HIP-30: Validator Network Optimization and Token Emission Reduction

@Casey I think we need to revisit the number of keys allows per validator and or per shard. With the current rules and proposed we hit an issue where we’d have around 20 active validators only vs ~130 now

  • looking at key usage of the current big validator : https://h.vstats.one/node-versions
  • many have way above 12 keys
  • 12 keys because with 100 keys per shard and 6% limitation per shard for a given validator, that give us 6 (keys) x 2 (shards)
  • we count already more than 20 validators having more than 12 keys
  • 20 x 12 = 204 which exceed above the 200 total keys spread between the two shards

the big ones are not necessarily the most active ones …

I actually thought with HIP30-3/4 we would even do a drastic change like 1 bls key per shard per validator per node. Ie the big validator can only register up to two keys (one in each shard).

However that may add other issue link to EPOS and the EMS being too low. The big validator and their delegator will earn less reward.

4 Likes

Just a thought, could we do away with epos entirely?

No upper or lower bounds or EMS.

We all need one key per shard. Each validator therefore is required to run a server for each shard.

Being elected is based off total stake.

Top 100 validators in stake are elected.

All validators return the same APY as a result.

@Casey

4 Likes

Agreed @FortuneValidator - it may be a good time to reconsider EPoS altogether. Those who designed it are no longer with Harmony and it adds a high level of complexity to any proposal that seeks to reduce the number of keys or nodes.

4 Likes

Indeed. It simplifies the process.

Guarantees 100 validators rather that the potential of 20 or so on the current proposal. Largely due to the pressure for larger validators to spin up more keys/servers being removed.

Validators gain stake by being active within the community which is ultimately what harmony needs (Engagement, coupled with quick node response times).

It also ensures an even balance of validators across two shards which helps the security aspect and shard maintenance/ updates.

3 Likes

I like this more than the original proposal.

Is there any reason a validator couldn’t be limited to only 1 key on 1 shard (no second shard)?

Wouldn’t that incentivize delegators to redelegate to higher rewards-earning validators, which would improve decentralization?

5 Likes

We’re close to our 2nd anniversary validating for Harmony and we learned and observed many things during this time. Ill try to gather here some of the aspects that we think that are good and bad for each proposal and what we think about that. When I say WE , Im talking about the brazillian community, since we discussed about that and I share some of their concerns as well.

1. Enhancing network performance through sharding and node optimization.

We all agree that having 4 shards back then was a big marketing, and brough some attention to harmony. But since it wasn’t properly used and the escalation didn’t ocurred as expected, having them is causing more problems than bringing solutions. So the infrastructure could be reduced to 2 shards and once the need arise it could be expanded again, provided it would be an option. We believe that it would be the best aproach regarding the shards issue.

2. Ensuring long-term sustainability with a reduced token emission rate.

Our main concern with the “long-term” sustainability is that its a broad range spectrum. It can engulfs anything. We have bad past examples (DAOs, RPCs, 1Alliance, etc…etc) that harmony didn’t managed very well their treasury. And it still didn’t learned the lesson if we look into the R1 issue brough by @TrickLuhDaKidz where he digged enough to uncover all of this. So, the community is genuinely worried to pay for something that won’t work, and its fair, they already had their share of pain. Their trust on harmony is damaged. Can’t stress enough how it is. Many of my delegators keep asking IF harmony will recover someday. I try to keep them calm and expecting positive things. But then, some bad info about something happens (was the R1 recovery plan this time, after that quoc shitshow). So, unless we have soke detailed info about that, on HOW the rewards will be spent, for how long, etc…and etc. I’m pretty they wont agree. Also, one of their worries was about the sell pressure that it would fall on the token, since a the staking APY will fall hard. How the staff plan to remain competitive with other protocols? Right now, 8% is not enough to attract a lot of investor, due to the past happenings. It could, but the hack was not solved yet, the funds are long gone by now. So HOW, in this world ( maybe in another parallel reality it could work), half or less APY than the actual will help to bring more investors?

3. Increasing validator rewards by reducing the number of nodes and redistributing rewards.

This could be a positive point. We would have been stressing the need for limiting the BLS keys and avoid the Lower Bound Divings, long enough. The fact that at least I’m not seeking for larger rewards. The leveling aspect of this proposal is positive in our point of view. Just can’t agree with this 200 slots limit, unless it means that it would be run entitely by validators, without internal nodes. Than it would mean we would have space for more external validators, having the network even nore secure and decentralized.

4. Encouraging new validators to join the ecosystem and fostering healthy competition

This is a point concerning the community as well. We saw many arrangements made by harmony staff within validators that didn’t helped or were present at the community, did lower bounds diving and harmony wasnt exacly clear about that arrangements and how it was benefiting the network. So, unless it is more detailed on how the pre-requisites for selecting the so called “network empowering validators” would be chosen, if these validatos would be required any performance to be remain elected or would be needed to be evaluated by the community every X months. Something would be need to be enforced to empower the community. Can’t have the staff spending at his discretion only.

So to finish this, I believe this proposal need to be more discussed AND enable more safeguards to be in place that make the community in charge of the changes not as a passive watcher.

3 Likes

Just realized I made a very silly math calculation here lol. it’s not 20 validators but 17. 17 x 12 = 204.

1 Like

No reason we could do that as well. However, 200 unique validator is a big number and nothing prevent current validator spinning up a “second” validator.

Yes, correct.

to me the issue with this is the stake imbalance. Big validator can have all the stakes without any penalty.

No internal nodes.

