[IDEA] Improve the Nakamoto Coefficient

1. Context The Nakamoto coefficient is a key indicator of the Hub’s decentralization. Unfortunately, it has been trending downward, which creates risks both for consensus security and for the resilience of the network against capture by a small number of validators. Today, some validators particularly centralized exchanges and a few large incumbents concentrate a disproportionate share of the voting power, mechanically reducing the Nakamoto coefficient. The community broadly agrees that we need to rebalance distribution, while still preserving: • the attractiveness of staking for delegators, • the economic sustainability of validators, • the simplicity of the governance and incentive model. Capture d’écran 2025-09-21 143654 2050×1056 283 KB cosmos-datanakamoto-index-2025-09-21 (4) 1116×270 70 KB 2. Problem Statement Without safeguards, large validators will continue to accumulate stake. Rational delegators usually favor the most visible or well-known validators, or those offering slightly higher yields, regardless of decentralization. Strict voting power caps are not seen as a solution, since they can be bypassed (for example, through duplicate…

Excerpt (1192 of 6854 characters). Read the whole post on the forum ↗

Seems like a legit discussion, I have to little knowledge to give any meaninfull input but the idea of limiting “whale validators” and giving smaller validators a better chance and a small “bonus” sounds good!

Hi,
Your proposal is missing some numbers to showcase the change

Nakamoto Bonus Impact

With the data from Mintscan, a 10% Nakamoto pool of the inflation :

• Pool total : 4.695M ATOM per years
• Bonus per validator : 26,086 ATOM
• APR for small validators : 17.22% → 37.61% (improve by +118%)
• **APR for big validators (top 66%) ** : 17.22% → 9.23% (reduce by -46.5%)

Self-Stake Requirement

With the rule of 1/250 self-stake :

• top validator self-stake requirement : 153,800 ATOM
• Total self-stake : 1.07M ATOM (0.23% du supply)
• Sybil cost (33% VP) : 35.2 ATOM at risk (0.01% slash) **1,763 ATOM at risk (0.5% slash)

Thank you a lot, I already added it into my 6th section.

This is fully possible in theory, because Cosmos Hub uses a flexible staking and reward distribution module that can be adjusted via inflation parameters, rewards, and self-bond requirements. The core idea — Nakamoto Bonus + 1/250 self-stake rule — is already solid, but it can be improved and made more precise without significantly changing the structure: Adjustments to the Nakamoto Bonus • Tiered distribution instead of fixed: Instead of distributing 10% equally to all validators, give more to validators with lower VP and less to medium/large ones. This still prevents the snowball effect, but rewards small validators who contribute to consensus. • Dynamic pool based on performance: Only validators with uptime ≥ 99% and no slashing receive the bonus. This reinforces responsibility and reliability. Adjustments to the self-stake rule • Instead of a fixed 1/250 for everyone: Make it progressive: large validators require a slightly higher self-stake (e.g., 1/200), while small validators keep 1/250. This prevents large actors from creating multiple ghost validators with minimal risk. Delegator considerations • Clearly inform delegators that choosing…

Excerpt (1198 of 2203 characters). Read the whole post on the forum ↗

I really appreciate your input and the spirit of trying to refine the idea. However, I think there is a fundamental concern of fairness and equality in what you are suggesting. With the current design, an equal distribution of the Nakamoto Bonus already acts as a redistributive mechanism. Every validator receives the same amount, but in relative terms this has a much stronger effect on smaller validators than on larger ones. In other words, equality at the rule level already creates redistribution in practice. Introducing tiers that give more to the smallest and less to the medium or large validators risks distorting this natural effect. It could also create perverse incentives, for example encouraging validators to deliberately remain “small” in order to capture a higher share of the bonus. Another important point is simplicity. The Cosmos Hub community generally prefers mechanisms that are easy to understand, audit, and implement. A flat equal distribution is transparent: everyone knows that a fixed percentage of inflation is redirected to a pool and shared equally among active validators. Adding layers such as tiered distribution, progressive self-stake ratios, or…

Excerpt (1189 of 2631 characters). Read the whole post on the forum ↗

First, thank you for starting this discussion, which could in fact be described as a reform of the staking module one that is closely tied to a potential governance reform I still need to address here . My first point is this: everything here comes down to compromise, deliberate choices, and political will. The current staking module does have the shortcomings you outline, but it also provides a major advantage: it requires no political intervention and self-regulates, with no harmful risk of validator duplication splitting their stake to chase extra yield. Let’s be clear: diverting part of the inflation for such a political reform carries obvious risks. While I fully support 1/250 self stake , the Nakamoto bonus could, in my view, only be implemented in two scenarios: • In small doses (certainly not 10 % of total inflation). Otherwise, the APR gap between the top and bottom of the validator set would directly encourage validator duplication and drive out less professional operators, who would be undercut by more sophisticated arbitrageurs. Self-stake alone wouldn’t restrain large capital holders. • By redefining validator access. One option would be to freeze the…

