d be eating a 28 day cooldown, several rounds of ETH gas fees, and a second 0.5% delegation burn. It would take months just to recover the fees and burn and to cover the opportunity cost of the 28-day cooldown, and it’s not really worth it unless I can see a substantial difference between p2p and the new indexer.

I can’t agree with this idea. It doesn’t help this “centralization” concern all that much (it might over a very long period of time, but by and large all it does is spread out the resulting rewards among more indexers) and it takes rewards away from large indexers and their delegators for no real reason other than because they’re large and spreading those rewards to smaller indexers and their delegators just because they’re small.

I already don’t agree with any sort of a cap, but at minimum it seems like it might at least be the most reasonable option assuming it is very thoroughly vetted, the Council believes it will help resolve any issue of centralization, and all potential unintended consequences are considered and resolved before implementation.

(Finally, I don’t think we need to be flagging derkhersh’s posts here. I understand the concern, but it sends the wrong message when an outspoken delegator is being flagged for being outspoken. This isn’t the same as real life where one person can quite literally talk over and drown out everyone else’s voices – forums are quite well suited to allow discussion even if one person posts more than others.)

1 Like

Hi mate, i have read the whole discussion. what exactly you feel i am missing while participating ?

1 Like

And issues attracting delegation that make them profitable. Just because there’ll always be a cut-off point for feasibility, it doesn’t mean we shouldn’t avoid limiting the amount of those it affects, to the degree it makes sense for the health of the network.

And i’m glad you re-iterated that this is just a discussion, or pre-proposal. My ideas in the OP are mine, and while they may be shared by others, they may not necessarily be the best solutions. It really boils down to how much we want to limit the effects of centralization, present or future.

Do we want 5 nodes on the network? Is 50 enough? 500…? Well, until we have L2 solutions and/or a high query volume, that’s possibly unlikely. But personally, i’d like to see us retain at least a 3-figure sum of Active Indexers in the next 12 months, after the effects of subgraphs/overheads start to make an impact.

The goal here isn’t allowing everyone to be feasible no matter their stake or actions, because that’s not possible. It’s simply to steer us in a direction where we retain a healthy number of Indexers, even if that number a little is subjective.

3 Likes

No real disagreements here. I previously asked what hypothetical numbers might look like without the big indexers because I think that matters. Payne suggested, for example, that an indexer needs 100k+1.6m delegation can only allocate to 5 subgraphs, an an indexer with 5m own stake and no delegation can allocate towards 100 subgraphs and turn a very modest profit. But how much does the big indexers actually affect this? If the answer is “a lot,” then this is a good justification for looking for a solution. If it’s “not much at all,” then the issue we’re trying to solve lies elsewhere.

It’s obviously a lot harder to quantify whether delegators will necessarily delegate to smaller indexers if the big indexers are out. Yeah, maybe I’d spread out some of my delegations down the list, but for example I don’t think I would ever delegate to a 100k indexer unless I knew them very well on a personal level. They just don’t have enough skin in the game for the APY to be worthwhile. And assuming other people delegated to p2p for similar reasons I and derkhersh apparently did, I think you’d see a large portion of their delegations spread not to the individual indexers far down the list, but more towards framework, figment, and other big names in the crypto staking space who are also offering competitive APYs.

Yeah, I agree we want to retain a healthy number of indexers. We might disagree on whether a solution is needed, what solution should be implemented if one is needed, and the details of the solution, but I think we’re all here wanting to see a good future for the network.

By the way, I don’t want to go off topic but this is related to our discussion about potential solutions to help alleviate centralization concerns. I was looking into Solana for completely unrelated reasons (because The Graph will be supporting it!), but their network UI is kind of intriguing to me. Check out, for example, the little note under the 7th validator: Dashboard | Solana Beach. Even including something similar on the network UI here might have some positive effects with very minimal cost as it would at least present to delegators the idea that they might consider decentralization as a factor in their choice of indexer.

1 Like

By the way, I don’t want to go off topic but this is related to our discussion about potential solutions to help alleviate centralization concerns. I was looking into Solana for completely unrelated reasons (because The Graph will be supporting it!), but their network UI is kind of intriguing to me. Check out, for example, the little note under the 7th validator: Solana Beach. Even including something similar on the network UI here might have some positive effects with very minimal cost as it would at least present to delegators the idea that they might consider decentralization as a factor in their choice of indexer.

