The “Snowden Training Film” (and Discrete Log Problem Progress)

Some Internet sleuths have turned up what may be the video Snowden made for Glenn Greenwald while he was trying to set up a secure channel to the source.[1]

While JYA has called the authenticity of the video into question, I’m going to proceed under the assumption that it’s real. Not least because, though the audio is obfuscated in an attempt to defeat voice recognition… it sounds an awful lot like Snowden.

Basically, I can totally see why Snowden completely failed in his quest to get Greenwald to install GPG.

Snowden makes the classic geek mistake (also, spy/source mistake — both tech and espionage can produce similar brands of “situational Aspergers”) of assuming Greenwald is more or less as motivated as he is.

Also, Snowden makes the OTHER classic geek mistake of assuming the other guy is more or less as smart and knowledgeable as he is, just not in this particular area. Watch the video and you’ll see what I mean.

As someone who actually does know how to use GPG (though I intentionally don’t)… I came away feeling more confused and less knowledgeable than I started! Okay, only a little, but the video is full of “you’re smart enough to make this connection so I won’t bother” moments.

Pedagogical criticism aside, it’s worth noting a few things about Snowden’s operational security measures.

One, everything is generic enough that it’s not possible to derive indicators which would identify Snowden. There simply aren’t enough “bits” of information.

Two, there is nothing about the video itself which would identify the creator as someone with “inside knowledge.” To be sure it’s very detailed, but he goes to painstaking care to mention things only in the public domain.

Three, though he only references public sources, he goes out of his way to highlight the RSA algorithm. Why is this significant?

Well, today some news broke about mathematicians making progress about the “discrete log problem.” [3] The discrete log problem has HUGE significance for security and cryptography, though this particular breakthrough isn’t quite that big of a deal. [4]

However, it’s been widely speculated that the NSA may have a much better solution — something which would render older public-key crypto toast, requiring the use of the elliptic curve type algorithms which the NSA recommends for government communications.

As it happens, the RSA algorithm which Snowden recommends is not vulnerable to discrete log-based attacks [5] — which suggests Snowden may “know something” here and be trying to make life a little tougher for the NSA. Nevertheless, if Poitras’ example is anything to go by, Snowden regarded even the best-used PGP as only suitable as an interim solution to establish even more secure communications. Which has even more terrifying implications.

[1]
http://vimeo.com/56881481
https://www.youtube.com/watch?v=9mvf8VwVjJY&

[2]
https://twitter.com/digitalfolklore/status/466704166501363712
@Cryptomeorg this the same email #Snowden is using in video?

https://twitter.com/Cryptomeorg/status/466741802603515904
@digitalfolklore No. anon108@tormail.org in the video is surely a spoof. If it is not, then poor lesson for clueless GG. Video may be hoax.

[3] http://www.sciencedaily.com/releases/2014/05/140515163739.htm

[4]
http://science.slashdot.org/comments.pl?sid=5168953&cid=47023565
As I understand it, they’ve simplified the problem to a compiexity of O(n^log(n))… this is still non-polynomial time… but the rate of complexity growth is effectively polynomial. If I understand correctly, that means that the additional security that was formerly thought to be obtained by merely doubling cryptographic key length must now be obtained by squaring it.

[5]
http://science.slashdot.org/comments.pl?sid=5168953&cid=47023791
SmartCards actually mostly rely on symmetric algorithms for most applications. The only commonly used public key algorithm is RSA, which is not
based on discrete logarithm. This leaves DSA, among the relatively common algorithms, but that is rarely used on SmartCards. What would be
interesting to know, is how EC-DSA is affected, since it is slowly replacing RSA because of the reduced key size.

[6] http://www.nytimes.com/2013/08/18/magazine/laura-poitras-snowden.html?pagewanted=all&_r=0

Advertisements
%d bloggers like this: