Assange is a very intelligent individual. In fact the people posting that Assange's pgp related post warrants laughter now won't get the last laugh (in due time).
NSA is known for breaking cryptosystems with implementation flaws, side channels, the bleeding edge of cryptanalysis (which in cases of things like padding oracles and chaining modes make practical differences), and by brute force (when key sizes are within their top notch cracking capability). Furthermore they are known to have sabotaged software to insert exploitable flaws and the CIA today will compromise compilers of specific individuals so that they compile backdoored binaries. Unlikely then, but replacing a popular hosted binary wouldn't have been beyond their capability.
It's not that unlikely they could crack some instances of PGP some of the time. Today the NSA docs reference being able to crack things like OTR sometimes, though unlikely.
Do you have any evidence that PGP cannot be broken? After reading through almost all of the snowden docs and 90's crypto wars and looking at the actions undertaken by the govt, how can the government possibly permit encryption that they can not break. Think about it in the lense of the post 9/11 hysteria. I understand the discrete logarithm problem and all the other important parts to show rsa is safe but computers do not produce truly random numbers and with that deterministic environment generating the critical primes for pgp I simple cannot believe that the NSA has not already enumerated a few trillion/++ factors.
PGP has 50,000 users counting by the keyservers, Tor has two million daily users. I don't think that's reasonable that 97.5% of the people who care enough to use Tor haven't used PGP in years. (Yes, that's a back-of-the-envelope calculation.)
My conclusion from that disparity is that PGP is broken by a flank attack: Its usability is so bad that the encryption goes unused.
