Rollups, data availability layers & modular blockchains: introductory meta post
Current monolithic blockchain designs are cripplingly inefficient. We are bound by the trilemma where higher throughputs necessarily mean lower security and/or decentralization. Modular architectures that specialize and split into discrete execution layers (e.g. rollups, volitions), security/consensus layer, and data availability layers (e.g. data shards) bring 100x-10,000x greater efficiency to the blockchain industry. This is the only way we scale to global ubiquity and millions of TPS, period. Further, we do this without materially compromising on security & decentralization. I expect 90+% of all blockchain activity to happen on zkRollups & volitions within a couple of years.
A simpler thesis: Long term, anything any monolithic blockchain (L1) can ever do, a single zkRollup can do it significantly better. And there can be hundreds or thousands of zkRs interoperating relatively seamlessly.
Here’s a collection of posts on the matter:
(Side-note: my thoughts have evolved over the last 6 months, and indeed, the more I’ve learned about rollups, the more magical it seems. For example: I wasn’t aware that you could have internally sharded zkRollups with full composability till very recently.)
Firstly, if you want to learn how rollups work, here’s the quintessential introduction to them: An Incomplete Guide to Rollups (vitalik.ca).
A brief summary of how we got here. From single-ledger monolithic chains, to multi-chain, to sharding, to rollups & data shards (affectionately called rads).
Why rads are the only sustainable solution to global scale. Sharded networks are better than single-chain, but rads & modular architectures are >100x superior still.
Volitions are the best of all worlds — an evolved manifestation of modular vision that goes beyond rollups. Once you have volitions, you don’t need anything else.
Addressing all myths and misconceptions about rollups. Tl;dr: Rollups are far superior to monolithic blockchains in almost every conceivable way.
Why security layers are the most challenging and there are only two networks — Bitcoin & Ethereum — that have built adequate network effects. It’s very likely volitions will simply settle on Ethereum (or if Bitcoin adds functionality for ZKP verification) for the foreseeable future, and just leverage multiple DA options if the DA on Ethereum is saturated.
Some of the possibilities with a modular architecture. Why “L1” and “L2” are very restrictive, and some fun thought experiments on extending nomenclature.
Ethereum’s unique place in new rollup-centric world and how rollups on Ethereum alone can scale to dozens of millions of TPS by the end of the decade. Further, a Twitter thread about this roadmap. Combine this with volitions, non-Ethereum rollups, state channels on rollups, self-sharded rollups composing together etc. etc., not to mention further advances in scalability tech, the blockchain industry finally has the answer to nearly infinite throughput.
Caution that this is the long game, and the transition to a rollup-centric Ethereum (and blockchain industry at large) will take a couple of years. Likewise, early rollups will have “training wheels” which will take time to decentralize progressively.
All hope is not lost for monolithic chains — they can pivot to the modular revolution like Ethereum & Polygon have done. Paths forward for monolithic chains, an example of Solana becoming a zkR to remain competitive with StarkNet and zkSync 2.0.
That’ll be enough to get you started. I believe at this point I have said everything I wanted to about rads and modular blockchains. If there’s something I missed, please let me know. I’ll, of course, continue to update you about the latest happenings in the space, but the overall thesis is complete from my perspective. There’s a lot more in my blog for further reading; indeed there’s even some from before I started the blog. For laughs, this is one of my earlier meme attempts— Rollups are 4th gen blockchains. The memes have evolved, but the thesis has remained unchanged. Alas, I was permabanned from r/cc, which is why I started this blog. (Sorry, couldn’t resist!)