With this soft opt-out feature, several scenarios are possible depending on several variables:
This feature seems critical to maintain a decentralized and healthy validator set in the Cosmos Hub so thanks @jtremback. Without this feature, after several consumer chains are onboarded, many validators would likely not be able to cover these additional costs, especially in the current market conditions
Delegators will be receiving the rewards from consumer chains, but they will not be covering any costs to run the additional nodes. So, delegators would be happy with many consumer chains to receive more rewards, while validators need to consider the balance of costs and revenues
With the soft opt-out feature, it is mentioned that validators opting out from the bottom 5% of voting power would still earn a ‘small share’ of the consumer chains rewards, however the value of this share is critical
If this share is the same as for validators running consumer chain nodes, then likely all validators eligible to soft opt-out would do it. The advantage is that delegators wouldn’t centralize more the stake in the top 95% voting power to maximize rewards from consumer chains
If validators from the bottom 5% voting power decide to opt-out but then they receive significantly less consumer chain rewards than validators in the top 95% voting power, this could likely lead to further stake centralization in the top 95% voting power or forcing the bottom 5% not to opt-out anyway to avoid losing delegators chasing larger consumer chain rewards
These dynamics might be less obvious with the first consumer chain onboarded, but once there are several consumer chains these issues could become critical. In fact these ideas could be used to not only avoid further stake centralization but to improve the decentralization of the Cosmos Hub. Ideas were mentioned before of a variable multiplier of staking rewards, being higher for the smaller validators to encourage more staking with them. This could be implemented with consumer chain rewards, creating a scale of rewards distribution in an inversely proportional way to voting power, so delegators would be encouraged to stake with smaller validators leading to a more balanced and decentralized validator set, what do you think @jtremback@Spaydh?
To clarify: the soft-opt out as currently designed does not affect rewards distribution at all. Whether or not they opt-out, all validators will receive rewards proportional to their stake. The share is only described as “small” because ~75 validators share 5% of the rewards pie.
Modifying the distribution mechanism is a powerful idea and we’ve been discussing mechanisms with Informal and others, but it’s also important to realize it could have strong adverse effects.
For example, weighting revenue towards the bottom of the set could incentivize Sybil attacks (large vals/holders could break into multiple smaller validators to capture multiple spots at the bottom of the list, kicking out honest validators). This might not be too much of a concern if only the consumer chain revenue is distributed with the new method as the majority of the yield will still come from ATOM inflation early on but should still be considered carefully by the Hub Community.
Many thanks for the clarifications. Modifying the distribution mechanism indeed involves a lot of complexities and risks but also if done properly some potential advantages as well. Here are some thoughts:
Firstly, as you mention ‘this might not be too much of a concern if only the consumer chain revenue is distributed with the new method as the majority of the yield will still come from ATOM inflation’ so sybil attacks would not be highly incentivized since the distribution mechanism modification would be only for a consumer chain
Other consumer chains may keep the current distribution mechanism so overall there could be an offsetting effect, and again not incentivizing sybil attacks
The modification of the distribution mechanism of a consumer chain doesn’t have to be a large change, just a small multiplier to encourage decentralization and staking with the smallest validators. Such a small multiplier may have important positive effects for decentralization improvements but not so much to encourage sybil attacks
The current 175 Cosmos Hub active validator set is well established and it is very easy to see and detect a new sybil validator, since their governance participation would be very low, moreover it should be easy to identify whether it is an honest new validator or a sybil from one of the large Cosmos Hub validators, and actions could be taken accordingly by governance
Assuming this distribution mechanism modification is implemented by Neutron, then the smallest validators will attract more delegation, increase in the ranking and as they increase the multiplier reduces, so this would be like a decaying curve of new delegations as they go higher in the ranking with a lower multiplier. Then the new smallest validators will repeat this process, and so on. This could lead to a much more balanced and decentralized validator set. Then, the value proposition of RS wouldn’t be just the security measured by the total value staked in the Cosmos Hub, but also the great decentralization of this stake, making 1/3 or 2/3 attacks much more complex, and hence the Cosmos Hub RS would become much more attractive to many projects
Once again, I’m not against any of these ideas, but they’re unrelated to the Neutron proposal. How the hub handles the fees it receives from ICS is for the hub to decide, not the consumer chains. A separate network upgrade proposals can be made to add the required logic to the Hub.
Hey @Cosmic_Validator , @lexa and I will soon be releasing an essay that explores the design space around this and why it’s important. Just wanted to mention it here because we can have a dedicated space on the forum to pursue this workstream & related conversation in that thread.
Thanks for your thoughtful comments here and will be looking forward to continuing the convo in there as soon as the essay is out.
The cosmos hub needs to adopt as many VMs as possible, Neutron was the first step in the right direction, by adopting more VMs we can only grow cosmos developer community, no other chain has such a diverse ecosystem as cosmos.