Thread by @BillyBobBaghold

L2 Mega Thread

I write threads mainly to educate and share my research findings, but also to refine and clarify my own ideas. Research and writing force you to understand something and help to test your assumptions and challenge biases.

I also like to trade narratives - this is why I rotated from #BSC to $MATIC to $SOL to $FTM ecos this year as narratives shifted. I’ve noticed what I think is the beginning of a new narrative, being the move to $ETH L2s.

So I thought I should start researching L2s to hopefully find some alpha for us and pick winners. So here it is below.

I borrow from @DCbuild3r’s superb article (and elsewhere - see further reading below), so if you want to dive deeper the article below is a great start. Thanks also to @cryptoseco for sharing it with me. I suggest you follow both.

mirror.xyz/dcbuilder.eth/…

L2s are an example of a modular approach to blockchain scalability (as opposed to monolithic). L1s in modular blockchains prioritise security and decentralization whereas L2s leverage this while enabling far great scalability.

If the L1 network were to go down, all L2s would also go down, since L2s rely on the L1. However, if an L2 were to go down, all of the funds are still safe and secured by the L1.

The pluses and minuses of Monolithic versus modular blockchains is outside the scope of this thread, but see the reddit and @bankless podcast links below for more info (even if biassed towards modular) reddit.com/r/ethereum/com…

L2 scaling solutions have two key features that distinguish different implementations; 1/ the type of cryptographic proof used; and 2/ whether data availability is on or off-chain.

I will start with data availability before going into more detail on proofs, since I predict the narrative battle will be more around proofs, at least initially

On-chain: basically all relevant data from the L2 is rolled up/bundled to the L1 via a cryptographic proof.

Off-chain: certain data is held off-chain by the L2. So generally less secure and decentralized, but much more scalable.

If data availability becomes more of a narrative battleground, I may go into more detail into the pros and cons, but this thread will focus more on proofs.

In terms of proofs, there are two main types - validity and fraud.

1. Validity proofs: You may have been seeing more references to validity proofs, since these are the sort of proofs ZKs (zero knowledge) use.

ZK validity proofs rely on cryptography - i.e. there is no room for human error or fraud. It is impossible for operators to commit an invalid or manipulated state. In summary, they are very secure.

Projects Using Validity Proofs

@zksync and StarkNet are two specialist ZK L2s which hope to be generally EVM compatible, but neither has yet launched - @zksync is in testnet, whereas StarkNet hopes to launch next week.

More under the radar ZK EVM compatible projects include the OG $SYS - which will release an EVM compatible ZK L2 next month and $MATIC.

You wouldn’t consider $MATIC under the radar generally, but did you know it was launching a ZK L2? It is doing so following its acquisition of ZK specialist Hermez. They expect to launch a testnet by the end of Q4 2021, with a mainnet launch in Q2 2022.

2. Fraud proofs: Optimistic roll-ups (“ORs”) use fraud proofs. Validators check that data going from the L2 back to the L1 is indeed correct. If an incorrect state transition is detected, validators will point to the invalid tx and revert while slashing malicious operators.

The validation/challenge process takes time (unlike ZKs) - this is why @optimism withdrawals to $ETH L1 take a week. However protocols like $METIS hope to speed up this process to a matter of minutes via their ranger and sequencer system next year.

Projects Using Fraud Proofs

Several OR chains have already launched. Optimism and @arbitrum ecos continue to be built out.

I think the most promising is $METIS (more info on this and @metisdao below) and I am also watching how the $BOBA ecosystem develops.

https://twitter.com/BillyBobBaghold/status/1459791867084787712?s=20

So which is superior, ORs or ZKs?

ZKs are generally thought to have superior tech since they rely on math as opposed to sequencers and game theory for security.

Another major technical advantage of ZKs is that they can enable on-chain privacy by default.

Privacy will be a massive narrative play at some point in this cycle and is key to blockchain adoption generally. @aztecnetwork is building a fully private chain which will support defi functionality. However they aren't EVM compatible due to tech and smart contract limitations

So as hinted above, the main drawback of ZKs relative to ORs is EVM compatibility - while specialist ZK projects are working on it and launching soon none have yet and their compatibility, while likely sufficient, isn’t 100%.

Which is the better trade?

So even if ZKs have superior tech in theory, they will still need the devs to build on them. However right now it's much easier to build on OR L2s like $METIS, Optimism, Arbitrum and $BOBA due to their EVM compatibility/equivalence.

OR ecos have been and will likely be built out much faster than ZKs’ for the next few months, particularly when dev incentives are considered. Crypto moves fast and so would ORs’ network effects as more projects build on them while waiting for the ZK L2s to be polished.

Whether this network effect will outweigh ZKs’ superior tech is TBD. In this regard it is worth bearing in mind two things - 1/ Crypto moves fast - just think that no one was talking about defi two years ago and

2/ Projects don’t need the best tech to pump - no one credible was claiming #BSC or $MATIC had the best tech earlier this year, but both chains still printed multiple Xs in a matter of months.

So in the short to medium term (next few months) I will mainly allocate to OR ecos like $METIS, as they will be built out faster and will likely provide the greater incentives and rewards to users. I have and will be allocating ZK projects, with a longer time horizon.

So far I am in $MUTE and $SYS and am looking at a few others.

Drawbacks of L2s?

While L2s have several benefits (primarily being low fees without compromising a great deal on security, scalability or decentralisation) this thread would not be complete without listing the drawbacks:

- L2s are not composable (yet) so interoperability is difficult if not impossible
- Oracle infrastructure is not yet fully developed
- Liquidity is fragmented thought L2s, although protocols like Hop Protocol are endeavouring to solve this

Anyway I hope you found this thread useful. I think we will begin to see a narrative shift from L1s to L2s. This isn’t to say all L1s will suffer - I am still very bullish on $FTM for example, but I am very mindful of the wider trend.

Further reading

mirror.xyz/dcbuilder.eth/…
blog.matter-labs.io/optimistic-vs-…
medium.com/aztec-protocol…
reddit.com/r/ethereum/com…

l2beat.com
zksync.io/zkevm/#what-ar…
medium.com/aztec-protocol…
syscoin.org/news/nevm-and-…

Most alts are bleeding or flat vs $BTC.

Except for L2s like $METIS $SYS $IMX $MUTE $ZKS $LRC.

L2 season is starting. I hope you paid attention.