Tim Gowers set up a new journal, Discrete Analysis — an arXiv overlay journal.

His new blog post shows that the Journal’s standards are set high:

I imagine most people reading this will already have heard that Terence Tao has solved the Erdős discrepancy problem. He has blogged about the solution in two posts, a first that shows how to reduce the problem to the Elliott conjecture in number theory, and a second that shows (i) that an averaged form of the conjecture is sufficient and (ii) that he can prove the averaged form. Two preprints covering (i) and (ii) are here and here: the one covering (i) has been submitted to Discrete Analysis.