Excerpt (1193 of 2170 characters). Read the whole post on the forum ↗

I develop with the Cosmos SDK, CometBFT, and my own wallet, and I have the ability to create plug-and-play modules. I can program, but my question is: how can it actually be possible to achieve the changes you mention? I understand your points: I will prioritize simplicity and neutrality of the mechanism. We can strengthen delegator education and transparency instead of adding complexity to the distribution."

Thank you for your thoughtful response. I think you raise important points, especially regarding the risks of APR gaps and validator duplication. On the Nakamoto Bonus, I agree that starting with 10% might be too aggressive. It could create too strong an APR disparity between small and large validators, which might lead to unintended behaviors such as validator splitting. I am open to adjusting this parameter and think a smaller initial allocation, for example around 5% of inflation, would be a more balanced starting point. We could then monitor the effects on APR distribution, validator sustainability, and the Nakamoto coefficient before considering any further adjustments. On the self-stake requirement, I am glad we are aligned. It is a simple, fair, and effective mechanism to ensure skin in the game and to make Sybil attacks more costly. Where I would respectfully disagree is the idea of ending permissionless validator entry. I believe this would undermine one of the Hub’s core principles: openness. If governance had to approve every new validator, it would introduce politicization, lobbying, and the risk of validator cartels. Validators would no longer simply be economic…

Excerpt (1199 of 1828 characters). Read the whole post on the forum ↗

I think that closing entry to new validators and maintaining a list of professional validators however un permissionless that may seem would help legitimize the creation of a Nakamoto bonus or any other reform aimed at improving the Hub’s decentralization.

You can check the response I shared here, where I explain how this approach could greatly enhance decentralization while turning validators into administrative agents of the Hub, validated by governance. Of course, this remains just one point of view and could face certain limitations.

Introducing a self-stake requirement is essential to strengthen the Hub’s security and ensure that validators truly have skin in the game. Without it, large actors can operate with minimal risk, while smaller validators carry proportionally more responsibility. However, enforcing this rule strictly from day one through slashing and jailing would create major disruption: several large validators are currently far from the target and would either need to exit suddenly or find significant capital on short notice. This could destabilize the active set and negatively impact delegators. To address this, the self-stake rule should not be implemented overnight, but rather through a phased approach. A gradual timeline gives validators enough time to adapt, while steadily reinforcing the credibility of the requirement. The goal is to reach full enforcement in a way that is both fair and predictable, maintaining network stability throughout the transition. Self-Stake Enforcement: 24-Month Progressive Implementation The self-stake requirement is essential to ensure validators have skin in the game. However, applying it instantly with slash and jail would destabilize the active set,…

Excerpt (1194 of 4106 characters). Read the whole post on the forum ↗

UPDATE Proposal: Improve the Nakamoto Coefficient through a 5% Bonus and Progressive Self-Stake 1. Context The Nakamoto coefficient is a key indicator of the Hub’s decentralization. Unfortunately, it has been declining, which increases the risks for consensus security and reduces the network’s resilience against capture by a small number of validators. Today, some validators, notably centralized exchanges and a few large incumbents, concentrate a disproportionate share of the voting power, mechanically reducing the Nakamoto coefficient. The community agrees on the need to rebalance this distribution, while still preserving: • the attractiveness of staking for delegators, • the economic sustainability of validators, • the simplicity of the governance and incentive model. Capture d’écran 2025-09-21 143654 2050×1056 283 KB cosmos-datanakamoto-index-2025-09-21 (4) 1116×270 70 KB 2. Problem Statement Without safeguards, large validators will continue to accumulate ever more stake. Delegators, acting rationally, often favor the most visible or best-known validators, or those offering slightly higher yields, without considering decentralization.…

