Encrypted, if you're using https everywhere (and
Wikipedia hasn't
intentionally or unintentionally compromised their
certificate).
But simple encryption that NSA can break at will.
No one will bother trying to break SSL/TLS. The NSA certainly doesn't
need to. They can just sign their own certificates and perform
man-in-the-middle attacks. Browsers will in most cases accept those
forged certificates, since the NSA can make sure that they are signed by
a CA trusted by many browsers.
HTTPS Everywhere (which I mentioned) includes a "Decentralized SSL
Observatory" to try to detect exactly this. If the NSA wants to keep their
spying a secret, they won't do a MITM attack, because they'd get caught.
I suspect if they were doing this with a significant portion of traffic,
they'd have been caught by now, and that it'd be a story I would have heard
of.
So what's left is breaking the encryption after the fact. I'm not aware of
how much difficulty this is (or even what encryption is used by Wikipedia),
but it's probably going to slow the process down to where they're less
likely to go on pure fishing expeditions. Once they have a target, sure,
but just to make lists of people viewing certain Wikipedia articles, I
doubt it.
Maybe if the algorithm itself has been broken, or NSA has a whole lot of
quantum computers the public doesn't know about, or something like that,
but otherwise, I don't see them doing this en-masse. Storing the encrypted
communications en-masse for later cracking, maybe.
Or maybe I'm wrong about the difficulty of breaking Wikipedia's HTTPS.
Anyone have any figures? Should Wikipedia be using stronger encryption?
(A quick search shows that there might be a problem with RC4:
http://nakedsecurity.sophos.com/2013/03/16/has-https-finally-been-cracked/)