To be honest I'm still catching up on a lot of the details with these stories, but it's this comment that made me wonder if there is a bit of hyperbole going around. Companies make leaks all the time on the most mundane things (hell Apple can barely contain the new iPhone that is usually coming out.) I'm hesitant to believe all the conspiracy theories going around that the companies willingly did this, or that it's as easy as it sounds.
> “They quite literally can watch your ideas form as you type,” said the “career intelligence officer” who gave documents describing PRISM to the Washington Post.
Think of the engineering needed at multiple endpoints to allow something like this...I can believe the NSA has some of the best cryptoanalysts, but our best front-end hackers as well (do they have their own Meteor.js)? I'm only being slightly pedantic here. If this is "literally" possible, then the amount of work needed for the collaborating companies is non-trivial. It's one thing to pass along uber-authentication credentials allowing an NSA-agent Zuckerberg-like privileges...that could be something implemented with as few as a couple people. But to bake in something right out of a hacker movie?
It just makes a few of the claims sound a bit suspect...because what is terribly frightening is the government's ability to aggregate and analyze this information in bulk, not to peep in on you as you're typing in real-time, which would be one of the least efficient ways ever to spy on the general American public's online activity.
And no, I'm not being an apologist for federal expansion of powers. I'm pointing out that some of what this source is leaking seems to be beyond reality, not because of technological sophistication, but because of the number of mundane moving parts and actors that would seemingly have to be involved (theoretically, wouldn't they have to have as many datacenters as all the companies they're vacuuming from?). The entire PowerPoint slide set looks like something a get-rich-quick contractor would whip up to win a fat contract that would never actually be scrutinized for viability.
But still, even if the government doesn't actually have the capability as described, it is wrong for them to not disavow it...it's not any good if their mindset is: "Oh we're not doing that...yet...but we would love to, some day)
I think "as you type" means different things to an intelligence officer and a programmer. The intelligence officer probably means "real time" (no need for a judge/approval/etc). A Google Wave like interface is probably not in the backend, but if the FISA route took weeks and the new kit takes seconds I could see "literally as you type" being a description that fits.
Yes...maybe I reacted too strongly to the often misused "literally."
But even real-time interception would not (seemingly) be that valuable to analysts. Assuming that suspects aren't dumb enough to use their own personal accounts and to talk of their plans explicitly, it seems analysts would want the historical aggregate of which alias accounts talked to which other alias accounts.
Hell, Facebook can already figure out if you're connected to someone even if you've never directly interacted on the service...if the allegations are true, I wonder how much of the requested data comes with pre-baked network-calculated goodness?
Obviously all conjecture here, but what came to mind for me is things like having a keyword filter on a bunch of accounts (whoever you're watching). The moment they Google for something you want to see you get an alert that it happened.
You do not need to have belief systems; knowledge is more powerful. And conspiracy theory is a loaded term. Remove both from yourself to process information better.
