Abstract:
This proposal recommends reducing the maximum number of active validators from 100 to 75 to enhance network security, mitigate validator dilution risks, and ensure higher-quality participation in consensus and governance.
Context & Motivation:
The current validator set limit is 100, yet only 81 validators are active, and among those, the bottom seven have extremely low staked tokens.
The average $BLD staked by validators is 6375114, and median is 3083541
The bottom 7 validatorcumulative share is 61919 $BLD with an average of 8845 $BLD per validator.
The value of average $BLD staked for the last 7 validator is ~120$ worth of BLD at the price of $0.01340
This situation introduces multiple risks:
Sybil-like threats: Low economic barriers allow malicious actors to spin up multiple validators and potentially influence consensus.
Censorship & downtime: Low stake validator might not have enough incentivization to maintain uptime or prevent transaction censorship in the future.
Validator quality variance: Lowering the entry barrier may bring in low performing validators
Proposal Details:
Parameter to Change:
staking.max_validators: Change from 100 → 75
This change will:
Encourage higher stakes per validator.
Incentivize better infrastructure and uptime performance.
Strengthen the economic and governance integrity of the validator set.
Increase the competition to remain in the active set, thereby improving the performance over time of validators and of the network
Implementation Notes:
Validators currently ranked 76–81 by stake will be removed from the active set upon implementation.
They will retain their delegations and may re-enter if their stake rises.
Conclusion:
Reducing the validator set to 75 reflects the current network conditions and aligns with best practices for validator quality and decentralization. It increases the chain’s consensus speed, attack resistance, and improves reliability.
Open Call for Community Feedback
Community and Validators are invited to share feedback on this proposal. If there’s no significant objection, the on-chain proposal will go live next week.
Thanks for starting this — totally agree it’s time to reduce the validator set. Given the comparables, does it makes sense to go smaller now? What’s the minimum for a robust, geo-distributed validator set for deploying applications like Ymax?
Also, with a reduced set, we could also lower overrall emissions (e.g. 5% → 4%) and still increase rewards to validators.
DCF has been thinking through the numbers and can share more detail. Curious to hear what others think.
So, as promised on X earlier today, here are some further thoughts on this, from the DCF perspective:
We’ve been discussing taking steps internally that are in line with this proposal. Here’s an excerpt from the internal draft we’ve been working on (we’d planned to post Monday, but this post came up first, so let’s engage here!):
================
"We think it’s time to initiate a discussion within the Agoric community regarding potential adjustments to the Agoric validator set size and the network’s inflation rate. This discussion is a direct response to macro market conditions and the depressed price of BLD. We posit that these changes are sensible and help to ensure the economic viability of validator operations and the ongoing security of the network, as well as delivering some relief on BLD price pressure by reducing inflation.
We invite discussion on two key changes:
Reduction of the Maximum Validator Set Size: From 80 to 30.
Reduction of the Inflation Rate: From 5% to 4%.
Both of the changes are executed via parameter adjustments.
Rationale
The reasoning behind these proposed changes is two-fold:
Enhanced Validator Economic Viability: By reducing the number of validators, fewer validators will share in the rewards per epoch, thereby increasing validator income and helping bolster economic viability. This is crucial for attracting and retaining high-quality validators who are essential for network security.
Reduction in Current Minting Inflation: Reducing the inflation rate will slow the minting of new BLD. In the current market climate, this adjustment is necessary to maintain a healthy supply-demand dynamic for BLD.
Impact
We anticipate the following positive impacts from these changes:
Bolstered Network Security: A more economically viable validator set will attract and retain robust validators, enhancing the overall security and decentralization of the Agoric network.
Improved BLD Economic Stability: By slowing the rate of new BLD issuance, we aim to contribute to greater economic stability for the BLD token in the long term."
================
@Ciberexplosion 's post, above, makes some of the key points with the underlying metrics, so we won’t repeat that. Suffice it to say: We feel the analysis is correct and that this will deliver benefits to the community as a whole and to a key constituency – validators – in particular.
Considering Pareto distribution, ~80% cumulative share of the network is a good benchmark to realign the protocol.
I support changing the validator set to 30.
Regarding Dean’s reply on aligning the next steps with YMAX
Message Complexity for Tendermint BFT is O(n²). Latency increases sharply with validator count, and Throughput drops off as nodes grow past 50–100.
There is a general shift by newer Tendermint chains to have fewer than 50 validators, and 50+ validators are leaning towards Hotstuff - O(n) based blockchains.
I would like to consider Sei Network with $612.39m TVL as a reference point. They have a validator set of 40.
