By Benjamin Simon

Let me make one point first. I recently participated in the latest funding round of Offchain Labs (Arbitrum development company), Mechanism Capital also participated. While it would be futile to pretend that our views in this article are objective, I hope this article will help readers understand some of the key differences between the two projects, biased as I may be.

All Rollup solutions follow a similar basic architecture and internal logic. Nonetheless, as we saw in the first part of this series, a single difference between Optimistic Rollups and ZK Rollups — how the respective “review processes” work — has implications for security, usability, and EVM compatibility. Many downstream differences.

first level title

early origin

First, some brief historical background on each project is in order. As it happens, both Arbitrum and Optimism have some unique origin stories.

Six and a half years ago, on a chilly morning in Princeton, a group of undergraduates working with Professor Ed Felten gave a talk on the project they had contracted to create: a blockchain-based arbitration system. With the goal of circumventing some of the expected scaling challenges of smart contract platforms, the plan is to design a blockchain that relies on a challenge and dispute resolution system to ease the computational workload of traditional miners. Dubbed "Arbitrum," the system would have suffered the same fate as most other promising academic computer science projects had not two ambitious doctoral students, Steven Goldfeder and Harry Kalodner, approached Felten a few years later. Build a robust Layer 2 solution from an initial concept. Shortly thereafter, Felten, Goldfeder, and Kalodner co-founded Offchain Labs and transformed Arbitrum from an abstract idea into a concrete reality.

first level title

Dispute Resolution: A Very Brief (Re)Introduction

first level title

A Preliminary Comparison of Arbitrum and Optimism in Dispute Resolution

The easiest way to describe the difference is that Optimism's dispute resolution relies more on the Ethereum Virtual Machine (EVM) than Arbitrum. When someone submits a challenge on Optimism, the entire transaction in question runs through the EVM. In contrast, Arbitrum uses an off-chain dispute resolution process to reduce disputes to a single step in a transaction. The protocol then sends this one-step assertion (rather than the entire transaction) to the EVM for final verification. Thus, conceptually, Optimism's dispute resolution process is much simpler than Arbitrum's.

image description

Source: OffChain Labs Development Center

Optimism's approach to dispute resolution—that is, running the entire transaction through the EVM—is not only conceptually simpler: it's also faster. There is no "multiple rounds" of back and forth as in Arbitrum's process. In fact, Optimism Rollups are often called "single round" for this reason, while Arbitrum Rollups are "multi rounds". In practice, this means that in the case of disputed transactions, the final confirmation on Ethereum is delayed longer in Arbitrum's case than in Optimism's case. As we explored in the first part of this series, the speed of dispute resolution matters because it determines how long it takes for users to swap tokens back to Ethereum from a Rollup.

first level title

rebuild compare

The fundamental trade-off between the two dispute resolution designs appears to be simply between speed and on-chain cost. But in fact, this is a bit too simple, because few people think that the controversy arises because of the following two reasons:

• Transaction processors on both Arbitrum and Optimism have no financial incentive to process fraudulent transactions. They are forced to put up collateral/bonds upfront, which gets slashed in case of fraudulent transactions.

• Parties monitoring the state of a Rollup are reluctant to submit false fraud proofs — in Optimism because the challenger has to pay the on-chain gas fee for a fraud proof, and in Arbitrum because the challenger has to provide its confiscated gas if the dispute fails security deposit.

So, if disputes are expected to be few and far between, why does the structure of the dispute resolution process matter?

While disputes are rare, Rollup must be designed to handle them when they can. Thus, the "controversial" case design affects the structure of the "non-controversial" case.

Since Optimism must be able to run every transaction through the EVM in the event of a dispute, it cannot process transactions that exceed Ethereum's gas limit because these transactions cannot be properly verified on-chain. Arbitrum, by contrast, can execute arbitrarily large transactions even if they exceed Ethereum's gas limit, because transactions are never batched through the EVM, but are first broken down into tiny "step assertions".

It's unclear how much of a practical limit Optimism's gas limits will place on applications. However, another, perhaps more important implication of the dispute resolution design difference is that Arbitrum can save gas by reducing the frequency of on-chain checkpoints (updating the "state root"). More specifically, Arbitrum can allocate a significant amount of off-chain computation to a single update, since that state root update could theoretically include (tiny) single-step fraud proofs for all transactions contained within it. Optimism, on the other hand, has to checkpoint on-chain after every transaction, significantly increasing its on-chain footprint.

first level title

Dispute Resolution and Potential Attack Vectors

A final point about these different dispute resolution processes is worth discussing: how resistant each design is to potential attacks. Above, we talked about economic incentives to thwart spam attacks. More specifically, neither Optimism nor Arbitrum validators are willing to submit unnecessary challenges.

But what about the case of a malicious attacker who doesn't mind incurring the economic cost of a spam rollup? In other words, what would happen if a person or entity was so committed to slowing down an Optimistic Rollup that they were willing to do so, even if it meant repeatedly paying for fake challenges?

As mentioned above, Optimism's dispute resolution process is simpler and faster than Arbitrum's because it only provides disputed transactions through the EVM. This speed is Optimism's advantage here, as disputes can be resolved quickly and without hindering future progress of the rollup chain.

The concern is "multiple rounds" of dispute resolution procedures, such as the one used by Arbitrum. In theory at least, spammers could stall a Rollup's progress by launching a series of consecutive challenges, each of which takes a considerable amount of time to solve. In fact, this is a problem that has plagued previous iterations of Arbitrum.

However, Arbitrum's newer protocol works for this problem, an elegant solution called "pipelining". Pipelining allows network validators to continue processing transactions for final approval, even if previously processed transactions are in dispute. This creates a "pipeline" of recently processed but not yet completed transactions, rather than a bottleneck that prevents orderers from processing transactions and network parties submitting challenges.

image description

in conclusion

in conclusion

Besides the design of the dispute resolution process, there are other notable differences between Arbitrum and Optimism, notably

• their codebase architecture, and

• Their approach to miner extractable value (MEV)

To summarize the differences very briefly: Optimism has a relatively minimalist codebase, while Arbitrum's is more complex and ambitious; Optimism has indicated in the past that it favors an MEV auction approach, while Arbitrum plans to implement a Fair Sorting Service (FSS). Naturally, both of these points of comparison deserve detailed treatment in separate articles. MEV, in particular, is a matter of philosophical debate between the two projects - although both are expected to use a trusted sequencer model for simplicity, at least in the early days after launch.

Ultimately, taking a step back from the protocol-level nuances (as important as they are), there are also "soft" things that differentiate these two heavyweights: bootstrap strategy, incentive design, and community spirit, to name a few. In fact, Optimistic Rollups will have to become their own world and not just an appendage of Ethereum if they are to be successful in the long run. Therefore, scaling is not so much an arms race as it is a war on multiple fronts. It may have one winner; it may have multiple. It may last for years; it may end sooner or later. This is sure to have a major impact on the future of cryptocurrencies.