What level of non-intrusive surveillance are you comfortable with?

So with all the news about the recently leaked Verizon records transfer, I'm curious what people's opinions are.

What level of surveillance (non-intrusive, as in happens without interfering with your daily lives) are you comfortable with in order to safeguard your life/society/country.

We can't live without a system of some sort, as doing so would mean that "bad people" would be able to operate unhindered. We at least need a method of attempting to stop bad stuff before it happens. I say attempting, because there is no realistic system that would enable that, and any other system would always have false positives (thinking a person is about do something bad but they're not), as well as false negatives (not catching a person doing bad things). Not to mention the fact that there is absolutely no way to gauge the effectiveness of a surveillance system as you will never know how many bad things you failed to prevent.

--- Opinion ---

If the government had come out and straight up said, "we prevented a large scale attack on US soil using the information we collected", You can be sure that the general attitude would be slightly different.

Personally, I don't see a problem with an Echelon type system being used, as the more data that is collected, the better assessment you can make in regards to threats. If the government wants the records of call logs, go for it. Does it "violate" my liberties? I guess that depends on the letter or the spirit of the liberties you're talking about. If it helps narrow down threats, then let them collect / monitor that information.

What I do see a problem with, is abuse of said system. Since there will always be a human element (assuming Skynet doesn't take over), at what point is a person considered to have "abused" the system, and what oversight should be in place to prevent abuse? That it and of itself is a paradox, as if bad people know they're being watched / monitored, then they're not going to be doing bad things, or try to switch to a method they think isn't being monitored, effectively making the previous surveillance useless.

So how do we effectively monitor digital activity while both a) letting the "good people" know exactly how they're being monitored (transparency), and b) not letting the "bad people" know that they're being monitored (obfuscation)?