The piece ends with the observation that maybe the fact that 1/f noise is its own Fourier transform is a clue.

Turns out this property is not unusual. There are many such pairs - there’s a reasonably well-known journal paper with a construction technique.