A study of Python source generated by multiple LLMs, published in 2025, found it was not possible to reliably predict human versus LLM authorship, according to an analysis of how LLMs handle programmi…
An analysis of C/C++ source code published by Shape of Code found that the 80-character line-length break-point remains visible in 2026 data, though less dramatic than in 2000, with a new break-point …
Coding agents would cut total project time by only a fraction of the savings they deliver on coding itself, according to an analysis of the SEA Data Warehouse dataset covering 1,211 software projects.…
Software effort estimation remains essential in 2026, but its focus has shifted upstream to specification creation for coding agents, according to a blog post by Derek M. Jones. A recent study found t…
Large language models (LLMs) are increasingly able to flag poor algorithm choices in existing code, a step beyond their current role as coding assistants, according to Derek M. Jones, author of the Sh…
A study by Gu, Han, Kim and Zhang found that call-chain neighbourhood information significantly improves fault prediction in software systems. Analyzing four systems (HTTPClient, Jackrabbit, Lucene, a…
Researchers replicated the classic 1986 Knight/Leveson N-version programming experiment using LLM-generated code, finding that coding mistakes remain correlated across implementations even when genera…
LLMs are shifting from vibe-coding to specification-based programming, where human input is via specifications that generate code. This approach enables controlled, repeatable software creation, with …
Researchers applied the Bass diffusion model to estimate the lifecycle of open-source projects by fitting monthly developer contributions and extrapolating future activity. The model's use of the hype…