I hate under educated views on topics they have no clue about.
Firstly , every company in the program cited have said THERE IS NO DIRECT ACCESS GIVEN to any US govt ( directly and unfettered) to their data centers and servers.
Second - the only data collected by other means is stored and not looked at , TILL a outside the US number triggers a specific inquiry
Third- The data collected is absent of Account holders name, it is the telephone bill one gets with just numbers on it and not the names of whom it belongs to
Fourth - Once a number form outside triggers showing up calling a " a number" in the US, then and ONLY then - do they HAVE YOU go to a judicial court to get a warrant to see WHOM that number belongs to!
Okay, so let's discuss the possibilities assuming the intelligence in the Guardian is accurate:
Firstly, what do they mean by direct. Secondly why is list what it is. There are omissions.
Let's look at the rest of the slide first.
The background is the submarine cable network. FAIRVIEW, ___ , BLARNEY. This isn't new NSA territory - they did it to Soviet cables in 1971. There's a 'Fairview' in a ton of places, one of which is Cheltenham - could be a US<->UK cable.
These taps get you raw wire data. Many sites are encrypted. If the NSA had broken the encryption platform it would only need upstream taps for US internal data. Coupled with the US ISP sinks and ECHELON, they'd have everything.
Working on this assumption, we can rule out any interception that involves simply copying the encrypted data after it leaves.
Now lets return to the questions we had:
Direct connection
1. Almost who uses Facebook could say they have a direct connection. People in various repressed regimes have to use proxies. So that's one definition, the 'non-proxy' direct'.
2. If you had a seperate fibre connection between these content providers and the NSA, that's another layer of direct. This is still perfectly above board - it would merely stop the NSA worrying about bandwidth problems when the data they are served via an approved request is given. So 'hardware' direct
3. The final definition is the one people bring up. Changing the definition of connection from 'any ability to send and receive' to 'the ability to remotely manipulate' gives us remote access. Again this could be with or without the fibre-link.
Permission
For a fibre link you'd need permission. But if all you are doing is option 2, then it's a reasonable request (albeit OTT unless you are making a lot of them). Given the Google statistics do not include FISA requests and they could be broad, we don't know the scope.
If option 3 was going on, Google would have to be complicit in some fashion (which could not involve senior management - plausible deniability is possible in theory). In an organisation like Google it's difficult to see that a single person doing favors for the NSA would work - there's checks and balances all over the place, you can't open a port up on a router without sign-offs. At minimum you'd need some middle management approval.
Denial
The denials focus on the scale of the requests and the idea of 'back-doors'. From Google's perspective it's not a backdoor, it's a front-door however. In addition they highlight that the transparency they provide is a false view. It's incomplete - it has to be - the orders they get tell them they can't disclose them, so they have to exclude the FISA powered orders from stats.
Regarding them hearing about the program title, well clearly - they aren't going to use the NSA code name when they ask for access are they?
In the U. S., many people have been arrested by the FBI after the FBI has supplied the knowledge and means to commit terrorist acts. These people would not be able to do this on their own because they were mentally incapable, did not have the knowledge or the explosives. These people were apparently identified by the FBI through their emails or social pages. Essentially, they were entrapped, yet the courts have found them guilty. I would like to know what role PRISM had in this spying on American citizens. It all just seems to coincidental.
The PRISM program is used to data mine essentially keyboard communications, along with probably some video and audio. Another program(s) is used to collect and store all telephone metadata. I am sure yet another program merges the information produced by these two programs to form a more complete picture. We know that the addresses on all paper mail we send is electronically scanned for sorting and routing, which means that the source and destination addresses are captured and stored in databases. This is done for USPS, Fedex, UPS, etc. Is the government also merging this metadata into their spying database?