Minimalist Human-in-the-Loop & LLM Output Ranking Arena with preference convergence.
Run interactive pairwise human-in-the-loop ranking and preference convergence on candidate outputs (text, code, markdown, images). Use to rank LLM responses, evaluate taglines or design concepts, capture user likes/dislikes, and iteratively evolve generation N+1 toward user preferences.
