But that article says the opposite thing, that a certain form of rejection really does work: "Thank you for developing a good approximation to the trisection o

But that article says the opposite thing, that a certain form of rejection really does work: "Thank you for developing a good approximation to the trisection of an angle! It has an average error of only 0.1 percent, according to this computer analysis of all angles from 0 to 180 degrees in steps of three. It compares favorably with past approximate trisections developed by X, Y and Z that work as follows..."