[Proposal] Decreasing Max Validator Set

Hi All! I’d like to open the topic of the maximum validator set, which is currently set at 80. There are only 74 validators actually in the set, which indicates this parameter is set too high.

Context and Importance of Max Validator Set

The Max Validator Set parameter (MaxValidators) determines how many validators are able to be part of the active set on Secret Network. In the past, this set was increased twice (proposal #59 - change to 70, proposal #85 - change to 80), and a proposal to increase it again failed (proposal #107 - change to 90). There was lively discussion on the pros and cons of the change - see the forum post for #59 for context.

The current validator set is too large to optimize given the price of SCRT and the delegations that are distributed to validators. Several of the validators with the lowest delegations are known to be inactive.

For reference, a validator with 5% commission should earn at least 200 USD per month with a delegation of 600K SCRT (based on validator feedback during the Tokenomics research.) With the current inflation and price of SCRT, the monthly earnings are significantly lower than this amount, which means that delegations would have to increase to achieve breakeven for many of the validators. Given the total staked supply, this is not likely to happen unless some of the current delegations are freed up and re-distributed.

Our Proposal

Validating on the Secret Network is getting more expensive, and some validators are not contributing a great deal to the network stability at this time. In order to improve the quality of the active validator set, we propose to reduce the number of validators in the set from 80 to 60. Since there are currently 74 active validators, this reduction will effectively reduce the set by 14.

The goal for this proposal is to:

  • increase network stability

  • redistribute delegations to higher performing validators

  • increase the competition to remain in the active set, thereby improving the performance over time of validators and of the network

  • prepare for the possible transition to higher performing hardware in the future

Some things that will be measured before and after the change:

  1. Average blocktime

  2. Missed blocks for validators.

  3. Network stability & performance

If the results are good, then a proposal to decrease the set to 50 Nodes will be submitted in the future.

Tokenomics Considerations

Based on the tokenomics research, there is a bias towards setting a minimum validator commission. This has no real bearing on this discussion since we have used an assumption of 5% minimum commission in the modeling.

Inflation changes may also have some impact on the outcome of this proposal. Based on the tokenomics research undertaken by SLabs, inflation is recommended to stay at its current level of 9%. There have been other proposals suggested to increase inflation temporarily, which would potentially change the economics for validators, but on the other hand this increase in inflation has been proposed to be reserved for building up a treasury to be used for liquidity provision and other initiatives. In the latter case, it would mean no material impact on the economics modeled for this proposal.

The research did not address the max validator parameter specifically, but there may be other considerations from the tokenomics research that we have not outlined here.

Request for Community Feedback

This proposal is open for community feedback. We welcome any thoughts, suggestions, or additional insights you care to share. We recommend you review the forum discussion for proposal #59 for some well thought-out arguments for and against changing the validator set. Your insights will be invaluable for refining this proposal and ensuring its success.

7 Likes

Seems good. People should be upgrading hardware as part of improving the network performance anyways, which will cost more, also many validators operate at a loss.

I don’t see sufficient value in reducing the max validator set. Secret Network is already on the lower end of validator counts in the Cosmos ecosystem and the benefits of doing so is not inherently clear.

Increase network stability:
Not able to speak to this on a technical matter. However, I would assume that validator count is not the main issues that need improving for network stability.

Redistribute delegations to higher performing validators:
The bottom fourteen validators that are in-scope to be removed hold a combined 1.68% of voting power between them.

Rank Name Total Staked (in SCRT) Commission Annual Income VP (%)
60 :seedling: ActiList.io 326,099.43 5.00% $1,208.05 0.21%
61 AgoraNodes 300,034.84 5.00% $1,111.49 0.19%
62 RC DAO 292,680.15 10.00% $2,168.49 0.18%
63 ELYSIUM 251,805.43 10.00% $1,865.65 0.16%
64 𝕊hadowRealm🥷🏼 231,487.08 9.00% $1,543.59 0.15%
65 Outlier Ventures 219,171.28 10.00% $1,623.86 0.14%
66 [ block pane ] 214,375.89 5.00% $794.16 0.14%
67 CosmosPug 183,542.54 5.00% $679.94 0.12%
68 secretnft.com 159,818.22 5.00% $592.05 0.10%
69 HBSecretNode 155,043.78 20.00% $2,297.46 0.10%
70 MTC DAO 140,698.54 5.00% $521.22 0.09%
71 Colossus 81,577.66 3.00% $181.32 0.05%
72 SpaceStake 75,646.31 1.00% $56.05 0.05%
73 cosmosrescue 37,743.97 5.00% $139.82 0.02%
74 figo 119.41 100.00% $8.85 0.00%

