Your premise:
> If thousands of daily reports warn of damage that must be fixed immediately and most are of no consequences (obviously)
is correct, but your conclusion:
> then the problem is that the reports don't accurately reflect the severity of the damage and the consequences of ignoring it.
does not follow. To put this in probabilistic terms, you need to consider the expected value of the consequences, not just the probability.
Here one of the terms in that equation is the negative cost of killing 150 people. That cost is so high that even a fractional percent chance of building collapse means that the recommendation to fix immediately is credible, correct, and does accurately reflect the consequences of ignoring it.