Since the launch of Open Staking, there has been a gradual decrease in the number of unique validators elected on the Harmony network. We are concerned about this trend because it runs counter to our goal of creating a decentralized network. This proposal is the first step we can take to gradually counter this issue.
On the one hand, it is still early days in our protocol and some consolidation is to be expected. You don’t become decentralized overnight. On the other hand, this is a critical window for validators to build a track record and attract stake. For these reasons we cannot act too drastically too quickly, but we also must be aware of the time-sensitive nature of this issue.
Thanks to everyone who contributed to the open staking improvement proposal thread. Your ideas and feedback have been important in informing the best ways to solve this problem. So far we’ve already taken some actions to support smaller validators in the following ways:
Undelegation delay has now been changed to 0 epochs instead of 7
Validators are now sorted by uptime on the staking dashboard
On the ‘Next Election’ tab of the analytics page we show all validators instead of only top 320
This is a good start but it is likely not enough. We are considering the following more significant change and would like your feedback. Here’s the proposal:
Increase the number of slots from 320 to 640
Increasing the number of slots will decrease the median stake and allow validators with fewer tokens to get elected
We will perform the increase in two batches of 160 each
This will give new validators time to ramp up and market themselves to attract stake
Increasing in batches enables us to assess any potential negative effects on the network such as latency increase due to larger committee size
This change should roughly cut the median stake in half, allowing more smaller validators to make the cutoff. While this should increase unique validator count, we realize that simply doubling the number of slots will not double the number of validators. We can only know the impact of this change when we’ve made it and had time for it to play out. That said, it should only help smaller validators.
The change of increasing the number of external nodes was already planned, so this only pushes the timeline forward. Eventually all 1000 nodes will be external. Please note that despite this change, Harmony’s internal nodes will retain 68% of the voting power for the safety of the network. We will release voting power for full decentralization once we have implemented last security features such as resharding and our randomness beacon.
Please share your thoughts and feedback on this proposal below. We know this could be a considerable change and we want to have community support before going ahead. We will host a follow up call this week to discuss in more detail. Before we make any changes, we will describe the specifics we have agreed upon in a blog post. We expect we can execute these changes over the following two weeks.
Though I actually agree with the proposal and I do believe it will improve the demographic of validators in the future, there are a few details that should be considered.
As I do not see a proposal to increase the rewards along with this, proposal, this will effectively halve the rewards per BLS key associated with a node correct?
Further to this point, due to the maximums on the quantity of keys per node, this will also effectively double the cost of operating a node? (if a validator is currently able to operate 10 keys per node today, this will require a second node to operate the 20 keys the validator will have to operate).
Lastly, I have yet to analyze the demographic of validators, but based on my quick review of the validators:
Currently the Lowest Voted in BID is 8.3 MM ONE tokens.
Doubling the number of slots will effectively halve the minimum BID to 4.15MM ONE.
Today there are only 2 slots that would be filled by validators with more than 4.15MM ONE and less than 8.3MM One bidding.
In conclusion I expect the demographic will improve the number of validators, but I want to clarify that in today’s open slots list a limited number of new validators will likely enter, the implication of this is that the reward for a single BLS key will halve and may skew the demographic towards those that can afford the lower reward / cost we will experience (which are larger node operators).
To ensure the objectives of this proposal are met, I believe that this should also pay special attention to the smaller new validators we are seeking to attract and should include added benefits to incentivize the reduced rewards these nodes will experience.
It’s a step in the right direction. I’m sure major validators will be hurt about this… but decentralization is key.
Another idea, how about just adding a generic “Stake” button on all wallets that will choose a (well) performing node at random and delegate to them?
Still offer a choice for people to choose… But for people who don’t care and just want to fat finger “stake” without thinking, why not have the code decide based on performance metrics? That should greatly increase decentralization as well.
Makes it easier on the average user to stake, and gives smaller nodes a shot to prove their worth.
Happy to see that Harmony team is trying to reduce decentralization. Appreciate the efforts there and I do think it is an important problem to solve but there isn’t an easy solution.
Few comments/questions:
1 - Assuming rest of the terms/conditions of the consensus protocol remain as-is, I think almost every validator will just double the number of keys. While it will certainly lower the entry point, I dont see how this alone will help bring more than 4-5 validators into the selection. You should also consider the below items.
2 - I understand there are already some considerations to support more validators with delegation from binance. Consideration should be given to re-balance the current distribution as well. Ideally, any such support should be equally divided with chosen validators instead of creating a tiered system where some set of vaidators have dis-proportionate advantage vs the others. This obviously wont fix any decentralization issues but will at least give every qualified validator an equal footing. If anyone is ahead at that time, that will be because of the community support that they have earned.
