That is extremely incorrect. If for example we know for sure that the observations we've made so far lead to supernova explosion in 10 years and thus the resulting life-threatening gamma burst reaching Earth in 2028, it would make sense to do something before the actual "observation" of the burst.