There has not been a lot of recent news in the ECC world, so this blog has been quiet.
As far as I am aware, there was little of importance to ECC announced at CRYPTO. Of course, the results of Lenstra, Hughes, Augier, Bos, Kleinjung and Wachter, as well as Heninger, Durumeric, Wustrow and Halderman, show that one must ensure any device that generates public keys has enough entropy. Such results apply to any public key cryptosystem, and elliptic curves are not immune.
The CRYPTO rump session can be found at this address http://crypto.2012.rump.cr.yp.to/. It featured two relevant presentations: Marc Joye discussed some convenient features of Edwards curves for side-channel resistant implementations, and there was a talk on the recent Japanese discrete log record in F_{3^{6*97}}.
On eprint, a notable recent paper is 2012/458 Computing small discrete logarithms faster by Dan Bernstein and Tanja Lange. The paper explains how to use precomputation to speed up the rho or kangaroo methods. This makes sense if one is going to be solving lots of instances of an ECDLP in the same “small” group or interval. Previous work on this problem was done by Kuhn and Struik, but they considered a small number of ECDLP instances. Instead, Bernstein and Lange push the method further to get a cube-root algorithm for the DLP in a group/interval.
— Steven Galbraith
Indeed the shared primes attack was an interesting finding hidden in plain sight…