3 - Dashboard - consider making below changes
randomly sort all eligible validators
new validators should be labeled “new” in up-time and returns
expected returns favor validators from first epoch and the ones that always remain elected. consider simplifying it and just have 3 values for expected returns: above average (good), average, below average (or bad)
4 - In the first proposal on 22nd May, marketing was a proposal. I see some progress being made there. When can we see marketing of smaller validators rolled out?
This post is just reflecting some of the thoughts. No approach is perfect. It is by no means meant to undermine the terrific work that the larger validators have done for the Harmony ecosystem. Many of them have been a long term supporters of the project, have contributed a lot, and the rewards received is justified.
I’m completely in favor of doubling the number of open slots.
At the moment, I am delegating to others because I cannot be a validator. But if I can be a validator at half the current level, I want to be a validator.
For economic rationality, those who don’t make the top 320 at the moment shouldn’t be doing self stake, but if the slots are opened up, there should be a lot of people who will be validators like me.
Larger validators may have more nodes and more costs, but it is the duty of the rich.
The question here is that there is a conflict between the culture of open, collaborative and community driven blockchain on which harmony seems to be identified. The reality is that currently Harmony has far fewer validators than an antagonistic concept such as Ripple which has 140. It seems that the number of current validators needs to be dramatically multiplied.
In my view, if you want to be totally consistent with your message of openness you must be strict in combating the current oligopoly situation in Open Staking.
Limit the number of bls keys per validator to 4. Do not allow more than one server per validator. Do not allow branches of original validators. With these simple measures you mathematically guarantee at least 160 different validators out of 640 slots.
The question here is to what extent you want to be consistent with the image that you project and involve more participants in OS, in this case setting the ground for a healthy competition is your duty.
With respect your proposal, diluting rewards through the increase of slots will be paid by delegators as current validators are going to pass on the cost of the new server by doubling the commission they charge.
Hey Nick, as suggested here : Open Staking: Improvement Proposals increasing only the number of slot will but not help entirely. Reason being is that, lots of validator are scripting their way down in the EPOS ranking slot so that if they are below the 0.85 median stake (or 0.7 in the future) so they have free raw-stake (ie bidding 0.5 would always give you 0.85). What it means as well, being in the low range is that they can take more slot and hence just once again discouraging the average validator.
The proposal I made would incentivized the validator to use as less keys as possible. Also it would also help the validator that can’t afford to use multi-key to earn a bit more than if they were to split and be potentially under the last slot.
We really recommend that is taken care about the number of slots, which will be opened. Opening too many at once could really impact the economics for validators and delegators. It has to remain profitable and validators to earn enough not to increase the fees too much and still be motivated to run the node professionally and on a proper-sized machine.
We personally don’t think is the best solution for it. There were other solutions which we proposed already long time ago but unfortunately didn’t got implemented like: https://github.com/harmony-one/staking-dashboard/issues/284
As for now, we think everything comes around the multi BLS and the incentivization of it. If after a certain number of keys it wouldn’t make sense economically to run even more keys, then that would limit it.
Another possible solution, which would rather be hard limitation, is to cap the max delegations per validator, some sort of validator capacity. For this some math simulations would be needed so that multiple parameters are considered, total number of validators desired on the network, token issuance, existing ONEs on the market currently and in the future. Is important that the limit is not too low so that theoretically all ONE tokens can be delegated but also not too high so that the rule would make sense.
Chainotetech, remember that your vision of a company that wants to maximize profits can easily be replaced by the enthusiastic collaboration of many community members. You are neither strictly necessary nor is your level of customer service indispensable.
if you already read the Juliun’s reply, you already know my thought on this, because I do agree with him 100%
I really don’t understand why keep making big changes to a staking system that is barely 2 weeks old.
This continuous looking for a way to show that “small guys” can really run a validator in a pos system where 60+ % of circulating supply is staked, when in reality they really can’t because it’s too expensive to achieve and most of all to maintain, it’s kinda “deceiving” for them.
3)this point is kinda the same I replied with on the previous proposal’s thread.
People joined, invested, moved funds, created validators based on rules, mechanics and REWARDS known and planned for months.
Having this huge amount of ONEs staked is a good thing for sure, but we already had to face:
A)“binance”
B) the 1.38 B unlocked and staked (very soon)
Now, with this proposal, rewards will be cut in half for every BLS, so even less rewards for everyone ( validator and delegators) if actual validators will not be able to double their slots numbers and/or will need to deploy new servers to retain those slots (-----> fees increase)
how many small delegators will keep their funds staked, when their rewards will become so ordinary or even worse than ordinary, compared other crypto opportunities, in just one month and most of all the FIRST one?
