Many health or environmental hazards are examples of unforeseen cumulative effects:
- PCBs in the environment
- Endocrine disruptors
- Ozone depletion catalysts
- Software bugs
Just consider software bugs. If a program has a million or more lines of code, at no time can one certify it bug free. You can only aver that none have shown up yet, while trying to determine whether a little bobble in performance comes from the software or something else (like the case of the quirky 2010 Prius ABS system). And will code buried in an old chip really work after thousands of thermal cycles and activated for the first time 10 years later (demanded of automotive air bag software today)?
Take pharmaceuticals. They do not work the same way on all patients, so a clinical trial is never complete confirmation of efficacy in every patient or absence of a fatal side effect in any patient.