Getting Up to Speed on Multi-Agent Systems, Part 4: Wave 2 (Why It Breaks)

By 2025, two things had happened. Wave-1 architectures were running in production (Anthropic had shipped its research system; the open-source ecosystem around orchestrator-worker patterns was maturing). The agentic coding turn had made clear that multi-agent was not the right tool for focused coding, and narrowed the interesting MAS question to “when we do use it, why does it break?” This wave is where I find the literature most useful, because it’s where empirical work finally catches up with the claims of wave 1. Getting Up to Speed on MAS Part 1. The LandscapePart 2. The VocabularyPart 3. Wave 1: Can Agents Coordinate At All?Part 4. Wave 2: Why It Breaks (you are here) Part 5. Debate, State, and Coordination (publishes April 28) Part 6. Verification Patterns (publishes April 29) Part 7. Benchmarks and What They Miss (publishes April 30) Part 8. Open Questions (publishes May 1) MAST: 14 Failure Modes from 1,600 Traces Why Do Multi-Agent LLM Systems Fail? (MAST) arXiv 2503.13657 ·…