Vibe Coding's Variability Void: New 'VbR' Lets LLMs Generate Clean Binaries

In an emerging AI-driven paradigm called 'vibe coding,' an LLM generates an entire program from a natural language prompt. But what happens to the variability that traditional software engineering carefully builds into code? A new paper from Xhevahire Tërnava presents an exploratory analysis of 10 vibe-coded C/C++ projects, revealing that there is near-zero in-artifact variability—meaning no compile-time or runtime alternatives. Instead, all variability decisions are resolved at a single new binding time: generation time, the moment the LLM produces the source code.

Rather than treating the lack of variability as a defect, the authors propose Variability by Regeneration (VbR), to their knowledge the first product-line approach in which the LLM acts as the derivation engine. VbR generates a purpose-built, dead code-free binary for each variant from a declarative specification, while a variant dispatcher transparently routes user requests to the matching binary. The paper formalizes VbR, contrasts it with classical software product line derivation, and demonstrates its full pipeline on a 'wc' product family. The key insight: for software product line engineering, variability in AI-generated software belongs in the specification, not in the code.