Encryption can be circumvented. It's hard, but doable for a state, when it targets one high-value suspect. But if everything is encrypted, they can't go "big data" on it and collect everything about everyone: if proper encryption is generalized, spying doesn't scale anymore. Big companies' ability to break the encryption between them an you is irrelevant: if they're the legitimate receiver of the communication, they…
Half the article is talking about scalable attacks on encryption. For example: "To an intelligence agency, a well-thought-out weakness can easily be worth a cover identity and five years of salary to a top-notch programmer. Anybody who puts in five good years on an open source project can get away with inserting a patch that 'on further inspection might not be optimal.'" I agree we should continue this arms race but…
Indeed, but my point is, in addition to be worth a lot, it's difficult to implement, and it's very fragile. Every time they use it, they gamble its secrecy, hence its effectiveness. So they won't use it for petty reasons, only for genuine national security matters.
The problem isn't that NSA works on ensuring national security: it's that the scope of what they consider national security, and thus justifies extra-legal measures in their eyes, grows unreasonably, becoming a threat to the robustness of democracy.
A spy agency's ideal environment is totalitarian, not democratic. For a starter, they're scared of free speech and accountability. There must be counter-powers, cancelling their natural tendency to push towards totalitarianism (this is in no way specific to the USA).
> If they decide they don't like you there are all sorts of other ways [than courts] they can screw with you.
Yes, but those ways don't scale. You can screw with a couple of people you dislike, but not with hundreds or thousands of people simultaneously.