SoloChance.org

I'm still building my intuitions around how exactly the architecture of L1/L2/L3 will shape up, but I currently believe there are a few substantive benefits to L3s: 1. settlement costs for L3s will be 10-100x cheaper. these costs are small but for devs they are still meaningful. 2. just in time bridging from L2 to L3 will be fast, cheaper, and significantly more capital efficient than going from L1 to L2 or L2 to L2. 3. the pathway for onramps to L3s will be much more streamlined because you have L2 stakeholders who will work hard to make them seamless via routing through L2. I think that on a long time period some or all of these may be solved by proof aggregation but in the short to medium time horizon am still working through how exactly that will shape up. I also think that my thinking here is also influenced by a belief that we will see much more homogeneity at L2, with the ultimate end game being that many L2s either become tightly coupled to each other (eg the superchain) and/or to L1 (via based sequencing and native zk prover support in the L1). in parallel, I expect we will see much more experimentation at L3 with different VMs (eg orbit, MoveVM, svm, parallel EVM), data availability decisions (eg Celestia, eigenDA, S3), and centralization characteristics. my best mental model for L3 right now is onchain connected servers - and I think there will be millions or billions of them. would love to hear other folks thoughts on all this - I'm still very much in the learning phase! and if you want to build an L3 on @base, lmk and I'll do everything I can do help!