Full decentralization of Ethereum layer 2s will happen in a few years, multiple teams told Cointelegraph in June.
The goal of decentralization has been held back by a variety of factors, including a desire to implement user experience and security improvements. However, progress is now being made toward decentralizing these networks, and “Stage 2” will be reached soon, the teams claimed.
The concept of “Stage 2 decentralization” refers to the final stage of Ethereum founder Vitalik Buterin’s list of layer-2 decentralization milestones. It is also called “full decentralization,” as it would theoretically make censorship nearly impossible to perform on the networks.
Ethereum layer 2s are networks that try to make Ethereum scale by compressing and validating its transactions on separate networks.
Decentralization is coming soon, some networks say
Linea founder Nicolas Liochon told Cointelegraph that “within the next few years,” users should expect most L2s to have reached Stage 2 of decentralization.
“Rollups prioritized other important upgrades which directly impact the user experience and security,” Liochon stated. “For example, the Dencun upgrade has lowered costs by an order of magnitude for users.”
But now that these upgrades have been implemented, progress toward reaching Stage 2 “has started.” According to Liochon, the team is planning to roll out a testnet with new decentralization features “in the coming months.”
Anthony Rose, chief technology officer of zkSync developer Matter Labs, expects the dominant L2s to reach Stage 2 “over the next two to thre years.”
Rose expects some consolidation in the space. Currently, more than 20 L2s have $100 million or more in total value locked (TVL). But in the future, “we will see a relatively small number of very large networks.” These networks will be made up of multiple individual chains, the largest of which will be completely decentralized.
“The performance of these systems can take advantage of certain things,” Rose stated. “So I think we’ll see the big ones that exist today being stage two [...] disposing of their training wheels over the next two to three years.”
Optimism co-founder Karl Floersch stated that the move to Stage 2 will be “swift and decisive.” In his view, reaching Stage 2 is difficult because it requires multiple proof implementations and because these implementations are “the most security-critical code and [require] unbelievable levels of assurance.” Even so, he expects his own network to reach this milestone soon.
Floersch also claimed that once a single network reaches Stage 2, others will follow quickly. This is because of the open-source nature of blockchain development:
“Once we get to Stage 2, a majority of L2s will get there too. That’s part of the beauty of building out open source ecosystems: the value driven by contributions can be shared by all the participants and projects who are using the standards.”
Steven Goldfeder, a co-founder of Arbitrum developer Offchain Labs, stated that his team’s network is “very, very close” to reaching Stage 2. Currently, Arbitrum’s fraud proofs can only be overridden by a nine out of 12 vote from its security council. ARB tokenholders elect members of this council.
Related: Optimism implements fault proofs, reaching Stage 1 decentralization
“I think two of them are from our team,” Goldfeder clarified. “So [...] our team doesn’t even have the veto power.”
To reach Stage 2, Arbitrum would need to remove the power of the security council to act except in those cases when there are “provable bugs.” This is “more of a question of community readiness than anything else,” Goldfeder stated. ARB tokenholders would need to vote for this change, and they haven’t done so at this point.
According to Goldfeder, the Arbitrum Security Council has never acted to override fraud proofs despite having the power to do so. Even so, the Arbitrum decentralized autonomous organization can vote to remove this emergency power if it chooses to do so, which would be “the last hurdle” to make the network Stage 2.
Vince Yang, CEO of layer-3 network zkLink, also estimated that full decentralization "won't take that long" and "would be less than two years." For Yang, much of the difficulty so far has stemmed from a lack of interest among users. “L2/L3s already inherit the security from Ethereum, making the industry’s actual demand for further decentralization relatively low,” he stated.
In addition, L2 decentralized autonomous organizations haven't necessarily wanted to give up their monopoly on revenue, so it’s been hard to convince them to share this revenue with independent actors. “There is an existing financial incentive for L2/L3 networks to maximize revenue on transactions, further discouraging attitude and behavior toward complete decentralization,” he claimed.
According to Yang, zkLink uses the layer-2 Linea as a settlement layer. It is a “validium” rather than a full rollup, as it stores compressed transaction data with a data availability committee (DAC) instead of directly on the settlement layer.
He claimed that zkLink is "actively seeking the industry's best practices" and "will and are ready to fully decentralize, awaiting the footsteps and decisions of the top Layer 2s."
Decentralization is a long way off, other networks claim
While the above networks were optimistic that full decentralization would be achieved soon, others claimed that this lofty ideal is still a long way off from being realized.
According to Metis co-founder Kevin Liu, most L2s are difficult to decentralize because they rely on token-voting systems for governance. “Traditional token voting often results in concentrated control by large tokenholders,” Liu stated.
To solve this problem, protocols need to decentralize their sequencers, which will ensure “that the network is owned by all participating nodes and stakers from the economy layer.” Liu claimed that Metis employs this model of governance, as “each sequencer node earns mining rewards based on the blocks they produce, and all onchain activities benefit the nodes themselves.”
Even so, Liu claimed that L2s are far from achieving the goal of full decentralization. “From a technical standpoint, I don’t believe pure decentralization for layer-2 solutions is achievable in the near future, given the current state of the technology and the ongoing need for upgrades,” he stated.
Still, Liu claimed that teams should keep working toward the goal of decentralization, even though the road ahead will be long and difficult. He stated:
“Decentralization is not just one technical layer, which is still not mature; decentralization is multilayer, and we believe our approach is the most practical one and will drive the progress of technical upgrades as well.”
Related: Layer 2 networks must decentralize sequencers or face the consequences
The stages of decentralization
The concept of “stages” of decentralization comes from Buterin’s November 2022 post, “Proposed milestones for rollups taking off training wheels.” In the post, Buterin suggests that layer-2 rollups could launch with temporary “training wheels” in the form of centralized control as long as they work toward decentralizing over time.
Buterin identified three stages of decentralization. To be labeled a “Stage 0 rollup,” a network needs to provide software that can be used to independently compute the rollup’s state. However, the development team can still censor transactions or block withdrawals by posting false state roots to Ethereum.
To be labeled a “Stage 1 rollup,” the network needs to have implemented fraud or validity proofs in its Ethereum smart contracts. This means that the team’s sequencer can no longer publish false state roots, as these will be rejected as invalid.
However, Stage 1 networks are allowed to have a security council that can override valid transactions. Executing an override requires an explicit action by the council, and this can only be done with six out of eight votes from it. At least two members of the council must not be from the development team.
In a Stage 2 rollup, the security council cannot override the network’s fraud proofs unless these proofs contradict themselves or each other or unless the network has not processed a transaction for at least seven days.
Blockchain analytics platform L2Beat regularly evaluates L2 networks and classifies them based on Buterin’s criteria.
The lack of decentralization in most layer 2s led to controversy in June, as the Linea team successfully paused withdrawals and censored one user’s address. The censored user was an attacker who had exploited the Volocore decentralized exchange. While many users were happy that the attacker’s withdrawals had been blocked, some argued that the action was proof that layer 2s were taking too long to decentralize.
Buterin himself has argued that rollups should progress to Stage 1 by the end of 2024 or should not be considered rollups at all.
Magazine: ‘Bitcoin Layer 2s’ aren’t really L2s at all: Here’s why that matters
Update (June 27, 2024 1:18 pm): This article has been updated to incorporate an additional statement from zkLink CEO Vince Yang.