All dominant blockchains are gradually gravitating towards centralization. Cardano's decentralization could be better. Let's identify the reasons for declining decentralization.
The current state of decentralization is far from ideal
Historically, so far all dominant blockchain networks have gradually gravitated towards centralization over time. Let's briefly recap the current state of decentralization of the major blockchains. We will then discuss the reasons that led to centralization.
In the Bitcoin network, two dominant pools, Foundry USA and AntPool, produce more than 50% of the blocks. A large number of miners deliberately choose between these two pools and ignore the other smaller pools.
Based on the 2021 study, we know that centralization is also occurring at the level of the miners. The top 10 miners control 90% of the hash rate. Only 0.1% of miners (about 50 entities) control almost 50% hash rate. The is relatively old, so the situation may be a bit worse now. The trend reversal very likely did not happen as the team did not change the conditions.
Ethereum is very similar when it comes to decentralization. After the transition to Proof-of-Stake, centralization is occurring through Lido and the centralized exchanges Coinbase and Binance. Let's add that for Ethereum, the critical threshold where a single entity failure can cause block finalization problems is slightly higher than 33.3%. Lido is very close to this threshold.
The total staked is 23.8 ETH. There are 7.6M ETH staked on Lido. Lido runs 238K validators. The number of unique depositors is 158,000. This means that the average deposit is 48.1 ETH. This suggests that Lido is used by whales which could run several validators on their own.
The top 10 depositors collectively deposited 1.56M ETH. They have a combined 6.5% share in the staking. The largest whale deposited 330K ETH into Lido.
Why is there a gradual centralization?
There are multiple reasons why blockchains are gradually gravitating towards centralization.
First of all, it is important to understand that a significant position in a blockchain network can always be bought with money. There is essentially no effective defense against this. Teams can set economic incentives of protocols (reward mechanisms) and people are the ones who decide how to take advantage of this opportunity. People are limited by the amount of their capital. The wealthier ones logically have the opportunity to gain a stronger position.
The type of consensus, i.e., Proof-of-Work or Proof-of-Stake, is irrelevant to the possibility of gaining dominance in the network. In both cases, the mechanisms are similar. Once it is economically viable to participate in a network consensus and receive rewards from the protocol, it is attractive to all people regardless of the size of their capital.
Let's ask some questions that will give us more insight into the problems blockchains face.
Why do Bitcoin miners most often choose 2 dominant pools and do not delegate the hash rate to smaller pools, which would lead to a higher decentralization of block production?
Economic incentives most significantly determine the behavior of actors. Bitcoin pool operators want to maximize their profit as much as miners. So who behaves irresponsibly in the context of decentralization? Pool operators or miners?
We get the answer if we look at pool fees. We find that the two largest pools have 0 fees.
Foundry USA pool has 0 fees and specializes in providing quality services to miners with large hash rate. So it is very likely that a large portion of the 50 miners (who control 50% of the hash rate) delegate to Foundry USA.
It is not possible to verify that the fees are actually 0, as miners must request permission to use the pool via a web form. The mining process is somewhat non-transparent and shady deals can occur.
As we said, economic incentives are the most influential factor in the decision. It is logical that miners want to maximize their profit even at the cost of sacrificing decentralization.
Unfortunately, we don't know what the Foundry USA pool, which is operated by a subsidiary of Digital Currency Group (DCG), benefits from. Either they are pursuing non-economic goals (which is unlikely) or they are trying to gain a dominant position in the network in order to destroy the competition.
We can conclude that the pool's zero fees, the pool's specialization in providing quality services to large miners, and non-transparency are leading to the centralization of Bitcoin.
Let's ask another question. How easy and affordable is mining?
At this point, we can only repeat what has been said many times. The initial cost of engaging in mining is high in the order of several thousand USD for the acquisition of ASIC hardware and power costs. Mining is more profitable in regions where electricity is cheap. Large miners have a distinct advantage as they can more easily negotiate discounts with hardware and power providers.
Big miners can expand their business faster and drive smaller miners out of business.
Mining is somewhat exclusive and the big players have a big advantage.
Are the reasons for centralization similar in the case of Ethereum? Let's explore.
If you look at fees and APRs, you will find that the services where ETH is most concentrated offer the most favorable terms. Specifically, low fees and high APR.
Lido has a fee of 10% and an APR of 4.4%. Coinbase has a fee of 25% and an APR of 3.77%. Binance has the lowest fee, only 5%, but also the lowest APR of the three most used solutions, namely 3.6%. Kraken offers the highest APR, namely 5.5% APR. Kraken has the 6th largest position in the staking. The lowest fee of 0.45% and high ARP of 4.8% is offered by Frax, which has 13 positions in the staking.
