Cannot be a configuration issue and I run leaderlogs every Epoch showing no slots allocated - 26 blocks minted prior to Epoch 320 when this issue started and 1 slot allocated with block minted at Epoch 351 which provides evidence that pool is operating as expected - if it was a configuration issue I would never have got the block in Epoch 351
Here is a research paper from 2020 that talks about such fees:
I’m not aware of any scientific paper that was published talking negatively about this fee. Can you post a link to which papers you are referring to please.
The term ‘fix fee’ is not used by researchers. This is just the name that coders gave it. Research papers usually refer to these things in proof oriented way such as:
When you hear things such as ‘fixed fee’ or minPoolCost that is usually dev talk.
There are over 100+ research papers available for public viewing here:
Reason for having this fixed was to implement research done trough Ouroboros research papers from 2018 to 2020 (you can find all of them in the link above).
Research pointed out that there must be:
A fee for pool operators to cover the cost
A fraction of reward as profit margin to pool operators
A way to dis-incentivize a type of Sybil attack that creates large amount of empty pools
What developers did is to create a fixed fee which address points 1. and 3. , while they created pool margin fee to address point 2.
There was a talk of making a margin fixed above zero, but that approach seems to go against research.
Further more the research pointed out that if fair reward scheme was implemented this would 100% lead to centralization of the whole system. So arguments to remove ‘Fixed fee’ for basis of fair reward scheme were already disproven.
This was very clear in the research paper I linked above. Here is what research found happens to fair reward schemes over 100 iterations:
Always leads to centralization.
The amount of 340 ADA was chosen in summer of 2020 when ADA was moving between $0.08 and $0.12, so I can definitely see why many people want this drastically reduced (or removed). If we use price of ADA as a guide this fee should be between 65 and 100 ADA.
Yes there can be a vote to change the amount of fixed fee.
There was no community on the beginning. All crypto projects start centralized with their creators and slowly become decentralized as they build out (if decentralization is the goal).
In that paper, the fixed fee is called “declared cost c_i” and it doesn’t really examine what happens if different pools declare different costs, if there has to be a minimum declared cost imposed on pool operators, …
The paper has proven that their has to be a cap, what we now call saturation. It does not really talk about if the declared cost is necessary. That is not the difference between the fair reward sharing scheme and the one with cap and margin.
How? We don’t have governance in place. Parameter updates are signed by IOG, CF, and/or Emurgo. I think, in the last – only half-friendly – discussions about these keys, Charles said he wants to transfer this to this imagined “members/merit-based organisation” (obviously a bit unaware of the problems of libertarians’ favourite child “merit”), but that is still far from democracy and voting.
That being said, I think that IOG not doing anything to protect some profits of high-ranking employees borders on conspiracy myth. Never attribute to malice that which is adequately explained by … laziness.
I was referring to findings in their results which clearly state 3 different fees/rewards pool operators get.
" It makes sense that the reward for the pool leader is different from the reward for pool members to compensate the pool leader for the cost it incurs by contributing to the collaborative project as well as to incentivize them to take the initiative to form a pool. We focus on reward schemes that distribute the pool reward as follows: the pool leader gets an amount to cover its cost of running the project as well as a fraction mj of the remaining amount which we call its (profit) margin. The remaining amount is distributed among the pool members, including the pool leader, proportionally to the stake that they contributed to the pool."
You are missing my point here. I was talking specifically about using an argument of Fair Reward Schemes. Trying to say “x so we can use Fair RSS” goes against research proving Fair RSS will centralize.
Standard track CIP. It literally states :
" 1. A Standards Track CIP describes any change that affects most or all Cardano implementations, such as a change to the network protocol, a change in block or transaction validity rules, or any change or addition that affects the interoperability of applications using Cardano. Standards Track CIPs consist of two parts, a design document and a reference implementation."
As for conspiracy, allegations, blame game, etc… My approach is “show proof or be ignored”. I’m not wasting my time on internet gossip, flame wars or name calling. As you said there is a lot more realistic problems that need to be addressed.
Wouldn’t call it “findings”. They put that in as an assumption – “it makes sense” and “we focus on” – without considering alternatives. But yes, it is clearly not the case that the research was done without fixed fees, albeit obviously without minimal fixed fees.
I think I’m missing that point, because I have seen nobody of the people arguing for no (minimal) fixed fee wanting a fair RSS in the sense of the paper, which is without any saturation.
Maybe, “fair” is used at some places in the discussion, but definitely not in that sense.
CIPs are not voted on. They are merged by CIP editors based on discussions and hopefully consensus. Moreover, if a CIP states that parameters should be changed, but the key holders of IOG, CF, and/or Emurgo do not want it and do not issue the corresponding transaction, nothing happens.
