In February, Coinbase announced its plan to launch its own native blockchain comprised of a layer 2 on Ethereum. On March 23, another and probably the most prominent layer 2 called Arbitrum launched its own token, setting an all-time high of most transactions on any layer 2 the same day. The next day, zkSync launched its highly anticipated Ethereum Virtual Machine (EVM) compatible layer 2 based on zero-knowledge (zk) proof, followed up by a similar launch by Polygon on March 27. The last few months have without question been the most productive months for layer 2 technology. Although they offer superior scalability and enhanced competitiveness to other cryptocurrencies, they are not without consequences, but let us first put the record straight on layer 2s.
The Blockchain Trilemma: What are Layer 2s?
Since the genesis block of Bitcoin in January 2009, every permissionless blockchain has been haunted by the blockchain trilemma. The latter concerns that decentralized networks cannot offer decentralization, security, and scalability as one, but solely two of those factors, leaving one in the dust. The two largest blockchains Bitcoin and Ethereum have for years been severely challenged by this matter. From the start, both opted to sacrifice scalability to not compromise decentralization and security. In consequence, Bitcoin can merely handle about 5 transactions per second, whereas Ethereum can handle around 15 transactions per second. This is not anyhow sufficient for a medium of exchange or a global settlement layer, so the lack of scalability has arguably been the greatest hindrance to both cryptocurrencies. As opposed to Bitcoin and Ethereum, newly launched blockchains such as Solana and Avalanche have sacrificed decentralization to massively boost scalability. However, as decentralization is the main point of crypto, it is not a favorable outcome either.
To resolve the trilemma, a few years ago, the Ethereum community opted to scale through rollups, also known as layer 2s. The fundamental principle of layer 2 is rather simple. Instead of executing a single transaction at a time on Ethereum or layer 1 by another term, you bundle a bunch of transactions from layer 2 to execute them in a batch off-chain before posting the transaction data as a single transaction to Ethereum. Although the final transaction is larger and more expensive than a regular layer 1 transaction, it is technically hundreds of transactions from a rollup, altogether reducing transaction fees by between 5 to 20 times.
There are two main approaches to ensure the validity of transactions from rollups, namely optimistic – and zero-knowledge (zk) rollups. The former assumes that transactions are valid by default, yet transactions have a challenging period of often 7 days during which anyone may prove the invalidity of the transaction. This is contrary to zk rollups that automatically compute cryptographic proofs, so that it does not need a challenging period. There are multiple advantages and trade-offs with each approach, however, a thorough deep dive is out of the scope of this piece.
It is important to state, though, that zk rollups arguably are superior, all things considered. In early-2021, co-creator of Ethereum Vitalik Buterin wrote: “…in the medium to long term ZK rollups will win out in all use cases as ZK-SNARK technology improves”. However, zk rollups are technically more advanced than optimistic rollups, so it has been a severe challenge to make them equal to Ethereum’s computation engine for layer 1, also known as Ethereum Virtual Machine (EVM). This has forced developers to rewrite applications considerably before deploying to a zk rollup, causing them to achieve less traction than optimistic rollups such as Arbitrum and Optimistic, as a result of fewer decentralized applications on zk rollups. This is why the recently launched zkEVM rollups by zkSync and Polygon, respectively, are of great importance to the ecosystem. The transaction costs of these rollups are rather high and few applications are yet deployed on them, but it is a step in the right direction.
Offering scalability at the expense of liquidity and decentralization
As specified, rollups primarily offer greatly more scalability, allowing a greater transaction throughput alongside lower transaction fees. The greatest part of transaction fees on layer 2s is a result of posting the transaction data of the layer 2 batch to Ethereum as so-called calldata. The latter is expensive, as it is permanently stored on the blockchain. This is somewhat unnecessary because rollups only need this data for a short period to assure validity. To reduce this cost significantly, the Ethereum community is working on an update known as Proto-Danksharding or EIP-4844. The update is set to introduce data blobs, in which rollups may store nearly all data. The data blobs are deleted 1 to 3 months after creation, so these data blobs are much cheaper than regular calldata. This allows the cost of rollups to be greatly reduced later this year when EIP-4844 is planned for release.
Moreover, an often-overlooked advantage of layer 2s is that they allow for tailored execution layers for certain applications and use-cases, instead of a single execution environment on layer 1 that may simply be less than optimal for every use-case such as gaming, which, for instance, may need faster transaction finality.
The Ethereum community has somewhat put all its eggs in one basket, since its roadmap to greater scalability in the foreseeable future is nearly almost about rollups, but that is not without severe trade-offs. Since its genesis block, Ethereum has strongly benefited from its first-mover advantage, allowing it to achieve a solid network effect well before there were any alternatives. This network effect is what still makes Ethereum the absolute biggest cryptocurrency for decentralized applications, but layer 2s does not fully appreciate this. Although they post their transaction data on Ethereum, they largely operate as sovereign blockchains, so users cannot use identical applications across rollups. This fragments liquidity in decentralized applications and erodes the network effect of Ethereum. The rollups are likewise a UX nightmare for users since they need to bridge tokens from one chain to another, which often entails heavy transaction fees and a challenging period. The community looks to solve this, yet a thorough fix should not be expected in the foreseeable future.
Next, decentralization is another concern with rollups. By and large, rollups inherit the security of Ethereum by posting the batch of transaction data thereon, however, the batching of transactions off-chain is largely not decentralized, simply as the computing to perform the batch known as the sequencer has until now been fully centralized by the teams of the individual layer 2s. Since it is centralized, the teams can theoretically front-run and censor certain transactions, causing them to not reach layer 1 at all. This is very much against the philosophy of crypto as a neutral settlement layer without censorship. Every rollup is public about the aim to decentralize their sequencer over time, but nobody knows exactly what degree of decentralization they may achieve.
To make matters worse, as it is still experimental technology, some rollups may be instantly upgraded, causing the rules to be changed from one second to another. This may in the worst case lead to the loss of funds. It appears that most rollups are trustworthy, but the main point of crypto is to not have to put trust in anyone but code and rollups are simply not there yet. They may reach that point eventually, but other more scalable layer 1s are slowly catching up on Ethereum.