I hate to point out the obvious but valuable NSA-type traffic is text and voice...I doubt they care 'bout youtube and netflix streaming traffic. I'm sure in 'absolute terms' that statement is true but it is a lie of omission.
I'd be willing to bet 0.00004% is probably close to 100% of all the metadata that crosses a router used by a US-based ISP or an international fiber line owned by a US company. :P
> but valuable NSA-type traffic is text and voice...I doubt they care 'bout youtube and netflix streaming traffic.
You don't think they monitor YouTube for terrorist video?
You don't think they monitor other video sharing sites for video of terrorist organisations releasing propaganda? That seems like useful intelligence, and they'd be foolish to ignore it.
What if they just ran their calculations something like this?
Suppose a 10MB youtube video is watched 100x. That's 1GB traffic. The NSA could record the video (10MB), then simply record who watched it (maybe a few dozen MB in http headers, etc -- call it 40MB)
They could report they only monitored 50MB of 1,000 MB, or 5%.
If the video was watched 1,000x, they're "monitoring" around .5%. 10k views, they're monitoring .05%.
Pretty easy to see how they can report a true number that also doesn't reveal the extent of their wiretapping.
I'd be willing to bet 0.00004% is probably close to 100% of all the metadata that crosses a router used by a US-based ISP or an international fiber line owned by a US company. :P