Removing dormant/sticky stake won’t result in less assigned blocks. Slot leadership is calculated by active stake.
Dormant/sticky stake is “active” as far as the slot leadership calculation is concerned. If you remove dormant stake from a pool that is actively minting blocks it will affect the number of assigned slots.
Exactly. And therefore more will be available to be paid out to active stakers in the future.
The early larger pools will simply split and encourage all their active stakers to move to the new pool. The “sticky stake” will remain in their old pool. Then they can wind up the fees in the old pool and that “sticky stake” won’t care because it represents Ada held by lost keys, people who have forgotten, or people who just don’t care. Now this early pool operator has an even more significant competitive advantage over other small pools. He can lower the fees for his new pool to lower than his costs because he is profiting more from his old pool due to it’s sticky stake.
This is the kind of business model you are competing with as a small new entrant, and it is unfair. We need to think about how we can level the playing field. Removing the structural benefit of “sticky stake” would help level the playing field a lot. If you are a small pool operator struggling to run a profitable pool then you will have more chance of success if competition is fair.
The docs were not written by the design architects and explanations are simplified. The Ouroboros papers make clear that staking is supposed to be an active role. One key assumption is that stakers will move if a pool starts behaving maliciously. We don’t have slashing in Cardano exactly because of this assumption and the “sticky stake” problem undermines that assumption somewhat.
Well, you should seek to level the playing field. Plus, you have the higher ground because the design architects agree with you.
Yes. Yes. Yes.
Now the newer entrants with smaller pools and no “sticky stake” might get more of a shot.
Then we can agree to disagree 
and I’m sure opinion will be divided on this one. Either way I don’t think this has anything to do with changing k.
How would you define “more passive”? I think you could easily define it as choosing a pool and actively committing your stake key to it. It’s not like people swore an oath to look after their keys, continually monitor their delegation and ensure it upheld their principals and the principals of the network (nor were they required to).
We could argue about intent, what “makes sense”, our interpretation of the docs or any other subjective opinions/topics, but the rules and behaviour are clear and encoded in the protocol that delegates signed up to. Changing this behaviour is almost certainly a governance issue and not going to be altered by changing k or by raising with the PC.
That is correct. This discussion diverted to alternatives to reach the goals that we were thinking could not be reached by raising k.
It will for that affected pool, but not for the total number of blocks for all pools together, because slot assignment is based on your stake relative to the total amount staked. If you get less slots, other pools will get more. I think that was the confusion.
That’s correct, I thought you meant in the same epoch, another confusion… 
Like I said before, that’s a false statement. There’s ni evidence for this. The data provided by @Homer_J shows that after last k raise, stake got more distributed and not all stake stayed with the same SPO as before because of pool splits, only part of it will. A lot of delegators probably just picked a pool because it was ranked high in a wallet and the SPO has no way to encourage them to stay with the pool group because the user doesn’t follow the SPO anywhere…
Thinking about it more, I agree that the docs are not a great reference point due to simplification/subjective interpretation.
Fair enough the intent may have been for delegates to be more active, but that was never enforced at a protocol level and I still think it’s unfair to impose it retrospectively. That being said, maybe it’s worth a CIP to suggest the idea outright and get further comment?
I agree that the dormant stake undermines the “protection” assumed by delegates moving away from bad actors. Obviously, the impact of that depends on how much dead stake there is - I’m again thinking that if the amount of dead stake is that significant then there are bigger problems
I would love a more level playing field! That’s why I was very pro lowering the min pool fee. I just want to do this in a way that doesn’t negatively affect delegates and I feel changing k or periodically kicking out inactive delegates will make a worse experience for minimal (or no) gain. Again, I think the latter suggestion would benefit from a CIP to create further discussion around it.
The suggestion to recognise stake as inactive after a period of time wouldn’t involve kicking anyone out of anything. I think what needs to happen is users simply need to be re-educated about what staking means. Staking is active participation in securing the network, by actively choosing a pool to stake with (partner with), and that you need to be prepared to change your stake should conditions change.