Each of these validators earn an immaterial amount and the vast majority of them operate at a loss. This income redistributed to larger validators will likely have little impact.

Increase the competition to remain in the active set, thereby improving the performance over time of validators and of the network:
Reducing the size of the validator set “forces” competition where it doesn’t have a need to exist. At present, there are eight validators with less than 100% uptime over the last 500 blocks. Three of the eight validators are below Rank 60.

Contrastingly, the remainder of the validators with less than 100% uptime over that period are all above Rank 35.

Prepare for the possible transition to higher performing hardware in the future:
As noted above, validators are already running at a loss. A possible transition in the future can happen and validators will naturally be culled if they’re not willing to run at a larger loss (assuming equal delegations and higher costs).

We’ve discussed reducing validator sets before when we were hovering at 79 validators in the 80 validator active set. It’s naturally shrunk to 73/74 validators, and will continue to do so.

1 Like

RE: Technical side

I’d like to see secret labs comment on a timeline for this, and they should do so as soon as possible so there is as much advance notice as possible. Hardware side has been discussed in a few places but here is the TLDR.

  • New hardware has a higher cost. Think $250-350 monthly in many cases.
  • Smaller set can certainly improve performance, this is fairly well accepted as true based on public data.

Ideally, a tighter set may actually encourage remaining validators to improve their setups though increased competition, which could be good with the upcoming launches.

  • Yes, it is well accepted that it can improve performance. However, there also appear to be several significant initiatives on the 2024 Roadmap to improve performance. Could someone suggest that reducing the validator set to 60 validators would move the needle at all in performance? Anyways, there is a diminishing marginal benefit to network performance with reductions in the validator set so it can only be done so much before it is relatively meaningless.

  • At the cost that you cite, even validators within the 40s are going to be breaking even, if not loss-making. Even if you redistribute the 6.61% of VP that is with the validators below Rank 40, it will barely improve the situation much for the remaining validators. Do we then cut the set to that? Or do we then accept that being loss-making is fine and validators can exit if they want?

  • What is being used as a metric for seeing who has “good setups”? Uptime (which is 100% for the vast majority of the validator set)?

Seems highly unlikely to come before launches. Keep in mind, a person can ddos APIs simply by using shade for a small number of queries / usage and if they get actual usage tx wise, we’ll see a lot of missed blocks.

Sure. I’m 90% talking about performance so ofc it’s fine for validators to exist if they want.

At this point the real metric is probably if they use old or new hardware. Without getting into the weeds, I think it’ll be that simple. There isn’t really new hardware that will perform as poorly as even the best old gen machines. Maybe uptime will still be somewhat wonky but that can be for other reasons, new hardware should handle actual tx load better. I have done testing that informs some of these things and it’s clear new hardware performs better.

  • Do you think that reducing the validator set from 74 to 60 will really change anything in terms of spamming queries to degrade Network performance? Seems like there are bigger issues that need resolving on that front?

  • Disclosure of validator setups is entirely optional though.

Spamming queries? No. But if shade is getting queries “legitimately” spammed it usually is accompanied by tx’s too and that’s where missed blocks and the users saying “network problems” comes into play.

Sure. But we can look to performance, and we could change it to show what cpu publicly. Point is still if competition increases, people may find ways to do a better job in these situations to prove worth.

1 Like

We are open to supporting this as also communicated before by mainly me in the Secret Governance telegram.

I think it would be good to try and reach out to all of the bottom 24 teams though and see if some people might want to spin down to give others a chance.

