Comment by mobeets

This approach implies the probability doesn’t depend on N. It only happens to be 1/2 for N=4 (the article goes into this). The trick is that you don’t know beforehand which semicircle all the points can land in, but your unwrapping step assumes you do.