We simply need to re-educate users that their stake won’t be recognised after say 12 months, unless they confirm that their chosen pool partner is still behaving in line with their expectations and values.
Then the protocol could be modified to reflect this timeout on stake recognition.
Automatically not recognising a delegate’s stake if they don’t reaffirm their choice within some period of time is kicking them out of consensus if they don’t perform a manual task. IMO that’s a worse experience for delegates than it is today, particularly for delegates that don’t care about being active in decentralisation and just want the rewards.
You could argue that we don’t want those kinds of users etc, but as I said previously I think adoption and utility are key and I feel the passive rewards are a big selling point for Cardano. Personally I’d rather have more users of a passive nature than potentially dissuading users because they don’t want to have to get more involved with a philosophy they may not care about.
Maybe I’m wrong though. Maybe if this were raised as an independent idea and voted on there would be overwhelming support for it including from end users.
FWIW though I don’t think you can “simply” re-educate users. IOG and the community could try and socialise the intended changes as much as possible, but I still doubt you would reach everyone. Even the ones you did reach I’m sure not all of them would understand the changes or what is being asked of them (more due to a lack of interest on their part than a lack of clarity in the communications).
OK, I see your point. In that case it would be better to just pay everyone a yield of 3% from the rewards pot irrespective of whether they are staked or not.
Sticky stake that is inactive and won’t move no matter what happens is not contributing to the security of Cardano’s consensus mechanism, in fact it is a negative. So yes, there is no point having it recognised as active stake because it actually makes Cardano more fragile.
Of course if we simply paid all wallets (unstaked and staked) 3% yield with active stakers paid nothing extra, then there would be no financial incentive to actively participate in staking as a partner with a pool. Though, there would be still an incentive to stake because those that actually care about security would presumably still choose to stake anyway, so perhaps the protocol might be more secure with only their stake impacting consensus? Similarly, I guess there would be some stake pool operators that would still run pools simply for security reasons, without expecting any pay?
And just like the staking rewards, this will probably go down rather quickly with the reserve being emptied.
It is maybe not that good of a idea to advertise “passive income” as an argument for Cardano, for staking. There already are lots of stake pools who advertise with phantasmal APYs of a year or two ago which gets closer and closer to what one could see as false advertising.
Stake pools could still get rewards for their pledge. Why not?
Perhaps, it’s just the delegation part that was not such a brilliant idea in hindsight?
- Network block latency vs. number of pools (both BPs and/or relays)?
In the numbers we are talking about - marginal effects either way.
The diffusion time is linked to the ‘valency’ (fanout factor) - current suggestions are reasonable for the range of 1,000 to 20,000 block producing and relay nodes in the system
Dear @earncoinpool ,
I am writing to you on behalf of the Parameters Committee to inform you about the status of your recent Parameter Change Proposal (PCP).
As you can read in the last update from the committee, the upcoming PCP-003 has been selected and so other extant PCP applications will be deferred as per the PCP process. This decision does not reflect a lack of merit in your proposal, but is rather the result of current prioritization and our limited capacity to process all the proposals.
The deferral of extant PCPs after a given triage and selection period provides a period for the authors to continue to refine their message, to accentuate the importance of the PCP for the next period, and to assess its ongoing relevance given developments (including the results of active PCPs) in the interim. When the Committee opens the next submission period, you can notify us that you wish to renew the PCP.
We encourage you, if you wish, to use this time to revisit and strengthen your proposal, incorporate any feedback you have received, and account for any relevant developments in the ecosystem. The Parameters Committee values your contribution to the Cardano community.
Should you have any questions or require further clarification, please do not hesitate to reach out to us.
Thank you for your understanding and for your dedication to improving the Cardano ecosystem.