Excerpt (1197 of 9764 characters). Read the whole post on the forum ↗

I just want to say I agree with this proposal and I think it hits the right balance. The way things are going right now, the biggest validators keep pulling in more and more stake, and it makes it harder for smaller ones to survive. That’s not healthy for the Hub long term. What I like about this idea is that it doesn’t try to punish anyone, it just gives smaller validators a fairer shot while still keeping larger ones profitable. The progressive self-stake requirement also makes sense to me, because it means everyone has real skin in the game instead of just spinning up multiple validators. That said, I’m expecting pushback. Some of the larger validators will probably either vote “no,” abstain, or try to drag things out to avoid quorum. That’s their interest, to keep the current system as it is. But delegators need to realize they don’t have to just sit back. Every one of us who stakes ATOM has the ability to vote, and if we want real decentralization, we can’t just leave it up to the top validators to decide for us. As a delegator, I care about yield, but I care just as much about security and resilience. If this goes through, it makes me more confident the Hub won’t end…

Excerpt (1197 of 1275 characters). Read the whole post on the forum ↗

Thank you for sharing this idea. We agree that improving the Nakamoto coefficient is an important discussion, but we believe that many in the ecosystem tend to misunderstand the distinction between decentralization in governance and decentralization in execution . Both are critical, but their implications differ — and in our view, one carries slightly more weight than the other. Governance vs. Execution Take the example of a chain with 100 validators, each holding 1% of voting power. In a Cosmos BFT consensus, the Nakamoto coefficient would be 33, meaning it would take 34 colluding validators to halt block production. Importantly, such an attack could only stop the chain’s operation — it could not censor individual transactions or unilaterally alter the state without reaching 66% consensus power. Now, imagine in this same chain that a single entity owns 50% of the token supply. Even if those tokens are distributed across many validators, this actor would control governance outright. In this scenario, the chain would look decentralized from a consensus standpoint but would be entirely centralized from a governance perspective. That entity could unilaterally push through…

Excerpt (1193 of 2540 characters). Read the whole post on the forum ↗

Yes, I agree with you on this point. That’s why the proposal you made in your other post is important, and I will vote in favor of it. I also believe that combining the reform of governance decentralization with the reform of execution makes sense as part of a broader, comprehensive reform.

Indeed, there is certainly a conflict of interest on the part of the validators. Their reputation and their alignment with the network’s overall health depend on not focusing solely on their own revenues. I believe they need to make an effort, and this combination of mechanisms aims not to hit them too hard. The goal is to gradually move toward a more decentralized and fairer system, while giving smaller validators a boost and preserving larger ones as much as possible.

This would not only make smaller validators more sustainable over time, but also incentivize delegations toward them thanks to increased APR visibility in the lower part of the set.

We fully agree on the importance of improving execution decentralization, and the Nakamoto coefficient provides a useful line of defense in this regard. However, it faces the same limitation we encounter with governance vote-power caps: validator duplication as a bypass mechanism.

This is why we believe governance must be addressed first. Additionally, we need to clarify that mitigating duplication requires a clear governance constitution that explicitly defines such infringements and establishes the associated penalties. This is another crucial intermediate step that needs to be cleared before proceeding with a solution like the Nakamoto Bonus or other methods that could be evaded through validator duplication or any similar tactics.

First off, thank you for taking the time to lay this out so clearly. I do agree with your distinction between execution decentralization and governance decentralization, both of these matter, but governance carries the heavier weight because of the power it has over protocol rules. You’re absolutely right that delegator passivity and concentrated influence are a fundamental risk that needs to be addressed. At the same time, I see the Nakamoto bonus and progressive self-stake as tools that can work alongside governance reforms rather than only after them. The snowball effect we’re seeing with the largest validators doesn’t just threaten execution decentralization; it also reinforces governance concentration by channeling ever more delegator voting power to the top. Breaking that cycle, even if it’s done modestly, that alone helps to reduce governance centralization risk at the same time. I also want to emphasize that my support here comes from the same place as yours: wanting what’s best for the Hub’s long-term health and credibility. I don’t see this proposal as the final word, but I do see it as an important piece of the bigger puzzle. IF we can combine steps like this with…

Excerpt (1199 of 1350 characters). Read the whole post on the forum ↗

Precisely, what’s being discussed doesn’t punish large validators outright, but it CREATES a space for smaller ones to breathe and be sustainable, which in turn strengthens the Hub as a whole.

I also like your point about APR visibility in the lower part of the set, that’s an underrated benefit. If delegators can see that supporting smaller validators isn’t just “good for decentralization” but also good for their own yield, that could shift behavior in a meaningful way.

The real test, though, is whether delegators step up. Validators will naturally have conflicts of interest, but at the end of the day delegators are the ones with the ability to vote and redirect stake. If they use that power, this proposal could actually deliver the healthier system you’re describing.