Nah, the building is (mostly) just as good at the end of that time. It's just a convenience to make the probabilities easier to conceptualize. If we designed buildings such that the design earthquake had a 5e-6 daily chance of occurring, we'd just be confused.
If there is a 10% chance of collapse every 2475 years, that is ~1.1/10million per day chance, or 0.11 micromorts. That assumes 100% death during a collapse, and that the person spends all day in one of these buildings.
0.11 micromorts/day is very low, which makes the design criteria reasonable. Of course, I suspect our ability to predict infrequent events with any degree of accuracy is unlikely.