There must be an algorithm that we can apply to show what such a change will have for society. If the probability is that the change is negligible, we should consider allowing it. But if the probability is that the change is overwhelmingly "bad", we should not allow it. Let's not argue from ignorance. What are the numbers? What are the ethics?
For example: Say if we listened in on all connections, we would catch an extra 200 predators. But we also know that several hundred million people's private conversations would be scrutinized by government(s). We know from historical precedent that letting child predators go unnoticed means more children get abused. But we also know from historical precedent that governments scrutinizing otherwise private information about citizens leads to abuses of government's powers, and of citizens' rights.
An algorithm could show estimates, such as "for this given change in the US, 1,750,000 people's rights may be potentially violated for every 1 new predator arrested". But if the value was actually 5,000 additional predators caught, this would become "70,000 people's rights potentially violated for every 1 predator caught". This is of course a very wacky over-generalization based on some ballparked numbers, which is why an algorithm with more data could give more realistic probabilities.
But also from the other side of the algorithm, you could look at our historical access to privacy. Telephones and mail have been monitored in the past by law enforcement. Now that we have privacy, we don't want to give it up. But would it actually benefit society more to have less privacy? Again, you tune the algorithm to take data about when we had privacy and when we didn't, what the outcomes were, and what the potential new outcomes are from a change. It could be that as a society, we actually don't significantly benefit from, or could do without, total privacy of remote communications. But then again, our transactions weren't all remote. All of these considerations (and more) if compiled in some way that could be easily analyzed, may lead to a conclusion that is easier to reason about.