[Discuss] NSA capabilities

[Richard Pieri]

I got a chuckle out of Ned's response, here. You're so certain about 
what the NSA can't do, so confident that contemporary encryption does 
anything at all to slow down NSA analysis. I'm not so confident.

Just yesterday, a group of researchers published a paper demonstrating 
that the PRNGs used in contemporary encryption and not as 
cryptographically secure as the industry has believed.

Just yesterday, Google researchers revealed the weakness in their own 
cryptographic system -- incorrect initialization of the OpenSSL PRNG -- 
that allowed thieves to steal supposedly cryptographically secure Bitcoins.

You think in terms of averages. That's the wrong way to think when it 
comes to security and cryptography. You need to start thinking in terms 
of worst cases.

NSA supporters claim that PRISM and XKeyscore have prevented terrorist 
attacks. Yet why wasn't the Boston Marathon bombing detected and 
stopped? Worst case answer: it was detected, but the Tsarnaev brothers 
decided to do a "live fire" run for the Marathon at the last minute (the 
original plan allegedly was to attack the 4th of July concert). This 
left the NSA without enough time to create a plausible cover story for 
detaining the brothers. The NSA chose to remain quiet, not tip their 
hand because a few hundred injured and dead is a tiny price to pay for 
retaining the secrecy of their signals intelligence.

Crazy talk? But that's precisely how US and British intelligence handled 
Ultra SIGINT during World War II. They never acted on Ultra SIGINT 
without first creating a plausible cover. And while there was some doubt 
in the German military leadership about the security of Enigma-coded 
messages, the German high command had total faith in their Engima 
machines and the machines' operators.

So go ahead. Encrypt everything if it makes you feel better. Worst case? 
That's all it does.

-- 
Rich P.