Yoav Goldberg Disputes Study Based on Only 4 Participants, Author Fires Back
Sskkooppiitt · x · 2026-08-19
NLP researcher Yoav Goldberg replied that he actually read the study in question: it started with 46 candidates, which gradually narrowed down to just 4 — the actual participants the research conclusions rest on, implying a small and questionable sample.
The study's author pushed back, accusing Goldberg of cherry-picking: the interviews were only the final phase of a much longer research process that included workshops and questionnaires, and asked whether Goldberg had run it through an LLM or just searched with Ctrl+F rather than reading it.
Related event: Yoav Goldberg Questions Study Based on Only 4 Participants(2 posts)→
More from Research
- Century-old Carathéodory conjecture likely disproved with AI help — haider1 · 2026-08-19
- DeepMind's Kevin Murphy: Enabling automated scientific discovery with LLMs — GlenBerseth · 2026-08-19
- Criticism of Anthropic's lack of research on internal activations in weird basins — repligate · 2026-08-19
- Google's Proteus Architecture Unlocks Progressive Memory for Long Context — inductionheads · 2026-08-19
- Revisiting Scaling Laws: Data Quality Emerges as Key to Model Breakthroughs — auto_grad_ · 2026-08-19
- Claude helps solve TII McEliece cryptographic challenge — matthew_d_green · 2026-08-19