Wouldn't the not so strong old buildings have collapsed during previous quakes?
Possibly, and then rebuilt. They rebuild houses every 20, 30 odd years anyway (according to https://www.theguardian.com/cities/2017/nov/16/japan-reusabl...), and temples too https://longnow.org/ideas/long-term-building-in-japan/.
They could have been further away from the epicenter of the previous ones, or the previous ones weren't strong enough to cause that specific structure to collapse, the previous ones might have weakened the structure and this one brought it down, a million different things factor into this.
Earthquakes aren't a perfectly repeatable occurrences you can run a regression test against.