Looking at the evolution of Web3 over the past few years, it's quite interesting.

Initially, everyone was focused on moving assets onto the chain. Later, it became clear that just having assets wasn't enough; the logic also needed to be on-chain, which is why smart contracts became popular. Then, the focus shifted to how governance could be brought on-chain. But now, there's an increasingly glaring problem—where are users' data, history, content, and identities?

If you really look at projects claiming to be on-chain applications, you'll notice an awkward phenomenon: assets and contracts are indeed on the chain, but the most core elements—users' content, interaction records, social relationships, behavioral traces—are still mostly stored in centralized databases. This isn't due to technical limitations; it's because there's been a lack of a reliable solution.

This is why Walrus has sparked discussions when it appeared. It didn't just shout slogans like "I'm cheap and fast"; instead, it asked a deeper question: if data is truly part of Web3, why can't it be verifiable, tamper-proof, and decentralized like on-chain assets?

Technologically, Walrus doesn't use traditional backup schemes. Instead, it employs a distributed structure based on erasure coding—you upload data, which is split into multiple chunks, then encoded mathematically to generate redundant fragments that are dispersed across nodes. In simple terms, it's not about "store one copy plus a few backups," but about creating a mathematically self-healing network.

With the current setup, as long as 60%-70% of the fragments are alive, the entire data can be fully reconstructed. This design truly opens up a new window.