1 Like

I understand this. However, the current proposal will lead to the potential of 17 network validators due to larger validators adding keys to get out the upper bounds. Asking a validator to stay in the upper bounds and generate a 0.something% is not reasonable and as such new keys will get added. The other proposal someone mention of a hard cap of the single key will end up with second validators being spun up and likely won’t pass a vote either. I was simply trying to find a middle ground that may appease a broader scope of validators.

Could you have it whereby no one validator can hold a certain % of overall network stake? Whereby no additional stake can be added or with a lower APY penalty? Binance validator stake size being the only issue I can see in removing epos.

1 Like

Just want to bring something up.
As far as I’m aware Binance has been doing us all a solid by sticking all their staked coins into one pool that is running at about 1.7% interest. I think this is because now the key bidding currently works, so If we changed how the keys work, would binances huge staking pool start taking up more of the interest and compete with all of us? They have 783,000,000 One staked currently not competing.

2 Likes

I’m ok with that.

A large validator needing to create a second validator and having to compete “from the bottom” again sounds fair.

I also think it would make the “burden of entry” easier for new validators to start on Harmony in the future, which could be beneficial if Casey hopes for Harmony to attract validator-developers (as per his Medium article). It was hard enough trying to get a new validator elected when there was 800-900 keys, i can only imagine how hard it’ll become with just 200.

My concern is the 20% of currently elected validators that would be removed (assuming 1 key per validator per shard is what’s implemented). There would likely be quite a few active and engaged validators who are culled. Doesn’t seem appropriate to ask them to do the ultimate sacrifice (killing their validator) to save Harmony from its self-inflicted wounds.

Does Harmony still intend to use the Foundation funds to “rebalance” the validators? What are the specific parameters that will be used? Can we have a list of current validators who would be eligible? @Casey

If we knew which validators would be part of that, then we could have a better idea of who would be “safe” from HIP-30.

Sounds like a good reason to implement it.

That isn’t a new threat. The possibility already exists for just 17 validators to control all of the keys once their stake becomes large enough.

I understand though, as currently proposed, that moves from a possibility to a potential reality though. Which is why I think Soph’s suggestion is good.

Not sure how that would work? What would the max % of total network stake be set at?

Example percentages:

• 1% (i.e., 100 keys) = 55.6 million ONE - 27 validators currently above this
• 3.5% (arbitrary middle ground) = 194.6 million - 6 above
• 6% (current 6% limit) = 333.6 million - 3 above

Wouldn’t an APY penalty be functionally similar to what EPoS currently does?

There’s also nothing preventing a validator that’s over this hypothetical % from spinning up a second validator. So in general I don’t view “spin up another validator” to be a compelling counterpoint. A “validator 2” would at least allow the possibility of more competition and decentralization.

3 Likes

The difference is right now there is no major incentive for this to happen. If you move to 1 with WITH epos then you just chucked a significant amount of validators in the upper bounds and thus would now have more a reason to add keys with extra servers or loose what they worked for ( community validators )

This is where my 1 key per shard per validator WITHOUT epos would remove this from being an issue. It guarantees 100 validators and forces network participation to gain trust and thus delegations.

2 Likes

I also have the same thoughts even before reading this thread. I will support this proposal if we remove the EPoS limit and compute rewards based on raw stake. This in addition to a 10% minimum fee increase to account for the 50% loss in rewards.

5 Likes

Having EpOS removed entirely, could lead to harmony downfall. Imagine rewards received by binance delegator being dumped entirely in the market. Aren’t you guys afrad of that? Some sort of % limit shoud be lifted if EpOS is removed. Having a cap in how much rewards a validator is allowed to receive is the way to prevent any of these occurrences. Removing EpOS could be ok if a cap system on reward per validator is enforced.

3 Likes

Indeed. I’m currently of the thinking of either:

  1. Remove epos
    Or
  2. Remove key limit from this proposal.
1 Like

@FortuneValidator and @RoboValidator how you guys plan to prevent the rewards to be prompt dumped on market by the massive validators (i.e kysen pool, binance, etc)? What are your thoughts about that? If it goes based on RAW stake, it will empower even more the big ones.

Reward dumps or sell pressure shouldn’t be a concern of validators. Those are free market principles that are outside of our control. Whether its Binance as the largest validator by stake weight or a validator with the lowest stake weight, there is no difference. Once the token emissions enter the circulating supply, they are subject to the free market. One could argue that by reallocating staking rewards to recovery and ecosystem development, sell pressure would increase because the funds would then be certainly used/sold for recovery and to fund developer expenses, including servers, salaries, contractual obligations, marketing, etc.

4 Likes

Binance would get 14% of the rewards. Or 2,707,249.6666667 ones monthly if the rewards are cut by 50% and based on raw stake. Sincerely…can’t think of this as a good solution without a cap.

2 Likes

This will go on TOP of all you said. Dont forget that 221m ones would be spent at staff discretion. Could be gone as soon as hit the treasury or in parcels. But it would be gone. And the ones that would received it, undoubtedly, will spend to pay their costs, since they’re receiving it as an incentive.

I doubt that it happens RN since many validators like me, didnt sold any rewards yet. Always reinvested in the protocol. So this sell pressure wont happen now. But you could count on that to happen later if the rewards are cut and spent that way.

1 Like

Agree with these two options. We already went through a lenghtly governance process to determine that a 6% key cap limit per shard was the most reasonable limit within the constraints of EPoS. Any further adjustments to key cap limits would almost certainly require an adjustment to EPoS bounds that offsets any negative impact to validator earnings or the elimination of EPoS altogether.

HIP-30 discussion should focus on the temporary reallocation of annual token emissions to recovery. Any additional measuras only add to the complexity and reduce the palatableness of the proposal.

3 Likes