Ways to maintain a diverse active collator set

Hi everyone,

The Moonbeam active collator set for all three networks (Moonbeam, Moonriver, Moonbase Alpha) is special because it includes individuals and organizations from various backgrounds and of varying sizes. Collators often work together in different ways, with the idea that collaboration in a competitive environment can help increase the stability of the chain, increase the visibility of the ecosystem, and that collators can play a part in creating a welcoming community.

Collators are rewarded for providing their services with GLMR as part of the chain tokenomics, and there is a simple formula running more collators = more GLMR rewards. However, because the active set is fixed, there is a risk that a single entity or backer could increase the number of collators in the set to a point where the optimal functioning of the chain could be threatened.

For instance, consider a future scenario where one organization controls 20% of the collators in the active set. That organization pushes an update to their collators via automation which breaks them. At that point there is the potential for a 1/5th increase in the number of missed blocks, causing the transaction queue to get backed up. This could also cause rounds to take longer resulting in an increased time for staking rewards to be dispersed. This organization could also start prioritizing transactions and blacklisting addresses that didn’t pay a fee.

There are many other reasons that a diverse set with limited nodes per organization / backing entity benefits the chain and the community but considering that Moonbeam is a permissionless network which includes open access to the collating function, maintaining that diversity lies with the community. This primarily comes in the form of staking – a well-informed community that agrees with maintaining diversity would use staking to help achieve this, spreading delegations over many entities with less nodes in the set.

However, even with informed staking by the community, it still leaves open the possibility that a highly financed organization could use private funds to push multiple nodes in the set. In addition, with the inevitable liquid staking options coming to Moonbeam, it’s possible that the issue could become more prevalent. These scenarios, and what the community could do to help prevent them has been a frequent topic on the collators channel of the Moonbeam Discord, and here is a summary of some of the suggestions that could help. Also included are suggestions from discussions with community collators:

  1. Raising the minimum bond will act as a deterrent to running multiple nodes and is a desirable action. There is acknowledgement that this may not stop very large whales as anonymity will allow any org to bypass community norms. Also, what would be an acceptable number of nodes by any one entity or organization?

  2. Use on chain registrars to verify identity and persuade the Foundation to create the ability to filter out nodes with NO identity from the collator list at apps.moonbeam.network (or put them in a separate list).

  3. See if the foundation could default-filter out all nodes from the staking webapp when any registered identity is discovered to be operating more than the community agreed maximum number of nodes.
    3a) This could be an automatic function and require no manual intervention by the Foundation.
    3b) It would bypass any on-chain censoring, there would be only censoring on the staking webapp frontend.
    3c) Precedent exists for this approach – Polygon’s frontend filters out all nodes with high backing to promote decentralization.
    3d) It does not solve the super-whale problem that can self-fund (e.g) 6 of their nodes, but it does increase their financing cost to the point it might not be worth it on a risk basis.

  4. A watchful community could also assist in identifying suspicious operator via a new communications channel with the Foundation or community channel.

  5. Other ‘nudging’ techniques should be considered to keep delegators informed and guided to a diversified selection of collators.

These are only a few of the suggestions that I’ve ‘collated’ from some of the community collators and wider community, and I hope it starts a good discussion on this topic including things that we can take action on.

15 Likes

Thank you @Jim_CertHum for raising this very important topic. I would hope there is broad support across the community for a diverse set of collators within the active set. You have raised some good examples of what could happen if this is not the case and I’m sure the list is much, much longer.

There are some very interesting ideas here on ways to mitigate this issue. We at the Foundation are very interested in hearing the opinions of various community members on this topic which will hopefully lead to one or more actionable paths which can be put to governance for the community to vote on and will be watching this thread closely!

4 Likes

Hi @Jim_CertHum

Great points and proposal. Regarding #1 and raising the minimum bond. Depending on the new proposed bond amount this may block community led validators from joining the active set vs capitals/whale backed validators. Do you see this as an issue? What options are there for community validators to have an opportunity to join?

Thanks for the response. To give CertHum’s opinion on your questions:

“Depending on the new proposed bond amount this may block community led validators from joining the active set vs capitals/whale backed validators. Do you see this as an issue?”

I see it as an issue for community led collators to join the waiting list, but not an issue to join the active set. The minimum stake needed to join the active set is currently over 2M GLMR. Changing the bond does not change the backing needed to be an active collator and we’ve yet to see a community led collator buy their way into the set without significant 3rd party backing.

