We create consistency checks from base forecasting questions, which we take from various sources (prediction market questions, synthetically generated from news, purely LLM-generated); ask the forecasters for probabilities, and check how consistent the predictions are (4/11)