Best regards,
Joaquín López
Secretary of the Parameters Committee
The K parameter is ID’d as the parameter central to decentralization and as such is used to tune decentralization.
Cardano experts are generally understood to believe that the design premise of K is sound and the research that justifies it is well-understood and correct.
Evidence from Cardano Foundation - “The k-parameter plays an important factor in keeping the Cardano blockchain adequately decentralized and encourages ada holders to delegate to a wider number of pools.”
Evidence from IOG - “The system… [through increases in K] incentivizes delegators to move their stake to a pool containing less ada, in order to maintain their yield and thus encourages the decentralization of the network.”
The mainnet poll took place in epochs 412 - 415 starting on May 15, 2023. This poll was conducted 12 months ago, which is more than enough time to respond.
There are no intervening steps in the way:
It was argued that the minPoolCost change decided upon ought to be executed before the PC addressed K. This occurred about 7 months ago now.
Some further justification would be needed for any further delay.
Entering self-governance phase puts added pressure on sufficient decentralization to prevent a dictatorship or an oligarchy from emerging.
SPOs have pursued actionable deliberation about Cardano’s decentralization parameters (primarily K, minPoolCost, a0) for ~3 years now– with inconclusive participation by experts and core developers– and the problem implicated in K hasn’t been agreed on yet or resolved.
If SPOs are to view this kind of conversation as tractable, the latest iteration of these discussions must produce agreement and resolution, or reasonable steps for achieving them.
If K is not the parameter used to fine tune decentralization, then what parameter is?
If K is not central to decentralization, then why does the K parameter exist?
Many reasons have been given to raise K and not to raise K. The parameter committee should layout which arguments that they find to be valid and which are not, as well as why. The committee should also describe the evidence it used to come to their conclusions on the arguments.
The committee should also explain which of these reasons it used to come to its recommendation.
Please refer to the original PCP for Community Support, what is the K-parameter, Why is it important to move K up and Is it safe to move K to 1,000 and why moving in one big move is recommended by IOG
Argument: Doubling K will double the necessary infrastructure costs of the network.
Argument: Increasing K will do more damage to SSPOs near saturation than to MPOs
Counterargument: According to the analysis that the Cardano Foundation referenced here by Community Member https://twitter.com/Ada4goodP in their “The Cardano Foundation’s Response to the Parameter Committee Recommendation in PCP-001”; shows by its own numbers that the oversaturated stake would be about 2 billion ADA staked in MPOs that would need to be redelegated. While only 542 million ADA would be oversaturated in Single Pools, that would need to be redelegated. To summarize, about 2 billion ADA in MPOs will need to be redelegated while about 542 million in Single Pools will need to be redelegated.
Analysis by Homer_J and others show last time K was adjusted, while MPOs spun up more pools their overall stake did decrease. Source: https://forum.cardano.org/t/pcp-k-parameter-earncoinpool/122701/61?u=earncoinpool
Argument: Existing SSPOs near saturation may choose to become MPOs
Counterargument: Pools choosing to become an MPO is a separate problem that should be addressed through the Cardano Problem Statements process. The problem is the enforcement or deterrent for subverting the K parameter. Maybe the Cardano Constitution could address what IOG called a Sybil attack: “…stakeholders creating multiple pools, either openly or covertly (what is known as a Sybil attack)…” - Evidence from IOG
Additional Evidence from IOG: “Be wary of ‘pool splitters’ - Pool operators that run multiple pools with small pledge hurt delegators and smaller operators. They hurt their delegators because they could have provided a higher amount of rewards by concentrating their pledge into a single pool; by not doing that, there are rewards that remain unclaimed. They hurt smaller and new operators, because they are forcing them to remain without delegates and hence making their operation unviable – without delegates a pool may be forced to close. So avoid pool operators that run multiple pools with pledge below saturation level.”
Additional Evidence from IOG: “Another danger is of stakeholders creating many pools, which can also lead to increased leverage – and even a Sybil attack.”