“What options are there for community validators to have an opportunity to join?”

The fact is that the active set only allows for a fixed number of collators, with infrequent increases to that number, and therefore changes to the set don’t often occur.

Of course, there is the Orbiters program which offers an opportunity for community members to participate in collating, but this may not be an option for everyone.

Perhaps if the community decided any one entity will only be permitted to run a maximum of 2 collators it would open up other slots in the active set that could be filled by community members. An increase in the total number of collators in active set would also open more slots.

However, the past shows that any new open slots, if taken by community collators, would eventually be overtaken by organizations with greater financial resources – unless there was a disincentive to them doing that.

3 Likes

It certainly seems unfair that little by little the nodes of the community are moving through the same big entities as always

I have some doubts, at least seeking to deposit in several nodes, normally it will be avoided, if the user seeks to maximize their rewards (higher share of the reward pool, of that collator)

Without trying to be authoritative, how can we confirm that a company does not run more than 2 or x, number of nodes, can an adress simply register like other person?

but I agree that can only be in the active set, at least the collators that have at least one available and verifiable contact, in case of errors or the user has doubts. ( it happened before)

So What options is the foundation investigating?

If I remember correctly, in the early months of 2022 the price of GLMR was over $8 and the min bond was 100,000 GLMR. To achieve an equivalent bond, with the current prices, we would expect to adjust that min bond to more than 2m GLMR. If we expect prices to fall further, we might want to set it closer to 3m for now. The greatest thing about community governance is that we can always change our minds, right? If we find a need in 6 months to adjust the bond up or down, we can simply do that as needed.

4 Likes

I agree with everything, only it is unlikely that it will be normal if people need to go through identification in order to leave their tokens staking for support on the network, this can alienate users.

I believe the proposal is only for the collator to go through identification. Delegators may continue to be anonymous. Does that help or did I miss your point?

Hey Jim, thanks for bringing it up. the option of increasing the self bond might make sense, as it might not make sense for large organizations/vcs to launch multiple nodes at once, and they can probably stop at one/two. but, it might also hurt small but technically strong organizations in launching their first node. yes, of course they have the opportunity to participate in the Orbiters program, but it probably is not suitable for everyone. If we are talking about setting up some kind of filter and so on, I think it can help to some extent, but as we know if we’re talking about large organizations with a lot of resources, it won’t hurt them to fund their nodes to stay on the active network. probably some kind of kyc, on-chain registars might help solve this problem to some extent. the only thing is that this rule is easily circumvented. but in that case the organization would need to change the name and so on.
in some ways, it departs from the idea of being - permissionless and decentralized. but perhaps some regulation and community rulemaking via governance would help make this happen without interference from the Foundation, meaning these changes would not be centralized.

I think part of the problem with selecting collators from the bottom list is that delegators are afraid that such a collator will be knocked out from the active set and delegators will have to wait 2-7 days without rewards. perhaps if delegators had the ability to change collators not on the active set about 2 rounds in advance, it would help delegate funds more boldly to collators from bottom of the list

2 Likes

Good points – I agree that the higher APY for lower staked collators does not seem to work well as a mechanism for redistributing existing stake. It would seem to work well for directing “new” stake to those lower staked collators.

Let’s explore the perception that raising the min bond will negatively impact new collators.

Q. What if an independent operator wants to join the active set with 100k glmr min bond?

If an independent operator wants to join the active set, they must first put up the min bond. This only gains them access to the “waiting list”. Here they will earn nothing on their 100k glmr while they wait for others to “delegate” to them… but since they are not in the active set, there is no incentive for anyone to delegate to them… so, the end result is that the independent operator must recruit other token holders who will be willing to earn nothing while they wait for the total stake to rise above 2.4m glmr. This is a very difficult task and the reason why almost all collators who fall into the waiting set eventually drop out.

Q. What if an independent operator wants to join the active set with 2.7m glmr min bond?

If an independent operator wants to join the active set but doesn’t have the bond amount, then they contract to run the collator on behalf of someone who does. They just need connections to find support for the min bond amount.

Conclusion:
The end result in both scenarios is the same. The aspiring independent operator wishing to join the active set must recruit token holders to support their effort.

It seems to me that the 2nd choice is more likely because at least the aspiring collator can offer the investor a share of the collator returns and thus boost the APY higher than staking alone. This is also a more stable scenario, as the primary investor with a higher APY than average is less likely to pull their funds and drop the collator out of the active set.

