Trying to push a coin off centre
Between 1979 and 2007 the Princeton Engineering Anomalies Research laboratory asked a straightforward question: can a person, by intention alone, bias the output of a random event generator?
PEAR ran an enormous number of trials. What they reported were effect sizes so small that they only became visible across hundreds of thousands of them, and the argument about what those numbers meant — an effect, a subtle bias in the equipment, or the ordinary behaviour of very large datasets — outlived the laboratory itself.
Protocol P-05 puts the same experiment in your hand. A live random stream runs; you try to push it high or low; the instrument shows you the running deviation from chance and the number of trials it took to get there.
It reports what happens. If you drift above chance it will say so, and it will also say how unremarkable that is at your trial count. The honest version of this experiment is better than a rigged one, because at some point everybody gets a run, and watching yourself want to believe the run is the whole lesson.