I think it is likely there are people who would be happy to spin down but havent done so because of potential negative backlash, if they got the chance to do so and keep a spot open to someone that does want to stay in (new joiners like CosmosRescue, SpaceStake, colossus) then they might take it.

Best,
Ertemann
Lavender.Five Nodes

1 Like

I support lowering the max validator count. I wouldn’t mind seeing it be less than 50.

1 Like

I‘ll also throw in my 2 cents here from a performance perspective:

  • Reducing the validator set improves CometBFT performance (https://www.inf.usi.ch/faculty/pedone/Paper/2021/srds2021a.pdf) and we have less enclaves that have to compute the same state.
  • Hardware upgrades with a smaller validator set are easier to enforce, simply put we have less validators with less coordination + each validator that is in the active set has a higher delegation
3 Likes

Going to 60 removes a few with known low quality equipment. Below 60 probably has barriers to passing, but sure it would be better for performance. I wouldn’t mind a lower number than is being proposed either, but the proposed number is probably the most realistic in terms of passing. Then I’m sure most of the hosted validators will likely all fall off once new hardware is enforced because those costs will balloon, though doubtful that happens this year before upcoming launches.

1 Like

Absolutely agree.

In addition to this, I’d like to add a few points.

  1. Why is the size of the validator set set because of the token price?

The current validator set is too large to optimize given the price of SCRT and the delegations that are distributed to validators.

Isn’t that a technical question? As I understand, the team has to decide, which size of validator set should be. Sure, the less validators - quicker consensus. OK. But that is the team question. They should say and adviese/decide what size of validator set should be. If the project needs speed, like Injective (60 validators) for example, or more decentralization like Cosmos (180 validators).

What should we do, if the price goes down 50%? Should we decrease it to just 10 spots or what?

For reference, a validator with 5% commission should earn at least 200 USD per month with a delegation of 600K SCRT (based on validator feedback during the Tokenomics research.)

No. Validators don’t have to get something. That is entrepreneurship, business and competition.
You get income and decide, is it OK for you or not. You know rules about expensive tech requirements and so on, so it is our (validators) question to take part in this project or not.

If validators do not agree - they shut down their servers. Project gets less validators → faster consensus. Every remaining validator gets more inflation and commissions, so they start to earn more. And that works until it is profitable (in addition to this, we can count profit not just in tokens/money, but for example as the amount of projects that this current validator serves).

Take a look at Solana. When the token price was close to $10, validators were selling their spots. But guys, who were interested, were buying them, holding losses until the price got back.

increase network stability

How? Secret Network doesn’t have holidays, as Solana does. Uptime 98-100% absolutely not critical. Throwing out of 14 validators won’t change anything in this case.

redistribute delegations to higher performing validators

Why can’t we do it right now? Without reducing the validator set?

increase the competition to remain in the active set, thereby improving the performance over time of validators and of the network

I don’t understand how it works. Validators won’t change servers just because… In the case of 60 validators they just will get a bit less load. They don’t throw flyers, that their validator is the best one. So, this proposal will just limit the competition competition.

prepare for the possible transition to higher performing hardware in the future

For this, validators HAVE to earn more. And they won’t because of the following point.

  1. We need a little math here.

As we can see at info about 14 left validators, total their income is about $15 000 per year. OK. somebody has 10% commission, somebody has 5%. OK. If all of them will set commission at 5%, income will be ~$10 000.

If we do very simple math, we will have the following: spreading this income at the first 60 validators will mean that the first 10!!! (they have MORE THAN 50% of voting power) will get half. They all have ~5% commission, so that is $5 000. They have so big stakes, that income won’t be a life change for them. But another 50 validators will get +$100 per year.

Maybe, if +$100 per year is significantly important for this validators, they better go to the job?
How can they upgrade servers for this money?

That is a fight for pennies, not for the money.

If we need to increase validators income, we can:

  1. Reduce inflation (Supply go down, price go up)
  2. Change tokenomics/burn tokens
  3. Develop projects and make ecosystem more popular.
2 Likes