When Stress Becomes Signal: Detecting Antifragility-Compatible Regimes in Multi-Agent LLM Systems

TL;DR

This paper introduces CAFE, a framework for detecting antifragility-compatible regimes in multi-agent LLM systems by analyzing how semantic stress affects their structure and potential for future learning.

Contribution

The paper presents CAFE, a novel statistical framework that identifies antifragility-compatible stress structures in multi-agent architectures through distributional analysis.

Findings

Semantic stress reduces average judged quality by about one third.

All architectures show positive distributional Jensen Gaps, indicating potential for antifragile learning.

CAFE can identify when architectures expose learnable stress structures despite performance degradation.

Abstract

Multi-agent LLM systems are increasingly used to solve complex tasks through decomposition, debate, specialization, and ensemble reasoning. However, these systems are usually evaluated in terms of robustness: whether performance is preserved under perturbation. This paper studies a different question: whether semantic stress exposes structured variation that could support future antifragile learning. We introduce CAFE (Cognitive Antifragility Framework for Evaluation), a statistical framework for detecting antifragility-compatible regimes in multi-agent architectures. CAFE models a controlled expected distribution of semantic stressors, reconstructs an architecture-specific observed effective stress distribution from multi-dimensional judge signals, and compares both distributions using a distributional Jensen Gap under a convex stress potential. A positive gap does not imply immediate performance improvement; instead, it indicates a convex-expansive deformation of the observed stress distribution, suggesting that the architecture exposes learnable stress structure. We evaluate CAFE on a banking-risk analysis benchmark with five multi-agent architectures: flat, hierarchical, debate, meta-adaptive, and ensemble. Across all architectures, semantic stress reduces average judged quality by roughly one third. Yet all architectures exhibit positive distributional Jensen Gaps with bootstrap confidence intervals above zero. These results show that immediate quality degradation can coexist with statistically detectable antifragility-compatible stress geometry. CAFE is therefore not an antifragile learner itself, but a measurement layer for identifying when and where antifragility learning may be worth applying.