Expanding Kintsugi Collators
8 Comments

Up to this point, Kintsugi collators have been centralized to allow the team to focus on delivering kBTC Vaults. With the introduction of the changes here we're proposing to fully decentralize the system by allowing anyone to become a collator. To enable registration we require two parameters to be set:

  • CandidacyBond is the amount of KINT required to be locked in order to register as a collator.
  • DesiredCandidates is the number of external collators that can register.

More info here: https://docs.interlay.io/#/collator/overview

At this point in time there is no established incentive program for collators, but we suggest to trial the current model so that Kintsugi is more resilient to network outages. Collators do collect transaction fees from the blocks they produce but this will likely not be enough to cover infrastructure costs. We welcome additional suggestions for how to reward collators in the future.

Please use this discussion to signal your support and help decide what values should be set for these parameters.

Reply
Up 2
Share
Comments

It is important to note that the CandidacyBond should not be set too low to avoid potential Denial of Service (DoS) attacks since the default CollatorSelection implementation does not slash misbehavior. If the deposit is cheap enough, malicious actors could repeatedly re-register and not produce blocks within the KickThreshold.

Reply
Up
  • Chains with no delegations tend to have a high self-bond (Shiden) as this is the only barrier-to-entry
  • Tx-fee reward fees leave collators in the red and usually have low self-bonds (Calamari)
    I think a low to medium self-bond would be appropriate, something like $1K to $3K

Due to the relatively low self-bond, you should go for a higher number of desired candidates to ensure enough value is staked. I would aim for 32-64. However, I recommend you increase the count gradually and perhaps manually assign the collator spots. Due to the low candidacy bond there is a danger that multiple slots are taken by a single entity which would be a bad start.

To conclude, at today's market prices (low barrier) and given the lack of collation incentives, I would err to a more censored approach. Looking forward to joining your collator set!

Ioannis | StakeBaby

Reply
Up

Another alternative is to use the illiquid stake-to-vote (vKINT) currency as implemented in this PR: https://github.com/interlay/interbtc/pull/688

Reply
Up

hi,
I think what Ioannis wrote seems quite reasonable. I would add an emphasis on inclusion of whitelisted collator selection from trusted collators that took part in the testnet.

Reply
Up 1

I think Ioannis's analysis is thorough and thoughtful and I support their recommendations.

Reply
Up

@Curu
I like the concept of whitelist, I would extend the range to vault operators older than 3 months (3 as example)

Marvel

Reply
Up

I tend to agree with what Ioannis described. The way I see it is there are basically three ways of rewarding collators sustainably

  • Introduce inflation where part of it is going to the collators (for instance Moonbeam)
  • Tx fees generated by the network (this is currently a problem for the whole Dotsama ecosystem, AFAIK no network has been able to generate enough tx fees to cover collator costs)
  • Every collator can make a proposal to the treasury to cover its costs every month (or it can be a fixed amount for everyone)

I am not sure how inflation rewards would fit into your tokenomics model so let's leave out this one. As was already mentioned, tx number in the network is rather low and it's hard to imagine it will change in the near future (unfortunately I believe this bear market will last for some time).
So this leaves us with the treasury option which is actually the most flexible one as well. It allows you to whitelist candidates and openly discuss their rewards in a form of spending proposals for your treasury to cover at least some of their infrastructure costs. I wonder what do you think about this? @gregdhill

Reply
Up 1

Regarding the size of the bond, I think there are two points of view that are worth mentioning.

  1. Some initial bond is definitely necessary so you prevent adversarial spamming of the network. However, it should be reasonably high not to discourage community collators (especially when rewards are not even enough to cover the infra costs)
  2. At the same time, there were plenty of people running testnet collators for the last year and it would be fair to reward them somehow for doing so. Either with whitelisting (which would allow them to join the set without a bond) or maybe with some one-off reward from the treasury - we can be creative here.

For 1), I think the fair amount is around 1-2k USD.

Reply
Up 1