Hacker News front page5 min readintermediate
Bend 2 and the Vibe-Coding Trap
Summary
The author argues that “vibe‑coding” with LLMs can cause engineers to reinvent existing solutions, using Bend 2—a language that asks LLMs to write long formal proofs—as a case study. By re‑implementing the same demo in SPARK/Ada, the author shows that established formal‑verification tools achieve the same guarantees with a tiny proof, highlighting the need for domain research before prompting LLM…
- Vibe‑coding lets you ship a large, broken design before you discover that a mature solution already exists.
- Bend’s 58‑line law spec and 442‑line LLM‑generated proof are dramatically larger than the equivalent SPARK implementation, which proves the same properties with a 12‑check GNATprove run.
- Formal verification is a well‑established field; ignoring it leads to unnecessary complexity and duplicated effort.
- LLMs will not point out existing tooling or state‑of‑the‑art techniques—you must do the research yourself.
If engineers rely on LLMs to generate entire languages or compilers without surveying the literature, they risk building solutions that are far less efficient, harder to maintain, and already superseded by open‑source projects. This wastes time, token costs, and can lock teams into sub‑optimal arch…
5/10
