This post aims to propose an adaptation in Astar tokenomics: halving collator rewards.
This proposal targets only Astar, Shiden is not concerned.
Actually collators earn 10% of block reward to maintain the network.
A collator can onboard the active set in Astar by bounding 3,200,000 ASTR tokens
By bounding 3.2M ASTR:
dApp Staking APY =~ 11.9%
Collator APY =~ 29.6% (+ dapp staking APY on rewards)
Having the APY rates above, collator reward appears too much attractive compared to dApp staking.
This brings in a lot of unknown collators, probably running on centralized infrastructure, having enough holdings to spin up a collator.
Tokens on collators are not staked and can be unbonded at any time within a 1 hour time frame.
dApp staking misses the amount locked on collators that could be staked to support promising projects
Of course, collators play an important role in the parachain because the performance relies on them. They still need to be incentivized, but we believe that an APY ~3% higher than dApp Staking can still cover the infrastructure cost for a collator and risk of slashing for real supporters of the network.
The current proposal here is to halve the collator reward from 10% to 5% of block reward.
The remaining 5% will will not be issued as reward, lowering the total annual inflation by 5%.
Encourage high stake collators to switch from collating to dApp staking
Reduce centralization in collator set
Decrease inflation by 5% which can be seen as a massive burn (=35M ASTR on 1 year)
Please post any comment here before passing the proposal to a community vote.
All three ‘problems’ mentioned can be tackled in different manner, which doesn’t penalise collators.
Afaik number of collators is limited, and is currently full, so no ‘new, unknown’ collators can actually join the active set.
I don’t really see the existing difference in APY between collators and staking as negative - collators undertake much more work and risk slashing if something goes sideways. Staking is risk free - stake your tokens and collect the rewards.
If inflation is a main concern here - initiating burns of treasury funds can be used, that model has proven results from other chains…
As this may appear from an actual collator point of view, this is not at all made to penalize collators.
This calculation has been made without even taking into account APY on collator reward, at the end of the day the collator rewards compounded into dapp staking are much higher than 30%.
Collators play an important role for the chain which is ensure performance of it but they don’t manage security so they do not need to be incentivized as so.
At current rate, collators earn $3345 worth of tokens per month, this is too attractive and as a result makes the active set crowded because of a too high profit.
The main objective here is reduce centralization of the network, not lowering inflation which is just a bonus effect of the proposal.
The proposal have great ideas. Maybe not all Collators will be happy with it but in my opinion it’s great idea move more ASTR to dApp Hub. Expecially if we mention a longer unbound period to avoid dumps on ASTR token.
I don’t think that if we reduce collator rewards it will reduce centralization. If we compare with Shiden network where at current rate collators earn 10 times less than on Astar, we can see that there active set is also almost full only few slots are free and Shiden very centralized right now. Also from my point of view Astar and Shiden metrics should be the same.
Another thing what I want to mention is that infrastructure costs with time become higher due to blockchain is growing and need to increase storage.
Instead of reducing collator rewards I would suggest to think about some hybrid model when permissionless collators can become a collator only if they have approved onchain identity by the team , but to get this approval collator should meet certain criteria, e.g. has server configuration with recommended metrics and node is running some time. May be increase whitelist collators and decrease permissionless for time being.
My opinion that collators should be with high quality with good infrastructure and this should be incentivized definitely much more than dApp staking where there are no risks, no 24/7 monitoring and slashes. I think current 10% is fair enough for collators job.
In our opinion, such a significant reduction in rewards is not too fair to the collators, because in addition to paying infrastructure costs, they would also have to earn something as a reward for their work.
It would be appropriate to reduce collator rewards by 1.5 times, but not 2. The halving of the rewards can be considered if the price of the token rises in the future.
Lowering the collator reward from 10% to 7,5% of the block reward will also help you to achieve the set goal, perhaps not so noticeable, but still.
You can also implement unbonding as in Kusama 7 days. Thus will not allow collator tokens to be dumped.
In addition, the conditions for the work of the collator in Astar are very strict (which at the same time pleases, as it gives confidence that the current collators are professionals in their field), this means careful and proper monitoring of the operation of the node, timely upgrades, and so on.
Shiden has the same APY, that’s why it also centralized.
But Shiden has another situation: if we halve collator reward, it may not be enough for financing a decent infra, this is why the proposal is not applied to Shiden.
The cost of storage is negligible here, with a pruned db it will increase only a few Gb per year.
Allowing only verified identities would be nothing permissionless, registrar is centralized by Astar team.
Same goes for a network made of whitelisted collators, we want to reduce the power of Astar team in the collator selection process.
In the current proposal the collator still earns 15% raw APY, combined with dapp staking this makes an overall 17% APY.
Considering a full set, collator would earn a minimum of $1673 equivalent (in actual bad market conditions) which leaves $285 left to pay infra cost. We all know most of collators pay average ~$80 for a decent bare metal server for the main collator.
Conditions for collators are strict only for whitelisted collator like you who only have infra cost to support, a minimum monthly reward of $1673 seems more than fair doesn’t it?
If I may add my thoughts on this, I would say that I don’t agree with this proposal as well.
As we may see in other networks, the better solution would definitely be implementing the unbonding period.
A good example would be:
Cosmos projects - 21 days
Polkadot projects - 28 days
Kusama - 7 days.
And Astar can set it’s own unbonding period same as Polkadot.
I am sure this would solve the issue better than cutting the collators rewards in 2, which is way to cruel, giving the fact of the strictness of maintaining the reliability of the nodes.
Just want one more time highlight your provided advantages and respond to each of them:
• Encourage high stake collators to switch from collating to dApp staking
This might work for new users who in decision point what to choose stake with dApp or start collator. As now we have full active set and for existing collators who have infrastructure up and running and rewards higher then dApp staking there is no reason to switch to dApp staking even after reward decrease.
• Reduce centralization in collator set
As I mentioned previously this also really low possibility as we have Shiden as example where rewards from percentage the same as in Astar, but in reality at current rate collators earn 10 times less than on Astar it’s almost in infra cost level if take into account backup server. And despite this active set in Shiden almost full and centralized.
• Decrease inflation by 5% which can be seen as a massive burn (=35M ASTR on 1 year)
This is definitely true, but as one of the comment above , inflation can be adjusted not only through collators who doing work to support network and rewarded only 10% of this inflation. There is 90% inflation from other areas and cut inflation only from collator part and not touching other parts is not fair from my point of view.
I as a collator of course not very happy to see such a proposal), especially now in the current market. But I understand the problems and wish the project good development and expansion. For this reason, I am ready to give up 50% of the rewards.
But can the project also accommodate us collators a little and increase the slashing downtime parameter, to several hours?
Slashing has actually been extended to 2 hours, 3 weeks ago
It would not be good to extend it more because of the flaw of Aura: when a collator is inactive, it harms the network block time.
This imperfection will be addressed in 2023 when we release Astar new consensus but it is going to take a pretty long time to develop. Waiting for it, we need to keep slashing to maintain the best block rate on Astar.