Argument: The smaller SSPOs who are the only SPO group to benefit from the increase are, on average, less likely to be successful long term than the SSPOs currently near saturation.
Counterargument: Have not seen the evidence for this. We have seen small, medium and large pools close. This seem like the problem maybe be the same as referenced above about enforcement/deterrence of subverting K. “…[Pool splitters] hurt smaller and new operators, because they are forcing them to remain without delegates and hence making their operation unviable – without delegates a pool may be forced to close.” - Evidence from IOG
If we are worried about the long term success of small pools, would increasing K and pushing over 2.5 billion ADA to be redelegated help their long term success?
Argument: Disrupting delegation needs a stronger justification
Counterargument: Staking is what secures the Cardano network. Actively staking and re-staking is paramount to that security. Cardano has liquid staking to make the process easy and not disruptive. If this argument is considered valid it’s hard to see how K could ever be increased if redelegating is considered disruptive. - Evidence
Only about 11% of active stake would need to be re-delegated due to being in over saturated pools - (2.5 billion in oversaturation / 22.6 billion actively staked = 11% of active stake.) Sources: Cardano Blockchain Explorer & https://lookerstudio.google.com/u/0/reporting/639b7209-36db-4aea-956e-6d220412ba9c/page/p_gti726jc6c
I would also note that as part of the Cardano Foundation Poll on K and minPoolCost there was a re-delegation phase so re-delegation was not seen as disruptive.
source: https://cardanofoundation.org/blog/entering-voltaire-poll-experiment-live-on-mainnet
Footnote: Thank you to Chris from Pond Pool and Ryan (Cerkoryn) for reviewing this supplement and providing feedback.
A special thank you to Matthew Capps. Matthew was a helpful sounding board in refining the conversation to key points to get this PCP resubmitted.
Thank you!
Thank you this very impressive compilation. As a small pool operator, I need every help to keep going on. The current situation makes it very hard to get delegation or even to keep it. And, IMO I am not alone with that problem. I also don’t fear multi pools because the only real problem IMO are exchange pools. The only solution to get delegation away from exchange pools is IMO education about the advantages of liquid staking. Forcing re-delegation should help understanding the process.
“Publishers note - this article was drafted collaboratively by members of the parameters committee.”
‘K’ Parameter: Update
We are pleased to see the continuing discussion on the ‘k’ parameter, which highlights the complexity of both the premises and the possible effects of changing ‘k’. We’d further like to thank the author of the public ‘PCP_K-Parameter_EarnCoinPool’ submission and everyone who is involved in and driving the ensuing discussion of network health optimization, including SPOs, researchers and others from across Cardano’s ecosystem.
Note that this response is not a regular PCP result, but a comment on this discussion of the ‘k’ parameter to support the community endeavor. In the view of the parameter committee, the supplement (link) to the original K PCP does a good job of representing the relevant data about stake distribution and summarizing the arguments for and against an increase. Although those arguing for a different course of action might frame it differently, the key elements are there. In gathering this history of the conversation, the supplement highlights a few open questions that need to be cleared up.
We believe there are four major points to address in order to bring the discussion to a point of decision:
In a PCP, the parameter committee considers a goal or plan that involves a change in Cardano’s parameters, provides facts about how the network will respond to those changes, and a recommendation about how to pursue that goal or plan, given those facts.
In the case of the present PCP, the goal appears to be the greater decentralisation of Cardano reward and stake distribution, and the plan to increase k to 1000 assumes that it is accurate to think of ‘k’ as a parameter that “fine tunes decentralization.” This is shown in part by the PCP’s extensive citations of material from IOG and Cardano Foundation explaining k’s original design rationale as a lever to increase decentralization during the maturation of network stake distribution.