That’s on me. I’ve seen it recently on Twitter in many forms and since Twitter was mentioned in the post I was replying to I anticipated similar arguments and lines of thinking.
True. The closest we have is Path to Active: Showing Acceptance in the Community. I shouldn’t have called it a vote. However, it can still be done. Can’t just give up because you assume that one of the entities will block it. Interested parties can propose CIP to reduce (or increase or whatever) min fee, no one is stopping them.
I don’t believe the case has been made that a separate fixed fee must fully cover number 1. above. I can’t see in the research paper where it makes the case that a separate fixed fee is necessary. Can you point it out?
Also number 3. is dis-incentivised by the 500 Ada deposit and lock-up period. So I don’t agree that the separate fixed fee is required to address number 3. above.
I think, the main problem of these game-theoretical analyses is that people are not purely utilitarian, not in the least.
In those models, a lot of our 3000 SPOs would just see that they can get more if they stop operating and start delegating. And the system would nicely go to exactly k fully saturated pools.
But they want to operate a stake pool. And the minimum fixed fee does make them kind of unattractive for delegators.
It might be that a larger fraction of those shouldn’t really operate a stake pool. That they do not have a clear path to a large enough stake even without a minimum fixed fee. But for the medium-sized of them, it would probably be easing if we can tell them “If you can somehow get 1 million stake, it will probably work out.” instead of “If you can somehow get 10 million stake, it will probably work out. Otherwise, forget it or see it as an expensive hobby.”.
Moreover, for most of those models multi-pool operators look exactly like multiple single-pool operators. Trying to fix that has to be done after the fact with pledge requirements/incentives, social pressure, … And if that works out good enough is anyone’s personal call. (I don’t think it’s as bad as often described.)
As I quoted above:
1.the pool leader gets an amount to cover its cost of running the project as well as
2. a fraction mj of the remaining amount which we call its (profit) margin
3.The remaining amount is distributed among the pool members, including the pool leader,* proportionally to the stake that they contributed to the pool
This clearly identifies 3 separate fees that are paid to pool operators in the research paper (*they refer to them as pool leaders).
Research is usually done in very limited models. These models usually try to answer a few questions about interaction of variables in that model. Controlled variables are mostly fixed not because they are ‘proven’ but because researchers need to make workable assumptions.
So once research is done and is handed over to engineers and coders to translate this research into code, they will follow the model described. Coders aren’t going to sit there deliberating on validity of each variable. They are just going to code variables to mirror as close as possible to the model they were presented. So if research says Sybil resistance against X can be achieved by controlling for variable Y which interacts with system in Y->sometingRewards way, they will try to code for it as close as possible.
They are of course a few different ways they could of coded those variables. They settled on this way and it appears to me that is very close to the model they got from researchers.
Now, is this the only way, best way, most efficient way? No. But it is the only scientifically modeled and tested way so far (for this family of PoS). There will be more research that will improve, prove, disprove and add to the current system but we are not there yet.
We already have proof from ISPOs that 500 ADA pool fee didn’t help with that type of attack at all. Also, it showed that 340 ADA min fee could be bypassed by different incentives. ISPO is a perfect real life example of stuff that research can’t account for and why we now need more research to deal with new problems.
Literally the point of a fixed fee. Part of the reason of having a fixed fee is to make pools with low pledge amounts less attractive since the whole reward incentive is based on pledge and delegation size. Reasoning goes that empty pool attacker can’t offer the same rewards as pools with medium to high pledge, delegators are incentivized to avoid empty pool attacker.
Just to clarify, I’m not defending the fee, I just don’t see any research that show a different variable with same or better qualities.
Yes, this actually has to be true in any model they make, otherwise they risk discriminating and gatekeeping.
I believe in incentivizing and making everything as automated as possible (and as simple as code can get). I think they can achieve the same effect if they removed the fix fee and made minimum margin fee locked for pools below some% of max pledge benefit. Lets say at the moment max pledge level for max benefits is 14 million ADA. You can lock pool margin at minimum 10% for any pool that has a pledge below 0.1% of max pledge level. So pools that have below 14k ADA pledge will be forced to take 10% margin and pools above that have full control of margin fee. When K moves up from 500 to 1000 this automatically drops to 7K and so on. No need to make up numbers like 340 or guess what will cost who or when. Just make it automatic with least amount of moving parts.
I hope after Basho enough of the research team will be free to start making comparative research. Right now they seem to be pushed into discovery type research instead of efficiency research.
In the full paper (https://arxiv.org/ftp/arxiv/papers/1807/1807.11218.pdf) in section 7 where the simulation results are, you can clearly see the only options a player has are:
- start a pool with margin m
- delegate to an existing pool
Players then decide each epoch based on their perceived gain with respect to their own stake and personal running costs for a pool (not a protocol level min fee cost, the real life hardware cost)