As at all times, Vitalik’s articles are tremendous attention-grabbing and tightly-packed with good data:
One of many makes use of he briefly describes, for L3, is ” L2 is for general-purpose scaling, L3 is for personalized scaling.”. Although it would sound easy at first, I feel this phrase hides deep and necessary implications
Customized scaling is probably going probably the most promising use of L3. One can allow exceedingly increased computational masses with application-specific rollup chains due to native consensus. Such utility chains (in distinction to conventional sidechains) would use the general-purpose base layer (L2) for dispute decision, thus deriving the safety of the bottom layer, by way of a 1-of-N safety mannequin.
The shift from world to native (or application-specific) consensus impacts how DApps relate to one another. This design would possibly come at the price of increased friction to composability, however that may be a concern that’s irrelevant for a broad class of purposes.
Alternatively, purposes acquire the flexibility to operate with out competing cannibalistically for the pc energy of validators. It’s attainable to keep away from community gentrification and make sure that issues outdoors DeFi survive.
Particularly, purposes that run on their very own optimistic rollup chains with interactive fraud proofs (as in comparison with ZK rollups) can assist computational masses corresponding to mainstream (e.g., involving billions of instruction steps and huge reminiscence tackle areas). It might allow a brand new class of decentralized purposes that use different VM architectures working complicated decentralized logic with mainstream software program stacks.
With the ability to customise the app chain and select the profile of validators can be a plus. Possibly everybody that cares a couple of cool decentralized MMORPG has a beefy pc that may course of a bunch of stuff. You additionally get to customise the sequencer per utility, perhaps a DEX might forsake market maker charges for computation.
Ideas shared by https://twitter.com/erickdemoura on twitter.
What do you suppose? I completely agree that we’re overusing world consensus and neglecting this concept of leveraging the 1-of-N safety mannequin to “localize” consensus and customise scaling.