Actually, it is not obvious that earthquakes should periodically strike a particular section of fault (even if there are "characteristic" earthquakes, which is not generally the case).

In the long run, there must be a balance between the stress loading from plate tectonics and the stress release from earthquakes. However, the absolute level of stress is poorly constrained, and may be much greater than the stress drop. So if the critical stress to start an earthquake, or the "sliding friction" to stop an earthquake varies from earthquake to earthquake, the recurrence intervals could be wildly variable, as seems to be observed in many places.

Not sure this comment comes across, but seismic gaps as a prediction framework have gone down in flames, although they may still work in some places.

I think Mike read this issue correctly.