Math scooping norms debate sparks Anthropic vs OpenAI discussion
On September 7, the math community debated whether scooping research others are actively working on is acceptable, sparked by @nihilunbounded pointing out math's unwritten social norm against scooping others' in-progress projects, while sarcastically noting it's increasingly clear Anthropic cares less about math than OpenAI.
Confirmed
- @littmath acknowledged scooping does happen in math, sometimes unintentionally, but he dislikes the practice and believes people should verify things before announcing results.
- @littmath clarified his stance: what he opposes is jumping ahead while research is still in progress; the stage of "work completed but paper not yet written" is another matter—that norm against scooping he fully agrees with.
- @littmath also argued math has a norm against scooping young researchers on small problems, but major open problems carry no such restriction; most problems have few people working on them simply because the field is so vast.
- @alzzyd added an observation: many unsolved problems seem unsolved because only one person works on each while everyone else tacitly avoids it—a pointed remark on the collaborative culture of math research.
- @nihilunbounded responded that scooping norms protect not just juniors—he's seen senior researchers affected too—and gave examples of major problems that also go un-scooped.
Why it matters
- The discussion extends academic scooping ethics into the AI lab context: if organizations like Anthropic or OpenAI use large models to do math "in parallel at scale," the traditional one-person-per-problem tacit understanding could break down—also the backdrop for @nihilunbounded's dissatisfaction with the two labs' differing attitudes.
- The discussion clarified that "scooping" isn't a single concept: jumping ahead on in-progress work versus preempting completed work are judged very differently by community norms, a useful reference for defining behavioral boundaries as AI participates in math research.
2026-09-07 ~ 2026-09-07 · 5 related posts
Primary sources
- [source] Math's unwritten no-scooping norm sparks debate over Anthropic vs OpenAI on math — alz_zyd_ · 2026-09-07
- Are hard math problems unsolved because only one person works on each? — alz_zyd_ · 2026-09-07
- [source] Math's no-scooping norm: young researchers protected, big open questions fair game — nihilunbounded · 2026-09-07
- [source] littmath: scooping finished-but-unpublished work is where the norm applies — littmath · 2026-09-07
- littmath weighs in: 'People should be sure before they announce' research results — littmath · 2026-09-07