Recently, I delved into the evolution of Ethereum wallets and found that the concept of account abstraction is becoming a key turning point in the entire ecosystem.

It's quite interesting. The wallets we currently use, like MetaMask, are actually based on externally owned accounts (EOAs). This design has been carried over since the Bitcoin era. But as blockchain applications become more complex, the limitations of EOAs are becoming more apparent—high private key management risk, lack of support for concurrent transactions, limited functionality. Often, you have to carefully safeguard your private keys, and losing them means your assets are gone forever.

Smart contract wallets (SCWs) offer an alternative approach. They use contract accounts as wallet addresses, enabling features like multi-signature, social recovery, and batch transactions—things EOAs simply can't do. More importantly, you don't have to directly manage private keys; you can control assets through social recovery or other mechanisms. This is the core idea of account abstraction (AA)—abstracting the account away from the signer.

The Ethereum community has been exploring this direction for many years. From Vitalik's initial idea in 2016, to later proposals like EIP-86, EIP-2938, EIP-3074, and now ERC-4337, each iteration aims to solve the issues of the previous solutions. Although EIP-3074 was proposed earlier, it was ultimately not adopted due to core protocol changes and security concerns. ERC-4337, on the other hand, was included in Ethereum's roadmap because it offers a lightweight implementation without modifying the core protocol.

The design of ERC-4337 is quite clever. It introduces the concept of UserOperation, which packages multiple user actions into a single transaction via a bundler. The process includes verification and execution cycles, with an entry point contract coordinating everything. There's also a Paymaster role that can pay for gas on behalf of users, greatly improving user experience.

But honestly, ERC-4337 isn't perfect. Vitalik himself pointed out three main issues: difficulty in user migration, additional gas costs (roughly 42k for a UserOperation vs. 21k for a normal transaction), and smart contract security risks. His recommendation is to push forward with ERC-4337 in the short term while gradually addressing its weaknesses through other EIPs.

Several projects are already exploring this space. Unipass is a traditional smart contract wallet featuring social recovery via email and multi-role key management. Candide is built on the ERC-4337 standard, combining Gnosis Safe's multi-signature framework with Stackup's implementation.

My observation is that future wallet forms are likely to be more B2B2C rather than purely consumer-facing products—offering direct user access but more importantly providing SDKs for other applications to integrate in-app wallets. Infrastructure like bundlers and aggregators are still operated centrally now, but in the long run, they may evolve into modular networks. However, since this involves value capture, the economic games among stakeholders will be quite complex.

In summary, ERC-4337 and the account abstraction direction are correct, but this is just the beginning. In the short term, more wallets and applications based on this standard will emerge, while long-term improvements will require protocol-level optimizations. For those evaluating related projects, it's essential to consider technical implementation, community engagement, and user experience across multiple dimensions.