I came across a tweet by Hayden Adams. I was going to say, chains are servers with just one superpower - strict global consensus, but hey, I’ve already written about that. So, instead, let’s examine Hayden’s claims.
First, let’s address the tweet being quoted, that “i cannot think of something that makes less fucking sense than a million rollups” -
Today, “Web2” runs on >100,000,000 servers. “Web3” (““ because its a very silly meme) necessarily more inefficient. Even at the endgame state of ZK proving, chains have to process signatures, generate proofs, feature some degree of replication, and are much harder to parallelize - indeed, I haven’t seen any mechanism which has achieved parallelization on the same application. If you think that “Web3” will take even 1% of “Web2’s” compute, then absolutely there’ll be millions of chains. And barring a couple of exceptions these will be rollups or rollup-like constructions because it’s impossible for a million chains to self-secure. Personally, I don’t think 1% is realistic, but I also think a few chains is far too pessimistic - the reality is probably somewhere in the middle. I do think there’s a possibility that a lot of these will be hybrid applications, with constructions like coprocessors where the lines between chain and server are blurred. Anyway, enough of a digression, let’s get back to Hayden’s tweet…
Immediately, the first point, “history can be verified” is wrong. All high-performance chains (L2s, L1s, doesn’t matter) will prune / are already pruning history and state, and eventually even high-resilience chains like Ethereum L1 have a roadmap for both history and state expiry. Bitcoin is probably going to be the only remaining relevant chain where history can be verified. This is not an issue at all, because the public blockchains are not meant for storage or history - they are meant for real-time consensus. Indeed, there are far better solutions for historical storage, which will be used by chains for history. As long as there’s one copy, it’s all good.
“Accessible to all + can’t be shut off” - once again, you can easily have a server that is accesible to all. You just need to extend it to a peer-to-peer network and it can’t be shut off.
“Interoperate with other servers” - traditional servers interoperate all the time, and to much greater extent than can ever be possible with public blockchains.
If you just need the above three things, you’re much better off with an open peer-to-peer network (e.g. BitTorrent) that doesn’t require strict global consensus - it’s significantly more efficient and decentralized.
The real superpower of chains, then, is… you guessed it… strict global consensus. This enables unique properties that lead to applications like Uniswap - I have discussed these in detail across multiple blog posts so I won’t rehash them here.
This is why I spent most of 2023 writing about this topic - when even the inventor of the most successful application can’t quite explain the strengths of public blockchains, and it’s such an unpopular topic in the face of infra and degeneracy obsessions that run rampant in crypto. Of course, it’s much more likely that I, an amateur blogger, am delusional and wasting my time.