5 Likes

"If we are talking about setting up some kind of filter and so on, I think it can help to some extent, but as we know if we’re talking about large organizations with a lot of resources, it won’t hurt them to fund their nodes to stay on the active network. "

Self-staking is a risky and costly business. It’s not a matter of access to funds, but rather a risk-reward ratio question. Financially speaking, having to commit $1M for 1 month (until they get delegations) vs 1 year has a fraction of the risk. So, unless these organizations are already sitting on these token reserves as part of their investment strategy (my case for the “wallet” collator), then, getting and keeping enough tokens to fund a collator may not make sense if they cannot secure delegations.

probably some kind of kyc, on-chain registars might help solve this problem to some extent. the only thing is that this rule is easily circumvented. but in that case the organization would need to change the name and so on. in some ways, it departs from the idea of being - permissionless and decentralized.

On-chain registrars are a decentralized ID solution, so I don’t think there is any departure here. Also, a registrar could be pretty strict in its ID procedure. I know this is not customary, but a registrar could demand entity paperwork and disclosure of relationships with other entities. After all, registrars get paid for their service.

3 Likes

It seems like there is general agreement on increasing the bond, but a poll will hopefully let those who haven’t commented also voice their opinion, so here is one:

  • Should the initial (locked) bond be increased to join the network as a collator?
  • No, I don’t think this will help maintain collator diversity.
  • Yes, I think this will help maintain collator diversity and it should be set to 1M GLMR / 10,000 MOVR.
  • Yes, I think this will help maintain collator diversity and it should be set to 2M GLMR / 20,000 MOVR.
  • Yes, I think this will help maintain collator diversity but I think it should be set to something else and I will add a comment below.

0 voters

we’ve seen it again today where binance brought in their 4th collator on Moonbeam Synclub-2 and pushed out another fellow collator.

So there definitely should be a change to keep a diverse collator set and not being controlled by just a few entities sharing 4 collators each

3 Likes

It is self-explanatory that the minBond needs to be adjusted.
Since the adjustment from 10k GLMR to 100k GLMR the token went from 5+ $ to 0.35$ so the effective minimum bond is now 14x less then it used to be or even more.
That’s why I voted for the 2m option in the poll above.

I’d also love to see filters in the official staking app to make DueDilligence easier for the ecosystem users.

Polkadot/Substrate Portal does the same and even preselects some filters to make life easier.
“One Validator Per Operator” for example is preselected and would be an awesome fit to be implemented in the app along with an “only with identity” filter.

6 Likes

I agree to the proposal to adjust the min bond to 2m GLMR tokens. I think it will help keep the stability of the collator set and the security of the network. Thanks for beginning this important debate.

2 Likes

I am glad to see that the collators have joined the discussion. I’m wondering if the main goal is to have 100 active collators in Moonbeam / Moonriver, and if the min bond is increased to 1/2 M, then with the next expansion of the set, probably only VCs/large organizations will be able to get into the active set, since it will simply become impossible for small organizations. if we think that 1/2M is not such a big amount now, then in the future it may grow significantly.
what do you guys think about this?

1 Like

Thanks a lot Jim for having compiled such a good description of the current situation and the possible ways to improve it.

We have seen how much community collators are involved in giving support to other fellow members of the MOVR/GLMR ecosystem and how well community collators perform in terms of monitoring and backup. Feeling a part of something is what pushes us to do it with dedication and love for detail. This community support, not of collators, but of the whole MOVR/GLMR participants has been one of the key elements of the project’s identity.

That has to be protected in my humble opinion.

The way to do that offers and accepts many options, but the easiest and most logical one is raising the min. bond. Over time, if needed, it can be changed again. A combination of that one with #2 would be my preferred option.

1 Like

Orbiters is still giving an opportunity right bud.

Even now if you notice the last few collators who joined are whales. Increasing the bonding amount means they cant withdraw it unless they leave so would make the whale think twice.

For sure what we’re proposing now can be reversed in the future!
If $GLMR goes back up in price we’d be surely happy to lower the bond again.
This is not a set-in-stone number for eternity.

I raised the exact same concern in Kusama Polkassembly in the discussion about lifting the minimum commission for validators Polkassembly

1 Like

I believe the team should control the entry of collators and not allow any one company to manage multiple collators. Increasing the entry threshold would completely eliminate the entry of small collator companies. Or let the community vote to add new collators.