I’m glad you noticed that. We (Staking Facilities) run Solana Beach and try to point some awareness to decentralization in ways like this, wherever we can.

Whether a note like ours is sufficient or impactful in every case, especially the case here, i’m not sure (depends how you feel about the Solana stake distribution).

But there’s only so much that can be done via education i think, because it depends who listens and their receptiveness to the “cause”. :relaxed:

2 Likes

0.5% of 350 mil = 1.75 mil GRT, not 17.5 mil

I was thinking about that too, yes. However, this would also be true for the previously discussed cap limits, so I didn’t introduce anything new in that sense. This is one area where I’m not sure how the community envisioned to control for that risk, maybe someone can chime in on that

Same as above, just even more excessive. Again, maybe someone from the indexer community can provide some thoughts. We already have indexers with multiple accounts today. How would a cap limit (as well as my suggestion) not simply lead to more splits of existing indexer accounts to circumvent the intended control mechanisms?

I’m not sure I would anticipate that, no. Would mega indexers not like the introduction of checks and balances? Yes, likely. Does that necessarily mean they get angry at specific small indexers who receive a little bit more support? I personally don’t see that

There is certainly a lot more to be defined and solved for. For example, the current cost of switching delegation to another indexer would make a successful implementation of this idea very challenging. This would have to be taken into context as well.

I read here that 57% of the delegated weight is already concentrated among the top 3 indexers and the current structure facilitates that trend to continue without any restrictions. To some, this may already feel like a “reality of centralization” not just a “threat of centralization”. I read that it’s game over for any indexer not having staked at least $5M once 100 subgraphs are on-chain. That’s less than 10 indexers currently. That certainly makes me concerned and I’m not even an indexer. I feel something needs to change that enables a reversal of the trend, so we can successfully be a decentralized community across all the participating network groups

1 Like

If you want to ignore what i said and implied about the limits of ‘reception’, then yes.

Is it really necessary to flag/hide Derkhersh’s posts? I understand Derkhersh is accounting for a large part of the posts in this topic, but so far apart from some sour tone I don’t see any bad faith in these posts. They actually attempt to provide constructive feedback and likely represent feelings shared by many within the Delegator class, who as-yet have not represented themselves to a large degree on this forum.

The Graph has many actors within its ecosystem. With different participants performing different actions with different incentives, tensions will naturally develop. The question is how we as a community handle those tensions. I would hope that all voices across the ecosystem are welcomed here as important members of the community. We should encourage Delegators to contribute more than they have, and I worry that actions like flagging/hiding Derkhersh’s posts conflict with that ethos.

2 Likes

Thank you for saying this. If nothing else, if my posts are going to be quoted here I should think the parent should be left so that anyone coming in later can read the full context of my post.

I haven’t flagged myself because it wouldn’t do any good but it makes sense to me after all because this thread is now half his replies full of pushback and sourness. Apart from Payne’s data-driven post and other 1-2, this all has turned into a mess of us vs them, because the discussion got derailed so many times. This probably also scares away other people that would participate and also stall the discussion of this proposal.

I haven’t seen anything more than pure feelings from delegators, because it’s all about money! Complaining that the evil indexer cartel will take away our APY. Not fair, good sir!

I’m sorry but, again, I don’t think it’s fair that delegators get so much public opinion weight vs indexers, while indexers literally do all the work. It took me 1 minute to delegate my stack and I haven’t had any demands since, because I know I’m not in the position to do so. A delegator is a network participant that can make some passive income on whatever holdings he has; if one doesn’t put his money where his mouth is by being an indexer or curator, then his opinion is at the bottom of the food chain.

If any of you participated in other governance discussions, you’ll recognize the moderation of these types of posts for the graph needs a massive step up from where it currently is.

1 Like

I agree there’s only so much education can do, but I also think we should definitely try it first. I know this number is very optimistic, but if even a warning on the network UI could change 10% of delegations, that would be 100 million GRT shifted to smaller indexers at no cost and no changes to the network.

