Aider tops LLM recommendations despite four months without updates
Developer marlenezw made a counterintuitive observation: whenever she asks an LLM about open-source coding harnesses, Aider — whose GitHub repo has seen no commits for about four months — consistently ranks first in recommendations, yet she barely knows anyone still using it. She suspects training data lag and floated a "data contamination" hypothesis.
Confirmed
- Aider consistently tops LLM recommendations while its GitHub has had no commits for roughly four months.
- marlenezw hardly knows any developer still using Aider.
- In practice, while testing Jev's context compression and comparing harnesses, she asked Opus 5 what Pi (Peter Steinberger's coding harness) was; the model had no idea and guessed "Pi is your colleague Sarah."
Unconfirmed
- "Data contamination" remains a hypothesis; no dataset-level evidence was given, and the causal link between Aider's dormancy and its top ranking is unverified.
Why it matters
- The discussion shows that training-data cutoffs can leave models badly outdated about tool ecosystems — a warning for developers relying on LLMs for tech selection.
- mitsuhiko (Armin Ronacher) joked that Aider "is written in Python and quite old — the perfect combination to seep into model weights," and Simon Willison made a similar quip about old idiomatic Python being "perfect for the weights," reflecting a shared community sense that old code is overrepresented in training data.
2026-09-23 ~ 2026-09-23 · 5 related posts
Primary sources
- [source] Why is dormant Aider still LLMs' top recommendation for open-source coding harnesses? — marlene_zw · 2026-09-23
- Why is dormant Aider (no commits in 4 months) still the top open-source harness LLMs recommend? — marlene_zw · 2026-09-23
- [source] Opus 5 hilariously guesses "Pi" is teammate Sarah when asked about coding harness — marlene_zw · 2026-09-23
- Mitsuhiko: Aider Is Entrenched in LLM Weights Because It's Old Python — mitsuhiko · 2026-09-23
- Simon Willison quips: old Python code is the perfect fit for model weights — marlene_zw · 2026-09-23