I'd just like to draw your attention to bitparking where a ~100GHs miner just solved a block for the pool. The "impossible" happened not once but twice - once in January and once in February. If you flip a penny 3 times and get heads 3 times in a row, what are the odds of getting heads on the 4th flip? If you have to think about this one, you are statistically confused.

What appears to have happened here is unlikely, but not impossible. All you need in order generate a block is one hash. It just needs to be the right one. Anyone with any amount of hashing power has some non-zero chance to generate a block.