Pete Horne on Warpcast

Content pfp

0 reply

0 recast

2 reactions

Pete Horne pfp

I have a question because “ethereum account abstraction” seems to be a hot topic. My question is - if the fundamental feature of decentralisation is end user control as principal with no agent or intermediary, and the user id is a 2^256 integer that the user must control - what is there to abstract??? An int???

2 replies

0 recast

3 reactions

Pete Horne pfp

An ethereum account is an abstraction itself. So it’s turtles all the way down to the fundamental key pair, and the requirement for a user keeping the skey themselves or give up their agency because they can’t be responsible for it. It seems like a trade off, so is the ETH community saying it’s trade off time???

2 replies

0 recast

0 reaction

nicholas 🧨 pfp

the coolest thing i've seen here yet is the proposal to use passkeys as signers on AA contracts. you can limit the privileges of the passkeys, or require multiple sign certain types of txs. passkeys are generated on device on anything with a modern browser. apple advanced protection & recovery contacts solve recovery..

1 reply

0 recast

1 reaction

nicholas 🧨 pfp

for the vast majority, who cannot be expected to bury metal wallets etc. the icloud keychain backup of the passkeys will be e2e. the biggest problems are apple binaries could rug the entire thing with a single update, and also secp256r1 may have been compromised by the nsa

1 reply

0 recast

0 reaction

Pete Horne pfp

Thanks. So in essence these are user tools, maybe supported by some ERC standards, but they are not leading to node protocol level changes?

1 reply

0 recast

0 reaction

nicholas 🧨 pfp

right now the bleeding edge UX is passkey on device is a signer on a MPC that controls an EOA that is a signer on an AA. maybe eventually the MPC and EOA get cut out of the picture and the 7212 sepc256r1 precompile lets folks sign with passkeys directly if AA gets enshrined in evm then boundary btwn erc/node blurs

1 reply

0 recast

0 reaction

nicholas 🧨 pfp

returning to your original question, idk what the second A in AA is intended to mean, but in practice 4337 separates (1) the account (0x..) from (2) the authentication scheme(s) and signer(s) that control it and (3) the account that pays the gas.

1 reply

0 recast

0 reaction

nicholas 🧨 pfp

So in the future you could imagine counterfactually deploying a AA smart wallet that has a Magic link (or passkey, or any future auth tech) as a signer, and gas is paid for by gelato or another subsidy provider

1 reply

0 recast

2 reactions

Pete Horne pfp

That acronym list made read around! Time well spent, tho. Lots to ponder.

2 replies

0 recast

0 reaction

mike rainbow (rainbow mike) ↑ pfp

mike rainbow (rainbow mike) ↑

@mikedemarais.eth

to pete’s core concern tho about new intermediaries that could potentially introduce fragility — i’d say that the paymaster model is directionally concerning in that regard, but i’ve been assuming that paymaster model is just an iterative step and won’t be around forever (without rly having looked into it)

0 reply

0 recast

2 reactions

nicholas 🧨 pfp

you may enjoy the second half of the latest ep of web3galaxybrain.com with @henri from privy.io which covers some of this material

1 reply

0 recast

1 reaction