I represent my view as a delegator and I have posted with far more than just “pure feelings” and about APY. I already mentioned above that delegators have often been treated as APY-chasing freeloaders while indexers do the work, and you just proved that misguided view persists. In fact, historically delegators have had near zero public opinion weight. This is quite literally the only time I have felt the need to speak up publicly as a delegator, and also coincidentally one of the first times–if not the first time I have seen any other delegator try to speak up. And now that we’re finally speaking up and trying to engage in a dialogue about a pre-proposal that might affect us, we’re told to shut up because we’re just mere delegators? Hell no. Delegators have collectively put in 3x more GRT in the network than indexers. Without delegators, there’s just the 600m indexer stake out of all the GRT available. Does anyone truly think that would be healthy for the network?

We need to revisit the indexer-delegator-network relationship. Indexers have costs and run the infrastructure, yes. They get 100% of your own rewards for doing that. It’s a pretty substantial reward. Now, delegators add to an indexer’s rewards and to network stability. An indexer who has delegators does not just get 100% of its own rewards, it also gets a cut (typically 10%) of all the rewards the delegator generates by delegating to the indexer. This is a win-win situation for everybody. It’s not delegators freeloading while indexers and the network get nothing out of it. There should be no “food chain” in which anyone is at top or bottom or where any one group should feel as if they have more or less of a say than another on a proposal that affects everyone (other than, of course, recognizing that the health of the network is paramount, and we have a council and people in charge of governance and decision-making whose voices most certainly carry weight :slight_smile: ).

I don’t want to get into a fight about all this because it’s off topic and I respect fattox and cryptovestor as council members who brought forth a pre-proposal on a subject they feel is important (even if I might disagree with some of their views). But I felt I had to respond because I absolutely abhor this type of treatment of delegators, and the statements made were neither true nor helpful.

And by the way, if it seems odd at all, I don’t respond directly to derkhersh because we’ve spoken elsewhere. I don’t agree with literally 100% of his views, but ours are generally aligned so I don’t feel the need to repeat much of what he says either.

Finally, back more towards the topic at hand, we seem to have gained at least one more new “big” indexer since this post. 0x643c5f5223ba68e06766b194b274c4d90dd42dc8.

1 Like

This thread needs a redo once we learn from it and put some governance and rules of engagement in place. Laissez-faire moderation does not work.

I’d urge y’all to hold off for now and go enjoy your weekend.

4 Likes

Hello, Vladimir from p2p.org here.

First of all I want to clarify that expressions like “P2P is not active in the community” are not particularly true. Yes, we are not very well presented on Discord and Forum (I am not proud of that) and don’t read it every day, but it doesn’t mean we don’t care. We have a separate telegram channel with over 1500 members and always help if any questions arise. We had a clear focus on attracting delegators and worked really hard towards that goal.

By being a relatively large indexer (~6,5% from total, without an ability to grow further as we reached the max cap) P2P puts its own skin in the game which is subject to 2,5% slashing for bad service quality. There is no slashing risk for delegators, so they should be confident about the indexer’s honesty and overall staking efficiency.

A lot of indexers are unknown and represent an additional risk of changing fee/reward cuts without announcing it publicly. Small indexers that are active and add value to the ecosystem in most cases receive many delegations being capable of reaching max capacity even without having huge self-bond (Ryabina is a good example).

With a higher delegation ratio, lower inflation and more subgraphs we can expect a higher demand for GRT so it can result in a different token price making claims about indexer profitability at this moment a pure speculation.

Another option to alleviate the opex of indexers is to use some of the upcoming ethereum L2s for staking and managing allocations.

We are here as believers in a fair and open network. Making a cap for the min stake or max delegation amount is not a good way to fight for decentralization - all that it achieves is helping less scrupulous indexers take a bigger share of the network. Participants that don’t mind operating multiple indexers under pseudonyms, circumventing the cap, are favored by this mechanism; honest to the spirit of restrictions are penalized.

Nevertheless, the question of decentralization is a very important topic. Here are a few ways how it is tackled in other networks without enforcing inefficient restrictions.

  • Enlightening of delegators about the topic
  • Including a breakeven line in explorers where stake is concentrated and suggesting to delegate to indexers below, not a stakeweight ordering, trust score, etc.
  • Foundation support for small but reliable indexers (subsidising for bringing value to the ecosystem, delegation policy, etc.)
  • Create a forum thread with indexer intros to make them visible for delegators
  • Non-linear slashing (make slashing risks higher for the largest validators)
5 Likes