Hacker News Reader

Comment by JasonFruit | Hacker News Reader

I have a hard time reading this with an open mind, because I mistrust the NSA so much. That being said, the processes described in the article sound like they will arrive at a fairly transparent (in a good sense), publicly-vetted algorithm --- or at least one where it's easy to perceive if it's not trustworthy.

I'm not sure, based on my conversations with people in the field, that I put so much stock in the "20 years 'til Shor" prediction. Most seem to be of the opinion that the best currently available encryption should be okay until we're very old.

View on HN

NSA put (effective) backdoor in Dual_EC_DRBG:

https://www.schneier.com/blog/archives/2007/11/the_strange_s...

Thread by someone directly involved on why the ISO rejected NSA ciphers in the past (hint: they refused to justify design decisions, lied, and attacked the credibility of people who had put out actually-secure crypto):

https://twitter.com/TomerAshur/status/988696306674630656

Either they're too incompetent to be trusted or are bad actors and should be treated as such.

> Either they're too incompetent to be trusted or are bad actors

I agree but am curious about one thing. If they aren't stupid won't they realize it's a terrible idea to embed weaknesses in standards destined to become so pervasive your own economy will rely on them (as well as your own military/intel)?

Also, history virtually guarantees that your most secret secrets are reasonably likely to become known to your adversaries and used against you with catastrophic consequences? Even if they could entirely subvert the standards process to plant their own backdoor that ends up everywhere, game theory dictates they might be creating their own greatest future weakness.

> If they aren't stupid won't they realize it's a terrible idea to embed weaknesses in standards destined to become so pervasive your own economy will rely on them (as well as your own military/intel)?

The Dual_EC_DRBG backdoor [1] is a asymmetric crypto based backdoor. Only the one who built the back-door has the key to access it. It's even impossible to prove the backdoor is there without knowing the key.

This way you don't weaken yourself or allies WRT 3rd parties when standardizing on that specific weakened algorithm (of course you need to closely guard your key).

[1] https://en.wikipedia.org/wiki/Dual_EC_DRBG

The military is still on unencrypted IRC, the dog has already been fucked for years on that one. The economy doesn't really matter, and for any person they deem does matter, they can just tell them under NSLs to use something that DJB made, instead, if they even understand the difference between protocols to begin with.

From what I'm aware, the military doesn't use IRC? IRC was inspired from BITNET, which was used by the DoD briefly and not the military as a whole? Military networks are encrypted at the network layer? User authentication is enforced at the session and presentation layers?

https://www.duffelblog.com/2012/10/wikileaks-releases-mirc-t...

Unencrypted IRC (esp on a secure network) sounds like a massive upgrade from what I'd answer if someone here said "Guess what the military just switched to for their secure chat??" (slack or some other commercial hosted web atrocity). :)

They can have highly competent researchers and still be institutionally stupid.

NSA fixed the S boxes in DES, protecting users from a type of attack that wasn't yet public knowledge. Many people wrongly assumed that the NSA was trying to weaken DES, but they strengthened it.

> That being said, the processes described in the article sound like they will arrive at a fairly transparent (in a good sense), publicly-vetted algorithm

Or a family of them for different applications anyway, which increases chances both that a) at least one of them is actually quantum resistant and b) at least one of them will have a flaw exploitable by even classical computers.

Reply on news.ycombinator.com