The article is wrong in that it assumes the initial state of the coin has as much duration as the subsequent states. If that was the case, the difference in dur

The article is wrong in that it assumes the initial state of the coin has as much duration as the subsequent states. If that was the case, the difference in duration would be the same as in the odd-even sequence : number 1 2 3 4 5 6 7 8 9 odd 1 1 2 2 3 3 4 4 5 even 0 1 1 2 2 3 3 4 4 difference 1 0 1 0 1 0 1 0 1 However, what happens here is different - you start at "half a state", meaning that the graph is a see-saw, as such: rotations 0.0 0.25 0.50 0.75 1.00 1.25 1.50 head 0.0 0.25 0.25 0.25 0.5 0.75 0.75 tail 0.0 0.0 0.25 0.5 0.5 0.5 0.75 difference 0 .25 0 -.25 0 .25 0