Without AA, there is no future of chain abstraction: What stage has AA reached?

Author: Austin King, Co-founder of Omni; Translation: Jinse Finance xiaozou

As I delved deeper into chain abstraction, one thing became extremely clear: without account abstraction (AA), there is no future for chain abstraction.

People have been talking about AA for years, so what is the current development of AA? According to my research, the latest developments and upcoming developments of AA will soon be put into production.

First, let's take a quick look at what account abstraction (AA) is?

Today, basically all transactions are "authorized" by signatures generated by private keys. This is bad in some ways, because it means:

1) People can't log in to their email, use their mobile phone, or 2face to authorize transactions.

2) Every transaction needs to be signed with that key, which effectively forces users to click a ton of buttons.

Account abstraction gives us a lot of huge benefits:

1) People can authorize transactions in a much easier way than being forced to remember 12 random words. Think about how many users this will unlock - imagine "sign in with Google" like you do on many websites today, and once you're logged in, you can also make crypto transactions.

2) You can bundle transactions.

3) You can create backup keys that you store with your parents or close friends to help you recover your account if you forget your password.

4) When a user wants to do something, it lets other parties handle the gas payment and submit it to the chain (solvers).

Okay, it's cool, so many advantages, but I at least feel like we've been talking about AA for years, so where is it at? Why are we still using MetaMask.

There are 3 key Ethereum Improvement Proposals (EIPs) that will help you understand where we are in terms of establishing account abstraction as a wallet standard.

EIP: The first and most important 4337

Personally, when I saw this chart, my first feeling was that it was too complicated. So what exactly does this picture depict?

Basically, 4337 allows users to submit actions they want to do on-chain to a third party. The third party then packages all of these requests together and is responsible for putting them on-chain. This is cool because there is no need to upgrade the EVM at all. Now you can actually use an account abstraction wallet like this. But why aren't you using it yet? Everyone complains about how bad MetaMask is, so if there is a better option on the market, why don't we all use it?

Here are some of the problems with ERC-4337 right now:

1) Existing wallets can’t upgrade to 4337

2) It’s more expensive in terms of gas costs (i.e. transaction fees will be higher)

3) Easy to audit - In the above diagram, if you create an AA wallet, it’s possible that the bundler (the party that submits transactions on-chain for you) that you rely on may one day decide: “You know what, I don’t like this protocol. I won’t let anyone submit transactions related to this protocol.” It’s a bit like your bank is actually an intermediary for you to send money to others. If your bank doesn’t want you to do that, then you can’t send money.

Okay, there are some problems, but can we at least take a small step in the right direction? What if we allowed existing wallets to upgrade to account abstraction wallets? This is the main purpose of EIP-3074. However, this would require changes to the EVM. It would introduce new opcodes that would allow existing wallets to say, “Yes, I want account abstraction”. The difficulty here is that this would require real changes to the EVM, whereas 4337 is kind of cool because it doesn’t even require any changes to the EVM. It’s all strictly based on the status quo, i.e. no hard fork required.

Cool, looks like a path forward? Of course there are security concerns. Sometimes people click on phishing links, accidentally authorize a transaction, and lose all their ETH. This would make phishing links even more dangerous, because if you authorize the wrong transaction, not only can they take all your ETH, they can also unwind all your DeFi positions, make public statements from your ENS, and so on. The difference is like the difference between “your Chase bank account was emptied” and “this person stole your identity and now has access to all your passwords everywhere.”

Okay, let’s take another 4-digit EIP and see if we can fix this: EIP-7702.

This EIP was recently introduced by our common god and savior Vitalik himself, and it is actually an alternative to 3074. It does not require new opcodes, requires very few changes to work with 4337, and can temporarily upgrade to an account abstraction wallet to reduce risk.

EIP Introduction:

4337 - Brings users account abstraction.

3074 - Allows existing wallets to upgrade to account abstraction wallets, but requires new EVM opcodes and introduces the risk of someone giving away their digital identity with a single accidental click on a signature.

7702 - A safe alternative to 3074 with fewer changes to the EVM.

So, now we know about 4337, 3074, and 7702 — but we still don’t know why we’re still using MetaMask. 7702 is still under discussion in the Ethereum research community, but there’s a chance it’ll be included in the upcoming hard fork.

So, let’s move on to what products we can use today to get a better wallet experience. I’m sure there are a lot of them, but I think the one that stands out is the Coinbase Smart Wallet. The wallet is fully in production, and honestly, it really does offer an amazing user experience. To test it out, it took me 37 seconds to do a Google search and install a wallet. All I had to do was scan a QR code on my phone. No 12 random words, no words like “mnemonic” or “cryptography” to scare off the average user — it just works. Now whenever I want to authorize a transaction, I just use FaceID.

Back to the original intention of writing this article, there are two important reasons:

1) For the application we are building, it is easier for users to get started.

2) User operations submitted through the account abstraction provide support for the rest of the chain abstraction.

I’ve been in contact with a lot of companies working on chain abstraction, and I think the reason they’re all so motivated is that they understand that by focusing on their core responsibilities and working with others, the chain abstraction pie will grow faster than any other single system that tries to improve user experience, capital efficiency, and transaction costs all at the same time.

I have more research to do on the permission layer, but after that, I’ll dive into the solver layer and share my research with you along the way.