[Proposal Draft] Increase Max Validators to 150

In the Cosmos Hub, the total number of active validators is currently capped at 100, ordered by total delegated Atoms. This number was originally proposed in the Cosmos whitepaper section titled “Limitations on the Number of Validators” (https://github.com/cosmos/cosmos/blob/master/WHITEPAPER.md#limitations-on-the-number-of-validators). This number was chosen as a relatively conservative estimate, as at the time of writing, it was unclear how many widely distributed nodes Tendermint consensus could scale to over the public internet.

However, since then, we have seen empirically through the running of the Game of Stakes incentivized testnet that Tendermint Core with Gaia state machine can operate with over 180 validators at reasonable average block times of <7 seconds. At the time of submission of this proposal, the minimum delegation to become a top 100 validator is ____ Atoms, a fairly high barrier to entry for new validators looking to enter the active validator set. The tradeoff to higher number of validators is that the size of commits becomes ~50% larger due to more precommits being included, increasing the network and storage costs for nodes. This, however can be resolved in the future with the integration of aggregate signatures.

In the Cosmos whitepaper, it states that the number of validators on the Hub will increase at a rate of 13% a year until it hits a cap of 300 validators. We propose scrapping this mechanism and instead increasing the max validators to 150 validators in the next chain upgrade with no further planned increases. Future increases to the validator set size will be originated through governance.

8 Likes

Looks good. Can you clarify if this is intended to be proposed prior to the next hard fork (and thus would be a change in the genesis configuration)?

However, since then, we have seen empirically through the running of the Game of Stakes incentivized testnet that Tendermint Core with Gaia state machine can operate with over 180 validators at reasonable average block times of <7 seconds.

why not increase the set to 180 then ?

Because it’s half of the original upper bound of 300, and a 50% increase over the current amount. It’s more human natural :slight_smile:

2 Likes

This plan seems flawed.

We already have a situation where the economic value of the network is not sufficient at current distribution for validators with less than a certain threshold of atoms to validate in an economically viable way.

The conversation of centralisation of stake is an ongoing one, although IMO there is not yet a sensible solution to this, but the network is young and forcing decentralisation in protocol is both premature and sets a precedent that we may wish to avoid, but I digress…

Adding 50% more validators at this point seems premature. We’ll go from c 50-60 validators struggling to survive to 100+. We should wait until either the network is more decentralised, or the economic value of the network increases sufficiently to make the addition of new validators economically realistic.

This problem of affordability of staking is exacerbated by those validators (Sikka, Sparkpool, Forbole and others) that insist on charging 0%. New validators will almost certainly see this is the de-facto standard they need to compete with to have a chance of attracting delegators, and you’ll price every single one of them out of the market.

The only real reason to add new validators is to increase the security of the network. At the moment such a proposal will fail to achieve this. We need to resolve other issues before this - the biggest of which is to drive value to the network by implementing IBC.

But I also implore all those charging 0% to consider the damage you are doing to the network and it’s smaller validators by continuing to charge 0% commission, as it is every validator’s responsibility to protect this network and it’s longevity.

3 Likes

We would strongly accept the proposed idea. The current min. stake requirement with over $100.000 is simply crazy. Compared to e.g. Ethereum 2.0 with 32 ETH - currently roughly $7k.

It’s not a good idea to create a tough competition for small validators to survive (=not to drop out from the active validator set). In the end, some of them might feel incentivized to contribute in case they are able to validate.

On the other hand, the potential danger (even if small) is that these validators simply “fork away” and/or create their own tendermint-based chain. So in my opinion, let’s be as “permissionless” as possible.

2 Likes

Makes sense to me, allows some of the lower validators who are building the delegator base to not live in fear of being pushed out.
Although, I would say this - Valid8r looking after the bottom of the set.

I see little negatives to this, increase it to 150.

1 Like

I disagree, not because this proposal has bad intentions, but it is hard to find a good reason to increase validator set to 150 at this moment.

The current network is very much controlled by top 15 validators who hold more than 2/3 of the voting power. Opening 50 spots for new validators will not have a significant impact with the current voting power distribution. Even in security perspective, adding 50 more validators does not seem like a big improvement.

From what I remember, while it is a good thing that Tendermint and gaia was able to handle 180 validators on Game of Stakes, stakes were not so evenly distributed back then, and we cannot be sure if running 150 validators with well decentralized voting power will be able to achieve <7 seconds block time. We should probably test the chain with increased transactions to see if the chain works well before we increase to 150 validators. (I believe problems could be bigger if we increase to 150 validators and have problems with increase transaction in the future)

Although I disagree with this proposal for the moment, after the network matures more and larger transactions are tested with 100 validators, this proposal will be a more agreeable one.

