Fetching from the wire…
01
02
03
04
05
06
07
08
Agents2026-08-14 · source-backed
arXiv 2608.12895 reports a preregistered evaluation over 18,000 missions: when either of two agents fails, both fail on 90.0% of those missions (log OR 6.66, 95% CI [6.38, 7.00]). That positive dependence invalidates the independence assumption behind compositional reliability math. Multiplying per-agent success rates systematically overstates pipeline reliability, and adding a second agent buys far less than the arithmetic promises.
Each link below shares sources, entities, or timing with this story.
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (agent, assumption, evaluation); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (agent, compositional); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (agent, evaluation); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (agent, either); pushes against this story (but).
Reported by the same outlet (arxiv.org); overlapping topics (adding, agent); pushes against this story (against).
Same source domain / Shared topic / Downstream implication
Reported by the same outlet (arxiv.org); overlapping topics (agent, fail); traces where this leads (downstream).
Same source domain / Shared topic / Tension
Reported by the same outlet (arxiv.org); overlapping topics (agent, compositional); pushes against this story (against).
Reported by the same outlet (arxiv.org); overlapping topics (agent, assumption); pushes against this story (but).