Applying the filter of falsifiability helps us sort through which theories are more robust. If they can't ever be proven false because we have no way of testing them, then the best we can do is try to determine their probability of being true.