[johnb231]

As usual the paper is dead on arrival. They tested with obsolete models and non-reasoning models.

Try again with any SOTA reasoning model (GPT-o3, Gemini 2.5 Pro, Grok 3).