Jas: AI-Paired Engineering as a Revival of N-Version Programming

2026-06-05

AI Summary

Traditional N-version programming is rarely feasible for solo developers due to its prohibitive human resource costs, thereby limiting the construction of highly reliable cross-platform software. This work proposes a novel approach that integrates AI-assisted development with the N-version paradigm by treating a precise, executable YAML specification as the single source of truth and implementing it in parallel across five native platforms—Rust, Swift, OCaml, Python, and Web—ensuring consistency through differential testing. A single developer completed all ports within approximately 120 evening hours, sharing 23,000 lines of specification while producing platform-specific implementations ranging from 0 to 95,000 lines of code. The results demonstrate that this method successfully recovers the fault-tolerance benefits of N-version programming while maintaining manageable development costs, thereby extending the feasibility of building complex cross-platform systems by individual developers.

Abstract

I report a case study in AI-paired software engineering: five working ports of a vector illustration application across Rust, Swift, OCaml, Python, and browser-based platforms, built by a single developer in approximately 120 evening hours. The methodology pairs AI-assisted implementation with two safeguards -- a precise executable YAML specification serving as the single source of truth, and parallel implementations functioning as a built-in differential-testing layer. The five ports share a 23{,}000-line specification; per-port native code ranges from 0 to roughly 95{,}000 lines, reflecting the specification's escape hatch. I argue that AI-paired engineering, conditional on these two safeguards, makes feasible scope of work that conventionally requires multiple developer-years, and frame the methodology as a revival of N-version programming, a 1980s approach abandoned on cost grounds that AI changes. The paper reports concrete artifacts and honest limitations of the single-developer case study.

Problem

• AI-paired engineering
• N-version programming
• multi-platform software development
• specification-driven development
• differential testing

Innovation

• AI-paired engineering
• N-version programming
• executable specification
• differential testing
• cross-platform development