However, as shown below, ‘k’ diverged significantly in practice from what the design rationale of ‘k’ predicted, due to other network and real world factors, and this divergence accentuates the tradeoffs associated with an increase in ‘k’. Both of these discrepancies with the PCP’s starting assumptions are important to be aware of for the ecosystem to decide what plan it wishes to follow to optimize stake and reward distribution.
For these reasons, the best way for the parameter committee to serve the PCP’s goal is to correct or recalibrate the premise about what can be expected of ‘k’’ so that the ecosystem can assess its options with accurate information, and pursue the goal of optimizing reward and stake distribution.
To preface this discussion we summarize the three major options we see for the ecosystem to pursue:
In their supplement to the main PCP submission, the author asks:
a) “If K is not the parameter used to fine tune decentralization, then what parameter is?”
b) “If K is not central to decentralization, then why does the K parameter exist?”
To a) – While ‘k’ has undoubtedly contributed to Cardano’s decentralization, it is not the sole factor. Apart from a0 and minpoolcost, other non-parameter features such as redelegation and liquid staking also play a role. A serious attempt to fine tune decentralization is likely to imply a larger project which looks at multiple parameters in addition to potential reformations to the reward sharing scheme.
To b) – ‘k’ is specifically designed to impact the target number of stake pools by creating disincentives for a pool to accumulate stake above some threshold. The Ouroboros incentives paper used a game-theoretical analysis which assumes that the number of stake pools would converge to ‘k’ over time. In practice, this is not the case, as the projected game theory does not fully reflect the complex real-world factors that have developed in the years since Cardano’s implementation. In particular, the original paper didn’t consider the impact and effects of: custodied staking services (exchanges), multi-pools, pool branding, loyalty, delegation stickiness, or the effects of non-financial incentives: ethical delegation (e.g. UNHCR) personal delegation (a friend, or particular operator).
If each stake pool was operated as an independent entity, and convergence happened as predicted by the model, then ‘k’ would be a direct measure of decentralisation, which can be defined in this report as the quantity of the distribution of control.
Metrics such as Edinburgh Decentralisation Index (EDI) clearly show that while ‘k’ is, as predicted, an indicator for the level of decentralisation, there is not a direct correspondence between ‘k’ and the actual level of decentralisation. In fact, Cardano now has many more than the target number of pools even accounting for multipool operators. This is around 1000 block producers per epoch, which is in-line with the Binomial distribution expectation of the target value of 500, and around 3000 total registered pools.
Looking at the number of registered stake pools, it appears that it is asymptotal around 3000, and more importantly, independent and not correlated to market conditions, price of Ada, or the amount of Ada in the reserves. Due to this, we can consider the total number of registered pools to be a proxy (with some unknown factor) for the number of entities in the Cardano ecosystem interested to run a stake pool. It would thus be fair to assume that the number of entities running a stake pool will not be altered significantly when changing ‘k’, and mainly the number of pools and balance of stake to some extent, would be affected.
Given the overall disconnect between k and stake distribution, due to real-world circumstances not accounted for in the original research, we can expect a number of limitations on the effectiveness of increasing ‘k’ to alter real decentralization.
An increase in ‘k’ would affect the cost of operating the network, which can be thought of in terms of financial or energy costs. While doubling ‘k’ would be unlikely to lead to a proportional doubling of cost, many pools are likely to be created and nodes spun up, following multi-pool and exchange pool splits.
The energy impact can be seen quantitatively in terms of carbon emissions through research by CCRI (CCRI Indices) and the report on the impact of Cardano in particular (https://carbon-ratings.com/dl/network-assessment-cardano-2024). As a result of this increase in global energy consumption, the total cost of block validation will increase and profitability of Cardano block production as a market will fall.
To put the applied methodology of CCRI into perspective, where the energy demand is related to the actual cost of consensus, the number of SPOs producing blocks per epoch is 1000. Those SPOs are estimated to each run 2 relays on average, as per the CCRI report, amounting to 3000 nodes actively participating in consensus per epoch, with many more (those SPOs not assigned a slot in an epoch) needing to keep up. There are two relevant corollaries:
In effect, increasing ‘k’ entails a tradeoff in the efficiency of the network for improvements in stake distribution. The significance of these tradeoffs in light of the potential benefits and alternative courses of action remains for the community to decide.
