I can't find the source, but I think it's Scott Aaronson who told a story of a device with two buttons, which students were invited to press as rendomly as possible, but training a simple Markov model allowed them to predict what button an individual would press next most of the time. Student after student tried to trick the predictor, and failed. Then this one guy comes along and mashes the buttons and the prediction accuracy never goes above 50%. When they asked how he was doing it, he said he "just used my free will".
I can't find the source, but I think it's Scott Aaronson who told a story of a device with two buttons, which students were invited to press as rendomly as possible, but training a simple Markov model allowed them to predict what button an individual would press next most of the time. Student after student tried to trick the predictor, and failed. Then this one guy comes along and mashes the buttons and the prediction accuracy never goes above 50%. When they asked how he was doing it, he said he "just used my free will".