If I remember correctly, the first set of tests was explained due, ironically enough, to not taking relativity into account when they were measuring the speeds. (The satellites they were using as their stopwatch were moving relatively slower than the neutrinos, causing an incorrect reading... or something like that; physics isn't my strong suit.) In the article cited in the OP, it states that they changed the experiment so that the neutrinos were bunched closer together, allowing for better accuracy in measuring when they get there. But there was no mention of the satellite timing issue. Does anyone know anything about that?