Re-understanding "modular blockchain" and Rollup

Last year, I was guilty of spreading the “modular blockchain” meme, although there were more influential players like Bankless, Celestia, The Daily Gwei, etc. who brought the term into the spotlight. This year, I haven’t really used the term “modular blockchain”.

It is abundantly clear that modularity is orders of magnitude more efficient than monolithic chains. The multiple failures of monolithic chains when tested at minimal stress make this clearer than ever. Modular execution layers also have a ton of work to do, but they have a significant lead over their monolithic counterparts.

I was wrong not on the technical side, but on the socioeconomic side. This has been highlighted before by Ali Atiia and Justin Drake‌. Consider the gold standard, an Ethereum rollup:

1. Execution layer: rollup

2. Settlement layer: Ethereum

3. Data Layer: Ethereum

By the way, I recently saw someone call an execution layer that processes data elsewhere a "rollup", which is not a rollup. Rollups must be settled on the same settlement layer and DA layer . The validity verification execution layer that publishes data on a different layer is called a validium and comes with additional assumptions. You can label them zkPorter, celestium, etc., but please don't call them rollups. The situation with fraud proofs is more complicated, so I'll skip it for now. The key point is that if the data availability is not agreed upon on the same protocol that verifies the state transition, then it is not a rollup.

A properly implemented rollup means you don’t have to trust it at all, and you can withdraw your funds from this rollup to Ethereum mainnet at any time. This is not completely foolproof, and in reality, you may have rollups with different security models and standards, but I would definitely expect all major rollups to provide some kind of clear exit mechanism that is isolated from the trusted rollup sequencer.

Different rollups have different security assumptions, you can have an immutable or enshrined rollup that provides the same security as Ethereum (assuming no vulnerabilities). To upgrade this rollup, you must use the EIP process, or deploy a new instance entirely. Many rollups will choose upgradability, which will be driven by token voting. This is an economic assumption similar to the proof-of-stake L1 upgrade, although rollups can try new upgrade mechanisms without tokens at all. There are other interesting risks that I won't discuss here (see Justin and Ali's comments above). Personally, I'm not worried about some of these risks, I believe a well-implemented rollup can achieve 99% of the level of a sacred rollup, but there will definitely be rollups with unimportant assumptions.

For EIP-4844 and later danksharding, we added a fourth layer: the expired history layer. This is a very simple 1-of-N trust assumption, but I'm going to add it to the mix anyway. Other data layers may choose not to expire history. So now you have:

1. Execution layer: rollup

2. Settlement layer: Ethereum

3. Data Layer: Ethereum

4. Historical layer: rollup and others

The ideal solution would be:

1. Execution layer: Ethereum

2. Settlement layer: Ethereum

3. Data Layer: Ethereum

4. Historical Layer: Ethereum

Of course, this doesn’t have to be Ethereum, it could be Bitcoin, but the idea is anything that provides strong security. Now, don’t be fooled into thinking this is a monolithic solution, it will be modular, but all held together by the holy (perfect) protocol:

1. Execution layer: holy rollup (e.g. zkEVM, or optimistic enshrined rollups after statelessness) (Note: before Justin Drake coined the term “holy rollup”, I called it “minimal rollup” in previous articles)

2. Settlement layer: Holy Settlement Layer (such as EL)

3. Data Layer: Holy Data Layer (e.g. danksharding)

4. History layer: Sacred history layer (e.g. Sacred Portal network?)

This leaves you with the protocols with the fewest assumptions and combines the maximum security into one protocol. As far as I know, Tezos is the only project currently built with this approach. And Ethereum will most likely have its own sacred rollup in the next few years, and of course, there will be external rollup options in addition to this. So you get the best of both worlds: maximum socioeconomic security + experimentation and diversity, which in turn will also affect the progress of the sacred layer.

So, as mentioned above, why not have a lot of “modular L1s”? What you want to do is aggregate, not segment, security and liquidity. Having a lot of “modular L1s” would be a very fragmented and insecure mess. However, I think having 2-3 modular L1s would be a desirable outcome. You would have at least one modular L1 that aggregates to a state-level maximum security, and then perhaps a long tail of niche modular L1s. In fact, most industries will consolidate over time, leaving 2-3 major players. The pressure will be even greater for proof-of-stake blockchains due to the security accumulation mentioned above.

I have always been very interested in volitions in the past, but it quickly runs into a bunch of assumptions:

1. Execution layer: volition

2. Settlement layer: Ethereum

3. Data layer: Ethereum, zkPorter, Celestia, Polygon Avail, adamantium, etc.

4. History layer: The solutions for the data layer are volition or other

You can see that this is clearly not as elegant as a modular L1, although it works just as well as a rollup if you’re in rollup mode. There are a lot of nuances here, but in short, you now need to trust another entity - the DA layer. The most interesting solution that still interests me is adamantium. Here, by custodying your own data, or choosing your data providers, you completely abandon trusting a different and weaker honest majority consensus. Honest minority consensus on a validium DA layer has not yet been investigated, but I believe has great potential as well. (One could argue that DACs fall into this category, but the problem is that DACs are permissioned. Also, note that honest minority DAs do not apply to rollups - only to validiums that verify state transitions on an honest majority settlement layer.)

In reality, we are likely to quickly move to a world where we see rollups, volitions, validiums, etc. coexist. With EIP-4844 and danksharding, Ethereum will have enough data capacity, but if demand on the blockchain grows exponentially, we will see additional data layers handle the excess demand. We will also see some monolithic chains persist through the network effects they are building now, and strong marketing, despite the inherent insecurity of monolithic cross-chain bridges and the severe inefficiencies of monolithic execution layers.

However, I am absolutely convinced that even the most stubborn monolithic chain projects will eventually adopt modular components. You simply can’t turn down a 1,000x efficiency improvement! Unless, of course, you don’t need scale or innovation.

Ideally, we can form a clear ranking of execution layers by security:

Holy rollups > rollups >> validiums > AnyTrust >> Monomer sidechains & alt-L1

Note: The original author is Polynya.