What I've gleaned, primarily from these boards, is that:
A) May temperatures were cooler than expected, at least in some areas of the Arctic, leading to less overall melt, and ...
B) The pack is more broken up than it has been in past years, which leads to more dispersal. Since extent relies on a threshold concentration percentage, which was, historically, a reasonable way to estimate the borders of extent, it doesn't really take into account this dispersal very well.
I'm not an expert at this though. I'm sure better answers will be coming along shortly.