ZK (Zero-Knowledge Proof) track now has many contestants, but when you really read through these projects' whitepapers and technical documents carefully, you'll find that many solutions are just improvements and upgrades based on the existing Ethereum tech stack, not true innovations.

The reason Dusk stands out is not just because it also builds a privacy public chain—such projects are plentiful. The key lies in their technical choices. To truly run privacy computations on-chain, these developers surprisingly built an exclusive virtual machine from scratch—Piecrust. Only those who truly understand the technology can appreciate how challenging this is.

There are practical considerations behind such choices. When EVM was originally designed, it didn't reserve space for zero-knowledge proof computations. Its 256-bit word architecture, when used for elliptic curve-based ZK proofs, is extremely inefficient. Rather than trying to force compatibility, the Dusk team took a different route, building the Piecrust virtual machine based on WebAssembly (WASM).

What is the most clever part? They introduced a Zero-Copy mechanism in memory handling, using the rkyv framework at the core. As a result, smart contracts can read data without going through the inefficient deserialization process typical of traditional virtual machines. Data is directly accessible in memory, which significantly boosts performance when verifying complex privacy Merkle trees—an order of magnitude difference.

Looking at the proof system choice, Dusk also put effort into it. They adopted the PLONK algorithm, specifically the version supporting "Universal Setup." This design is more flexible in engineering and better suited for the long-term needs of privacy computing.

From the entire tech stack perspective, these choices are not made impulsively but are reasonable trade-offs for specific scenarios—optimizing for privacy computation, even if it means reinventing the wheel. This is the part worth pondering.