CIP: Improved Rewards Scheme Parameters

Developers CIPs
21

Yes it should removed. But, it’s not as clear as it looks.

So, to fully answer to this we need to check in which part of the RSS the cost (and margin) have any impact on (the Reward Splitting), if there is any.

Based on the paper:

  • The op reward has 3 terms (simplified to 2):
    • r_operator= cost + (f-cost) * margin + (f-cost)*(1-margin) * pledge/total_stake where:
    • pledge / total_stake, the ratio of the pledge to the pool’s total stake.
    • the first term is the cost, which is a one time fee independent (not really see later) to the total stake (pledge + delegations) of the pool
    • the 2nd term is that the SPO will get that portion (margin) of the rest’s reward based on the margin they’ve set (a.k.a operator’s and/or owner(s)’ reward based on pledge),
    • the 3rd term is rest rewards split between the pledge (as ratio) and members e.g., if pledge is 50% of the total stake (pledge + members) therefore 50% for the SPO and 50% goes to the members (based on their delegation).
  • And the members reward have 1 term:
    • r_members= (f-cost) * (1-margin) * (total_stake - pledge)/total_stake where:
    • (total_stake - pledge)/total_stake is the ratio of the delegators stake to the pool’s total stake
    • It is simply the leftover of the pool’s epoch reward.

Also, keep in mind that the reward splitting has conditions too, meaning members will get nothing and the operator will get all reward when the pool’s epoch reward is less then the cost.

What does it mean at all?
It means that the cost might have impact on bootstrapping new pools, meaning that the delegators are not incentivised to delegate to a new pool if the pool has some small pledge, when the total stake’s reward is less then some value (pool reward should be much more than the cost to compensate the stakeholders lost on their profits). The cost has impact only on new pools and not on established pools who have enough total stake (pledge + delegations) as a fixed 340ADA compared to some 15K rewards on a half-saturade pool (~2%).

But, one question. Will the stakeholders be incentivised to delegate to a pool which has low pledge (less than 1 block per epoch)? Perhaps or perhaps not.

There are a lot of question can arise even with some small changes, so I am not sure how to go further.

So, here we should think of it. How is it relevant to the game? What other things have any impact on this?
It depends on. Just think of a pool which has enough pledge (~500K)

What the graph says, the pools are not incentivised, for example, for having one pool with their all owned stakes as pledge, but splitting their stake to smaller pledges and having multiple pools. The whys are in the graph everybody can find it.

22

The Staking Game is about to reach the Nash equilibirium, when only k number of almost fully saturated pools and some tail pools exist.
It has harsh consequences.

  • Therefore, many-many pools will and should be extinct, sad but expected, similar to natural selection.
    • Ofc, we want our pool(s) to survive others can die.
    • So, we do not want 3.3k pools, when k is 500, but some other nr. like 550 or even 600 etc., but not 3.3k.
  • We do not want distinguish between small and large pledged pools
    • is a small pledged pool for some ideology is worse than some whale with lot of money?
    • One is good for higher profits others are good for charity, ideology etc.
    • But, the system should not give a fuck if there is no enough pledge and not enough delegators, that means the surviving strategy is inefficient for those pools.
  • Sybil attack is not about that many pools arises (most of them will die out eventually), but about that individual pools can gain lot of power despite they do not have enough pledge. Remember delegated stake is the power, ofc with some control.
  • The system SHOULD NOT GIVE A FUCK what individual SPOs want, as it’s only aim is having Nash equilibrium with only k number of pools from individuals, based on their initial investment i.e., pledge.
  • Changing k can have huge impact on the network, imagine set it to 2000 now, the earned rewards could not even enough to cover the costs of the pools as a theoretcial example.
  • We can argue on numbers, a0 example, which is the best? 0.3 or 0.7 even 1.0, what is the impact on it? Should we allow only x amount of pledge? E.g., minimum 1mADA? Why not 3m or 100k? Even, I hardly reach just the survafe of the relevant questions could be asked and related to the game.

So, I am shaking my head that ppl make firm statements and have strong opinions on this very-very complex thing.

2 Likes
23

@_ilap , I’m a fan of the work. Keep it up! Extend the scales beyond 1 pool to multipool operators.

Remember:
The original RSS paper was flawed with analysis done poorly and no stake peer reviewers who didn’t reject it for obvious flaws (no comparison of 5 to 10 equation forms, no MAV?!, no dozens of trials per combination for an average, tiny pool counts, ignoring multipools, no transition from reserve payout to fee based rewards). RSSv1 was always designed to centralize, but it failed and centralized more. RSSv1 was always designed to allow larger operators to have structural anticompetitive advantages over smaller operators. (The yield vs size slope is positive, not negative). It was designed to be self serving, not selfless.

I’m working slowly when I can to do RSSv2. Decentralization and an epic and growing MAV is an unsolved problem in Blockchain, so RSSv2 is necessary. I’m keeping accountable in the Matrix CPS RSSv2 room, however maybe I should start a nice long format thread here and share updates in both spots.

3 Likes