Show HN: Predict LLM Predictions

Adversarial play might be interesting. I'm assuming the goal here is to use the feedback as training input, fitness to expectations and like.

So a random chance of a bad actor adversary? Or high scores for lying? "The traitor" style

Team models? I suppose I'm arguing for gamification with a leader board and hubristic bragging rights.