Why does the Optimism fund withdrawal to the Ethereum mainnet require a 7-day challenge period?

Author: @yyy

Coinbase announced its entry into the market with an L2 based on OP Stack ------ @BuildOnBase, reigniting the sentiment of being "Bullish on Optimism." Taking this opportunity, this tweet discusses the challenge period of Optimism.

It is well known that withdrawing funds from Optimism's Layer 2 to the Ethereum mainnet requires a 7-day challenge period.

So why specifically 7 days? Wouldn't 3 days or 5 days suffice?

Before answering this question, let's first understand why a challenge period is necessary.

Rollups improve the performance of the Ethereum mainnet by separating the execution layer from Ethereum. This means that transactions are executed on Layer 2, but to inherit the security of the Ethereum mainnet, transactions that occur on Layer 2 must be verified on the mainnet.

In simple terms, the Ethereum mainnet needs to confirm that the transactions submitted by the sequencer indeed occurred on Layer 2, i.e., the validity of the transactions.

Rollups are divided into zk-Rollup and Optimistic Rollup based on the different methods of validating transaction validity.

The validation methods are mainly classified into two categories: zero-knowledge proofs and fraud proofs. Under the zero-knowledge proof mechanism, transactions have near-instant finality.

For ORUs based on fraud proofs, the finality of transactions depends on the length of the challenge period, which explains why withdrawing funds from Optimism's Layer 2 to the mainnet takes 7 days.

Under the ORUs mechanism, it is optimistically assumed that the Rollup transactions submitted by the sequencer are valid. Then, during the challenge period, if validators discover invalid transactions, they can initiate a fraud proof.

If no one challenges after the challenge period, the transaction is deemed valid, and at that point, the transaction has finality.

But does the challenge period really need to be as long as 7 days? Let's simplify the entire challenge process: validators discover invalid transactions by reconstructing the state root and then issue a fraud proof. This series of actions takes time, which means the challenge period cannot be 0.

Let's assume the above challenge process is very quick, completing in 10 blocks (about 2 minutes).

Of course, such a quick response is unlikely, so let's slow down the time by 10 times, to 100 blocks (about 20 minutes). This time is still far less than 50400 blocks (7 days).

The above discussion is based on the most ideal situation. What if there are malicious actors in the network? How long would it take to detect malicious behavior?

Or to put it another way, how much longer does the challenge period need to be to eliminate malicious behavior? The malicious behavior referred to here is the act of publishing a large number of transactions over a period to spike gas prices, thereby preventing fraud proofs from being accepted by validators.

Here, let's make another assumption: suppose there is a $2 billion risk exposure on Optimism, and a malicious actor is willing to spend $1 million in gas fees (priority fees) per block to "race" transactions containing fraud proofs.

In this case, the maximum time they could "hijack" the network would be 2000 blocks (6.67 hours). Of course, the examples given earlier are based on some very simple and crude assumptions.

It seems that from any perspective, the challenge period does not need to be as long as 7 days. It is difficult to understand the nonlinear relationship between the length of the challenge period and security.

Perhaps 7 is a very meaningful number for the founders of Optimism, just like how people speculate that Satoshi Nakamoto is a football enthusiast because of Bitcoin's 4-year halving cycle.

So, can the challenge period for ORUs be shortened to 3-5 days without compromising security? Yes, but it doesn't significantly optimize user experience.

What about shortening the challenge period to 1-2 days? The argument that this could effectively optimize user experience seems to make sense.

@0xMantle, an ORU incubated by BitDAO, can shorten the challenge period to 1-2 days. The optimization logic of the challenge period is based on the MPC solution.

In simple terms, the sequencer does not directly submit the batch of transactions to Ethereum but first submits it to the MPC network, where MPC nodes jointly sign it. Once the transaction is confirmed as valid, it is then submitted to Ethereum.

Therefore, the core difference is:

Optimism submits the batch containing the state root as the execution result to Ethereum;

Whereas Mantle submits the batch containing the jointly signed state root by MPC nodes as the execution result to Ethereum.

The goal is to reduce the trust risk of Rollup execution results.

However, MPC verification is ultimately still an off-chain verification solution, which is better than nothing, providing users with a bit of psychological comfort.

To make a somewhat inappropriate analogy, the MPC verification solution is like a semi-decentralized zk solution, where the former's security is based on economic assumptions, while the latter's is based on cryptographic verification.

This concludes some thoughts and discussions on the challenge period of ORUs, which may not have strong logic but serves as a starting point for further exploration. BTW, I am very much looking forward to seeing ORUs with significantly shortened challenge periods in the future (at a time-sharing level).