Short-term OP, long-term ZK?

Because Vitalik Buterin said this, many people think it is a "golden rule", but the actual situation is much more complicated.

The common comparisons have been discussed online for a long time - OP is based on game theory < ZK is based on mathematics; OP1 has a two-week withdrawal period < ZK has a withdrawal period of only a few minutes to a few hours; OP has better EVM compatibility > ZK compatibility still has a long way to go... No more nonsense, Vitalik's long-term and short-term comments are generally based on the above three fundamentals

However, game theory vs. mathematics - users can't experience any difference at all, and they don't care at all; two-week withdrawal period vs. a few minutes or hours - these differences are all smoothed out by the bridge; EVM compatibility - OP is indeed better now, but as ZK slowly develops, this will eventually be smoothed out

So let me say something different

1. First of all, performance

A common saying is that the TPS of the ZK system is higher than that of the OP system. The main reason for this is that the compression ratio of ZK is higher than that of OP. In other words, when submitting compressed transaction data to L1, ZK can submit more transactions than OP due to its higher compression ratio, which naturally makes the TPS higher.

However, this statement ignores the huge cost and time required to generate proofs for ZK.

Therefore, the performance of OP vs ZK is likely to rise alternately, and eventually reach the same result - the following is purely speculation and may not be accurate

1. ZK has just been launched - OP TPS is high, because the cost and time of ZK proof far outweigh the advantages brought by the compression ratio

2. ZK's Prover architecture is relatively mature - ZK TPS is high, and with the emergence of FPGA or ASIC machines, the cost and time of ZK proof have been greatly reduced, and the advantage of compression ratio has begun to emerge.

3. Pro-Danksharding is launched - OP and ZK TPS are almost the same again, because L1 does not use Call Data, but uses a Blob with much larger space and lower cost as DA. Therefore, the compression ratio advantage will be far less obvious than in the current Call Data era. The small advantage of compression ratio is basically offset by the small disadvantage proved by ZK. The theoretical TPS limit of OP and ZK is basically limited to the hardware processing power of the Sequencer.

2. The second is the actual advantages of ZK in the market

Cryptography > Game Theory, the withdrawal period is much shorter than OP, etc. are the advantages of ZK technology, but not necessarily in the market. The blockchain world has never been "technology-only". Just like ETH has switched from POW to POS, there are still many old people and technical experts who are tirelessly publishing articles to prove that POW is better than POS, and you have to admit that many of their arguments are really reasonable.

However, it doesn’t matter. The market believes that POS is the future of the new public chain (excluding BTC). What can you do?

So what is the actual advantage of ZK in the market? I can think of two

1. As a "prominent discipline" in the current blockchain technology, ZK may bring a new industry chain to the forefront, just as POW brought a series of upstream and downstream industry chains such as mining machines (from CPU to GPU to FPGA to ASIC), mining pools, mining farms, and computing power derivatives to the forefront. ZK may also bring a similar POW-based industry chain from proof to verification, based on hardware supporting services.

2. ZK can do more - such as implementing privacy functions (Aztec), such as Vitalik's recent article "What kind of layer3 is meaningful?" - which mentions a scenario where the Arb native token "cross-chains" to Optimism (Wrap method). Because it relies on ETH L1 at the same time, the Wrap contract on Optimism can completely bypass the current "unsafe" bridges by reading the Merkle proof of the deposit contract receipt uploaded to L1 by Arb. However, in theory, OP-type L2 deposits should wait for a fraud window period (7 days) to pass before they are considered safe, so it is difficult to do this. If it is replaced with ZK, this scenario will be no problem.

3. Finally, let’s talk about the end of ZK and OP

Do you think that ZK will be usable soon because ZK-sync, Scroll, etc. have already started Alpha testnet?

That’s too naive. ZK still has a long way to go. For example, the circuit code of the official ZKEVM of the Ethereum Foundation is more than 30,000 lines. According to Vitalik Buterin’s original words, it requires “very long development and continuous testing” and “it will be impossible to fully rely on the security brought by the ZK system within a few years.”

Of course, although OP is leading, it has not yet completed the journey. For example, due to the change of OVM architecture, Optimisim has not yet launched the core fraud proof function of OP, and many people don’t know

So what might the end be like? Vitalik Buterin also gave a discussion, which I personally think is quite reliable. It may be like this in a few years.

What does it look like? All in this discussion thread

Specifically

Before zkEVM matures, OP is the main focus, and ZK is the auxiliary

1. Release Block

2. Wait 24 hours

3. If there is no fraud challenge during the period, publish ZKP and Finalize block.

Else (challenging), introduce Governance and use the 2of3 model to determine the final result

After zkEVM matures, ZK will be the main component and OP will be the auxiliary component.

1. Release Block

2. Regularly publish ZKPs

3. IF ZKP is released normally during the specified period, Finalize

Else (ZKP is not released normally during this period, whether Prover is down or there is a bug), the system switches to the Optimistic mechanism until the ZK mechanism is restored