When there is sufficient data, the chance of another earthquake is indeed lower early in the cycle, but after about 2/3 of the way into the earthquake cycle, the chance of a quake seems not to rise much more. So overdue is not the same as that the chance of an earthquake is much higher than otherwise.

It has NOT been observed, either, that the longer since the last quake, the larger is the next one. Earthquake cycles do not follow the simplest imaginable model.

As the authors themselves say, there may be more going on in the system - interaction with nearby earthquakes, episodic aseismic slip, missing events, spurious more local events, faults with alternating activity.