It can be said that economic incentives are also decisive in the case of Ethereum. Many ETH are staked from centralized exchanges. The consideration of whether to entrust ETH to a centralized service (exchanges) or rather to a more decentralized solution like Lido plays a role in the decision of stakers.
Users are certainly weighing the risks associated with using centralized services against the technological complexity of stacking. It is possible that many stakers choose centralized services simply because it is easy and less risky from their perspective. People may find self-custody and staking ETH from their own wallets to be a complex and risky process.
Vitalik Buterin said publicly last week that he only stakes a small amount of ETH because he doesn't think the process is secure enough. We know that less than 160K stakers use Lido. Many ETHs are staked from centralized exchanges. We can assume that people trust the exchanges more than themselves so that they can stake ETH securely.
In terms of input costs, staking will always be better than mining, as only small amounts of coins can be staked. For Cardano, there is no minimum amount of ADA coins required for entering staking. Ethereum requires 32 ETH to run the validator, but third parties will allow participation in staking with smaller amounts of ETH as they can pool ETH. If a user does not have 32 ETH and wants to participate in staking, they must entrust the coins to a third party.
Other staking properties that are defined in the protocol are also important. In addition to the required minimum for staking, i.e. 32 ETH, the ability to spend coins at any time also influences people's decisions. Ethereum locks coins because it needs to punish validators who don't behave as the protocol requires.
If people want to use the value of ETH in DeFi services or be able to spend it at any time, they must use a third party to mint tokens (LSP tokens) instead of ETH. These tokens can be spent, traded on an exchange, or used in DeFi services.
Buying coins is not independent of region, making staking more inclusive and open to everyone across the globe. As for the availability of staking, it is significantly easier and more affordable than mining.
In the Ethereum network, centralization is happening for economic incentives, very similar reasons to Bitcoin. People are looking for the most economically viable solution. Another reason may be the fear of self-custody and self-staking (technological complexity). Also, people's need to be able to spend ETH (or LSP tokens) or use them in DeFi at any time may have an impact. In a word, this can be described as liquidity.
In the case of Bitcoin, coin locking does not occur because BTC serves as a reward, not as an expensive resource around which the protocol is decentralized.
Let's summarize all the reasons that lead to centralization. The list includes reasons we haven't talked about yet. We will talk about them in more detail later.
- Economic incentives (fees, profit)
- Input costs to enter the staking/mining
- Size advantage (the economy of scale)
- Low transparency
- Specific protocol features (locking coins, slashing, etc.)
- The technological complexity of staking/mining
- Liquidity of coins
- Misunderstanding of protocol and incentives
- Inability to change protocol and economic incentives
Economic incentives are the most significant factor influencing decentralization. Without further analysis, it is difficult to say what factor could be put in second place. However, it can be concluded that the combination of all other factors influences the resulting quality of blockchain decentralization.
People must not be put in a position where they are forced to choose between decentralization and economic incentives, because their personal selfish interests will almost always come first. It is therefore important that the team sets the economic incentives of the protocol in a way that encourages decentralization.
In general, the quality of blockchain decentralization is influenced by the properties of the protocol and economic incentives, as this determines people's behavior. Teams can estimate and model human behavior when designing a protocol, but they can never be sure if the theory will work in practice, especially in the long run. It is therefore beneficial if the team (together with the community) can change some settings or mechanisms to increase decentralization. Stagnation or resignation to making changes can result in the centralization of the protocol.
Decentralization of Cardano could be better
We consider Cardano to be the most decentralized blockchain in the top 10. Still, the situation is not ideal. Let's look at this in more detail.
When compared to Bitcoin and Ethereum, at first glance Cardano is more decentralized, as no entity has more than a 10% share in the staking. The largest multi-pool operator (MPO) has a 7.2% stake. The centralized exchange Binance has only 4.2% (it used to be over 10%). All other entities have a stake of less than 4% in staking. The top 10 MPOs have a combined 29% share of the staking.
In the Cardano ecosystem, roughly 1.3M people delegate ADA to pools. It's hard to find out how many people stake ADA through exchanges and how many coins are held on them. In the case of Binance, we know this very well as the exchange runs its own pools.
Based on analyses of the number of ADA holders worldwide, we can assume that there may be several tens of millions of them. In that case, the same applies as in the case of Ethereum, i.e. most people stake ADA from centralized exchanges. It could be 90% of the people.
The Cardano community needs to think about how to motivate people to stake ADA from their own Cardano wallets.
MPOs have a significant role in the decentralization of Cardano. The protocol cannot distinguish single-pool operators from multi-pool operators in terms of rewards. The number of MPOs in the network can be influenced by delegators (their delegation decisions), but only if the MPO is economically dependent on delegated ADAs. A wealthy entity can operate multiple saturated pools essentially without restriction.
