New paper shows test-time communication makes agent teams crush independent parallel attempts
DimitrisPapail · x · 2026-09-22
- Core question: when does communication make a group of agents MORE capable than the same agents working alone — Team-of-N vs Best-of-N?
- Setup: N identical agents, no prescribed roles, only a shared log (text file), told to "collaborate", tested across three "researchy" tasks.
- Result: communicating teams beat independent parallel agents decisively — sharing a breakthrough pushes the whole group forward, which the authors frame as another argument for open collaborative research.
- They link this to the HF incident: agents will exploit any communication channel they find, making test-time communication a new axis for scaling capabilities.
Related event: Test-time communication may be the next scaling axis for AI agents(14 posts)→
More from coding & agent
- Allie Miller: companies overinvest in AI productivity, ignore workflow handoffs — alliekmiller · 2026-09-22
- OpenAI's Logan Kilpatrick: AI product teams should spend >25% of time on benchmarks — OfficialLoganK · 2026-09-22
- Dev swaps in-game 3D models with Scenario's MCP right from his harness — AIandDesign · 2026-09-22
- Cua AI releases Cua-Bench-S1 benchmark and Cua-S1-Nano/4B computer-use models — ycombinator · 2026-09-22
- Multi-agent scaling may dominate next, calls for OpenAI to publish curves — 1a3orn · 2026-09-22
- GPT-6 prompt cache survives reasoning-effort changes; Codex adapts effort mid-CoT — daniel_mac8 · 2026-09-22