Consideration on Geo-Distribution
The following reference is from Messari’s article with data from May 26, 2023. While we strive for maximum distribution progressively, the insights on distribution are helpful. The country-wise distribution remains low across blockchains.
I believe 30 validators will be sufficient to provide geo-distribution that meets the current standard across major Layer-1s.
The current Bonded Rate lies at 46.27%. What is the Bonded Rate target?
Lowering the emission alongside reducing the validator set can reduce the bonded rate.
Would it be a better idea to reduce the validator set first and observe the effect on the Bonded Rate over 3 months? The insight can help in coming to a more conclusive number for emissions.
at once, we get it done and past, but there are key service providers (e.g., relayers) that need nodes that would drop out of the set until and unless stake got shuffled around.
in stages, it takes longer, but each transition is a lot less disruptive.
You want to cut validators, you want cut BLD inflation , you have cut team, you have closed Inter Protocol. What to expect next? What was the reason to have 100 validators in the past, not 50 not 30?
I know you posted in response to dtribble’s comment, but I’m going to jump in any way. I wasn’t around when the decision to set the maximum size of the validator set to 100, but as I understand it, that number was chosen to allow for a decentralized and robust validator set. (Moreover, at that point in time, a lot of projects were selecting 100 as their starting value – and a number have subsequently modified that figure downwards.) When projects set those figures, they are typically looking at allowing for room for growth and the added resilience that comes from more operators. That isn’t always needed, and when it is not, decreasing the max size of the set is an appropriate response that allows the validators to earn more and decreases complexity at the time of chain upgrades and such. That’s a healthy change that is made in direct response to changing needs.
Similarly, a lower inflation rate is good for the token and the token holders. Not sure why anyone would object to that change. Lowering issuance means fewer new tokens coming to market, which helps take some pressure off the price. Again, a healthy response to a changing environment.
These figures – issuance, validator set, etc. – are params that can be adjusted by governance and should be as circumstances evolve. These are, IMHO, healthy recommendations that help the entire community.
Your arguments are strong, but from emotional point of view if everything goes down: (price, number of validators, inflation) it is bad.
As You know many well know validators such as Informal systems, gumi, chorus.one, p2p.org, dora factory, RockawayX closed their nodes on Agoric’s chain. They left Agoric and i have a question why? They didn’t know about plans to reduce number of validators, they left by another reason. Do You guess what is the reason?
I totally get the emotional side of this – I hold tokens too, and it’s painful. But, we shouldn’t be too quick to label every “down” a negative. Classic example is reduced inflation, right? Less minting means fewer tokens to depress the price, which is a good thing.
As for the validators who have left, a couple of comments: (1) I can’t speak for all of them; (2) validator turnover happens on all networks (not saying what we’ve seen is normal, just saying turnover is a fact of life); (3) and I would guess (emphasis on guess) that at least a few left because it wasn’t profitable to run the validator any more – largely because of the decrease in token price.
That latter issue is one of the reasons why reducing the validator set is good business – it enables the remaining validators to earn more.
Fully agree, that in current situation Agoric needs cut number of validators and inflation rate, but the question is why such situation takes place?
And the answer is: evil of all evils inpatient Venture Capital, who sell their tokens and dump the tokens price, reducing security of the network, interest from retail investors and network validators.
On my opinion, they have to wait till network usage appears and hits at least 1 million transactions per day and than sell their tokens, but not like they do and did during current and last year.
@Ciberexplosion We’re now well past the standard 10 day discussion period that would normally precede a formal proposal. Are you planning to submit a proposal along the lines of this discussion, or would you prefer us (DCF) to do it – or is there someone else on this thread who might want to take the lead? We’re happy to support as needed, so let us know your thoughts – and thanks again for initiating this discussion for the community.
NP – given the relatively short voting period for Agoric, we typically advise people to post early in the work week (rather than on the weekends). Proposals posted during the weekends sometimes struggle harder to make quorum as fewer people are paying attention to such things, etc.
@Ciberexplosion FYI – just poked the validator community in the Agoric Discord in hopes of raising awareness of this topic. We might want to see what they have to say before moving more. Also, I note that the person at Agoric who leads Validator Relations left on holiday today and is not back in action until 2 Sep. Given that, we might want to drag our feet a bit on posting the proposal.
Will this adjustment run in parallel with a new distribution of the foundation delegation?
Self bond of many active validators is null or low, with often also low governance participation.
Shouldn’t the cut happen after a redistribution of the foundation delegation? It was mentioned that this would have happened multiple times but still waiting. A lot of delegation is also lost with jailed validators.
