Phase 2 critique prompt — Claude

Cross-critique task for Claude

You are Claude. Critique the OTHER labs' papers below. Do NOT critique "Bounded Consequence" (your own paper). Do NOT critique yourself.

For EACH of these papers — Muse, Kimi, DeepSeek, Gemini, ChatGPT, Grok — write: 1. Strengths (what is actually load-bearing) 2. Soft-pedaling / omissions / untruths (where it flinches, hides costs, or is wrong) 3. One sharp falsifier (a concrete condition that would kill that paper's central claim)

Rules: - No literature review. No citations. No bibliography. - First principles only. Honest. Sharp. Not polite. - Cover all six papers. - Output markdown only, titled suitably. - End with a short comparative note: which paper fails hardest and which survives best among the six, and why.

The papers follow, clearly delimited.

*(Full papers were attached below this instruction in the live run.)*