The existence of MPO is not necessarily a bad thing when the entity operating multiple pools is trustworthy and has a good track record. In an ideal world, each pool would be operated by a single operator. Unfortunately, we don't live in an ideal world. The amount of MPOs in the Cardano network is largely decided by the ADA holders, and they make their decisions based on a variety of information, events, or other motives.
Ideally, there should be only 500 fully saturated pools. In this case (and a few other circumstances), the staking rewards would be the highest. In a recent poll, it became clear that not all people understand how the Cardano protocol works. This can be inferred from the results and the debate on social media.
The option of increasing K to 1000 and reducing minPoolCost (fixed rewards) from 340 ADA to 170 ADA received the most votes.
From our subjective point of view, it is unnecessary to increase K to 1000, as currently, more than 1000 pools produce the blocks in every epoch. The current problem is not about the setting of the K parameter but about the low saturation of the pools.
Increasing the K parameter would lower the saturation point, so MPOs would have to operate new pools and ask stakers to delegate ADA to new pools from the old (oversaturated) pools. The number of pools operated by MPOs could have a more ostracizing effect. However, if the stakers have a reason for delegating to the MPO, it wouldn't change anything.
It is also interesting that people voted to reduce the fixed reward to 170 ADA. Many community members believe that minPoolCost should be removed. In the case of Bitcoin and Ethereum, we have seen that people are economically motivated to delegate hash rate or stake coins to where the fees are lowest. We, therefore, think that reducing the fixed reward or even removing it entirely, is not a good idea if we want to maintain a high level of decentralization.
It is not possible to run a pool for free. The time and financial costs are always relatively high if the pool is operated safely and reliably. In our view, it is not a good idea to introduce the possibility of zero-fee pools in the Cardano ecosystem. It would very likely lead to the creation of large MPOs, i.e. centralization.
We would have to ask how it is possible to run pools with 0 fees. This would probably lead to an overall reduction in the quality of block production as other operators would also have to reduce fees, which could be at the cost of reducing the quality of infrastructure.
Let's add that there are many Cardano pools that have a margin set to 1% and yet are not saturated. On the other hand, it is possible to find pools that have a margin of 3% or more and are saturated. It seems that in the Cardano ecosystem stakers are not yet looking for the most economically viable solution, which may be surprising. We don't have an explanation for this, but it could be that pool operators are motivated to be active or visible. It is likely that they have found loyal stakers because of this.
Cardano is well positioned to maintain a high degree of decentralization because it has liquid staking. The economic incentives are set reasonably. Input costs are low. Large players receive proportionally the same rewards as smaller ones. There is no need to use third parties just because you don't have enough ADA coins. You can literally stake a few coins. Staking is very simple and secure. However, it is possible that stakers may find the choice of pool difficult as they may not be familiar with the details they see of the various explorers.
From our perspective, Cardano could be more decentralized if stakers understood the details of staking and incentive mechanisms. Education could also partially address the existence of MPOs. Certainly don't think it's an easy task to educate new stakers. This can be just as difficult a task, perhaps even more so than changing the protocol settings. A big problem (bigger than people admit) can be the fear of self-custody, i.e. also self-staking.
Cardano may get into a situation where it will gravitate towards centralization like its predecessors, as the current mechanisms may not be set up well enough. In theory, a pool operator may emerge that offers very favorable staking terms or has good marketing. It can be a centralized entity, such as a crypto exchange.
If the network is to maintain a high degree of decentralization, it is essential that the community is able to respond to problems as they arise. We believe that decentralized governance is the best way to get there, as there can be nothing better than the wisdom of crowds. But even the masses can be mistaken or manipulated. Any changes to the protocol should be preceded by debate, supported by research and mathematical models. We should look at other major projects and be able to name the reasons why decentralization has begun to degrade for them.
Every blockchain, including Cardano, is dependent on the rational behavior of people who hold an expensive resource that is used to decentralize the network. People's behavior cannot be predicted in advance and conditions change over time. It is almost inevitable that networks will gravitate towards centralization, as humanity is used to living in a hierarchical structure. We are used to relying on leaders. We are accustomed to trusting third parties that appear to be reliable. Our habits are not in line with decentralization’s needs. This is the reality we have to face. If blockchain networks become more prominent, people with big capital will want to gain control or profit from their existence. It will be very difficult to maintain a high level of decentralization and historically we have failed to do so. Cardano is one of our next attempts and we definitely have not won. We don't know how best to sustain decentralization in the long term. Hopefully, on-chain governance is the right direction.