The apparent goal of the PCP and ensuing discussion is the improvement of Cardano reward and stake distribution. Rewards are not distributed evenly, especially for block producers with small numbers of blocks produced per epoch.
The rewards sharing scheme (RSS) is a complex system with many potential design optimizations that may be more effective than increasing ‘k’ by itself. Addressing the range of design factors affecting the target mechanism of rewards distribution first would arguably be many-fold more productive than manipulating ‘k’ in isolation.
It is not within the scope of the Parameter Committee to propose design improvements to the RSS, but we believe this is a fruitful line of approach for optimizing network health in ways that those both for and against an increase in ‘k’ can agree on. Some areas raised in the discussion so far that could be pursued by the community include:
Using a ‘minVarFee’ as suggested above by community members. Although this is probably not a large development effort, it would require a hard fork.
To amortise rewards below some threshold of expected production (e.g. 10 blocks). This would also need a hard fork.
Distribute rewards in a less linear fashion, e.g. so that smaller pools received relatively more reward. This would also require a hard fork, and would need some research investigation, since the original game theory would be affected to some extent.
Include metrics such as reputation when determining block production/rewards. This is proposed in Minotaur, but is still only a research proposal.
Proxy stake - A major issue is how to address SPOs who are delegating “proxy” stake (i.e. stake that they do not own, but which they control). SPOs typically refer to these as “exchanges”, but there are other actors in the ecosystem. Proxy stake will not flow down to other SPOs as a result of increasing ‘k’, but will just be allocated to new pools that are owned by the same operators. Increasing ‘a0’ (as some suggest) will actually allow operators with a large proxy stake to gain a greater share of the rewards if they choose to pledge rather than delegate some or all of their “borrowed” stake. Again, reviewing this is important.
Sticky stake - Another issue, also identified by the community, is that “sticky stake” and current re-delegation mechanisms in wallets do mean that there is less frequent re-delegation than the game theory would have assumed. Removing “active” stake is a possibility and a small technical change to the rewards scheme. It has some negatives, also noted by the community, mostly around reducing the amount of stake that participates in block production (a major metric of actual decentralisation - passive participation is still participation in the system).
While this doesn’t comprise a direct response to the original proposition of the K-1000 PCP, the committee does have a recommendation– we believe Cardano will be best served by community members organizing to optimize the rewards sharing scheme via the CIP process and Intersect. You can reach out to the Membership and Community Committee to support you in establishing a presence and momentum for the initiative.
We hope that this reply reaffirms the overall goal of optimizing network health and clarifies the range of options for pursuing it, so that the Cardano community can take action. We’d like to conclude by reiterating that the Committee is very pleased to see this discussion proceed; special thanks to the author of the PCP proposal.
A more radical approach: Many problems in the stake distribution can be attributed to human behavior. Thus one approach would be a programmatic/automatic stake delegation to all available pools in the system. Every x epochs there is a stake redistribution depending on the performance of a stake pool derived from the actual blocks produced. With this system in place all the available ADA would always be assigned to the set of the best performing pools in the system and fully saturate the avalable pools. In this scenario a k parameter would make more sense to calibrate the overall performance of the network.
Heads up - I wanted to let everyone know that the article provided by the Parameter Committee was revised this morning after community input in the past few days about the clarity of its message. You can see the changes made by navigating to the top right of the post and clicking on the pencil icon there.
I hope the function and message of the reply is clearer now, and like the article says, I think now’s the time to take action. Looking forward to seeing the initiative move forward, and hope we can take a comprehensive, strategic approach, including the use of K if needed. It’s a great endeavor for on chain governance to cut its teeth on.