In terms of the economics, I should have explained this in the original post. Doubling the number of slots to 640 means that the rewards per slot will be half of what they are now at 320. Meanwhile, the cost to run a single node will remain the same. That means that net yields (staking rewards minus node operating cost) will go down. However, given that node operating costs are relatively low ($30 per month? on the cloud) I would imagine that the material impact on the bottom line for validators will be small. This is where your guys’ impact would be extremely helpful. What do your profit margins look like, how would this change affect your cost structure and profitability?
In terms of the few outstanding bids above 4.15M ONE currently showing on the protocol, I am not surprised that there are few bids there because there’s no reason for someone to bid that given the current median. There may be people with enough tokens to make 4.15M ONE bid who are currently delegating because they can’t run their own validator. My point is that we won’t know how this will impact on the number of unique validators until we’ve done it. Looking at current bids on the chain is likely not a reliable reference.
@ApertureDataScience glad to have you join the forum! We are pursuing more wallets who can integrate staking functionality. At that point, it would be up to the wallets to decide how to choose the validators. I will say that it’s tricky to trust algorithms for this kind of thing. Could lead to more centralization.
Hey @SmartStake, great points and thanks for your feedback.
I agree. We won’t know how this will change the validators until we’ve done it. Though I do believe that many validators from the beginning have been pushed out due to the minimum bid rising above their stake. I would hope that keeping this minimum bid low would allow those validators to come back and hopefully new validators would have an easier time bootstrapping.
2 & 4. We (the Harmony team) don’t control any delegation but we can certainly help raise awareness for smaller validators. I believe that @Maggie is leading the validator marketing efforts. We will do what we can.
These are good suggestions. I have shared them with Gizem and Sahil.
@Vega I agree with what you said. We need to increase the number of unique validators to align with our ethos of decentralization and openness. We are together on this!
However, limiting the number of bls keys per validator to 4 and not allowing more than 1 server per validators I don’t think would be effective to solve the problem.
Limiting BLS keys is mostly cosmetic. Any large validator would just create multiple accounts and claim the same number of seats. The friction for them would come in the operational overhead of having so many different accounts with which to balance stake and attract delegation rather than just one account. This may be somewhat of a deterrence, but I’m not convinced.
Limiting the number of servers per validator I don’t think is enforceable at the protocol level. Plus many validators may intentionally want to run multiple VPS’s in order to have redundancy.
@rongjian may be able to add more details for these points.
Hey @sophoah I don’t totally follow you. Can you explain what you mean a little more? I would also love to learn more about your proposal. Can you repost it here?
Overall I believe this is a good inital step to lower the barrier of becoming a validator, but I don’t feel like it makes enough of a difference to change the nature of the game.
Personally my holding would have allowed me to secure a validator slot all this time, but I chose to drop out of the validator-game very early on, because I don’t see much benefit over staking at all. Here are my reasons:
Fees for staking are negligible. In fact they are lower than the server-cost and the time cost it takes me to maintain the nodes. Validating myself would only be worth it at much higher ONE-valuations (but I suppose the fees would adjust accordingly).
Running your own validator can be extremely inefficient: If you’re a holder of f.e. 5m ONE and the current BLS bid is 4m, you’re essentially wasting 1m. Having all of it staked would actually be cheaper than validating yourself. Staking the overhead makes you very inflexible and at risk of dropping out at the next Bid-increase.
Staying in the game requires a significant amount of personal resources. Due to the gamified nature of EPOS, you could drop out by being outbid, so you have to particularly pay attention at Epoch changes. This can be very time-consuming as it happens frequently and at unfavourable times.
Bottom line I think it’s definitely a good step, but won’t solve the problem of centralization. Due to the complexity of the game, I see stakes further centralizing to a few big players who have the manpower & resources to maintain efficient bids further down the line. I simply don’t see a reason to spin up a validator, as staking will remain more profitable than validating myself.
You’d need to make validating more attractive compared to staking. Currently staking wins by far.
Increasing the number of validators having Binance holding almost 2/3 of the total stake and distributing its stake in just 20 validators is an impossible task with the current rules. Either ask them to redistribute their massive stake with all validators even the not elected or just make enterprise accounts for exchanges.
Promote comission free harmony staking pools limitting max delegation to the median.
Pretty difficult task to carry out with just soft measures.
let’s assume between the bid in the range 0.85 - 1.15 (of the median stake) and we create 3 sub range:
1/ 0.85 - 0.95
2/ 0-95 - 1.05
3/ 1.05 - 1.15
for each sub range, we applied an additional logic in the effective stake calculation as follow ie :
1/ (0.85 - 0.95) of those bid would have 10% (or any other %) of their effective stake being taken away or slashed
2/ (0-95 - 1.05) nothing will happen for them
3/ (1.05 - 1.15) of those bid per keys would receive the 10% (or any other %) taken away from 1/
as mentioned above, we want to give a bigger reward to the validator that would use less keys (ie having a higher bid per key) and hence free up more slot
feel free to ping me in discord/tg if this is not clear enough