2 Likes

No one is forcing decentralization. It’s allowing for decentralization. Who are we to decide what is “economically realistic” for others? If new validator operators feel that they want to enter the validation space and have a strategy for “economic feasibility” and success, this is enabling them to do so without needing to instantly secure the >$100K delegation needed to just enter the validator set right now.

In the discussion around delegation centralization, we’ve been discussing the issue of delegators not looking at validators outside the top 20 or so. Well, you know who they’re definitely not looking at? Someone not in the validator set at all. Reducing the barrier to enter the validator set will allow new operators to enter and be able to compete rather than locking them out completely.

I have addressed this claim here: [Governance] Limit validators from 0% commission fee

Were they not? Didn’t all the validators start with the same amount of stake? It was probably more evenly distributed than the mainnet?

No one is forcing decentralization

I was referring the numerous suggestions that have been floated recently to make protocol level changes to punish delegators (economically, or otherwise) whom delegate to top X validators.

Who are we to decide what is “economically realistic” for others?

Neither of us are in the position to do so, for obvious reasons. However, there are a significant number of validators towards the lower end of the table who are talking about this on a daily basis because it is simply not affordable to run a validator when they cannot earn any kind of revenues because they feel obliged to charge 0% to stand any chance of attracting delegators.

If it’s a significant problem at 100 validators, adding another 50 to the set and diluting what rewards there are sure isn’t going to help the situation. The number of validators charing <= 3% is demonstrably greater than it was ten weeks ago.

My question of new validators feeling they need to charge zero, or near-zero to be able to compete for delegations when they are starting out, is not answered. As you rightly said, you are able to do this because you have full time jobs, have cultivated the reputation by working on the project and are able to get low/no-cost infrastructure. Sure, business is business, but by charging zero percent you are forcing any new entrant to make that decision between charging zero/near-zero commission and making a pittance or never getting much in the way of delegations and making a pittance.

Oh this was a mistake in my part. I only thought about how GOS ended not how it began. My apologies

Not punishment, but incentives. This is quite different.

I do agree with Joe that centralization is a more crucial issue right now. Validators charging 0% fee are damaging the network in multiple ways. (Forcing centralization, making StaaS in theory unpractical, destroy competition amongst validators)

Nonetheless, I do not see any problem with adding another 50 validator slots. It means little effort, and what is the worst case? New validators try their luck and fail? In the best case, we’ll have a set of new validators that foster the decentralization of the network.

I am pro for this proposal for the following three reasons:

  1. I see no negative effect in doing this, on the contrary it can provide a higher exposure for the entire cosmos ecosystem
  2. Possible some new valuable guys or girls will join the validators group, people with high potential that may provide themselves as great team players and contributors on the long run
  3. This will show to everyone that Tendermint BFT can run a large-scale distributed network with 150 validators and I think that this may create a first timer event on a main net.

The only inherit “negative” is the 50% increase in storage space due to precommits and votes. Regardless, I am in favor.

My position is against the increase of 50% of the slots. These are my reasons:

  • It wont help to solve any actual problem of the network.
  • It wont increase actual network value.
  • It can be consider a call to work for free, without any reward.

If the “talent recruitment” is something we aim with this proposal, I think there is a way better to do it. Any team of talent people below the 100 may be rewarded and pushed up into the 100 by delegating the communty founds. That team could stay below the 100 for a time, and earning exactly the same as bein the 149 in the propossed scenario. Zero. If their contribution to the network is good enought we could “cherry pick” those teams and include them in the top 100 by governance.

Actually you can’t say for sure, it’s just your opinion for the first two points, maybe it will not increase actual network value but it will increase future network value (providing a clear example how a large number of validators can run on Tendermint BFT is a reason for network value increasing, for example)

The thing is by increasing the slot numbers is bringing no negative effect, and just think about it, everybody loves open-free systems (humanity fought and died for freedom in the history) by starting to keep Cosmos a closed hard to enter system we will not be sending the right message.

Totally agree that this is not going to solve any actual problems of the network (who are not so critical at the current stage) but on the other hand it brings no negative effect either. Regarding your 3rd point, well it can’t be considered an exploitation as exploitation has restrictive components (for example restriction from freedom, restriction from resources, etc). People who will enter will do it from their own free-will, they will not be forced by any means.

1 Like

for my understand if with 100 validators is not decentralized, why is going to be decentralized with 150 validators?
the first step is to do some like Harmony, change the code first, then open more slots for newcomers can compete in fair terms, if now open the doors to 150, then is going to be more centralized and tomorrow or someday when the code change for cape the figure of the king validator is going to be more difficult for newcomers for what cosmos is doing now. IMO no have to explain more because is easy to understand.

I am in favor because it will decrease risk of sybil attacks lowering entrance boundary.

I am against fixing to 150, it should be possible to increase further in the future when deemed necessary.