Looks like the "0% luck for a day" has finally come to a close at 27+ hours. That was BRUTAL!!!!
I don't think it's luck. There's a pretty consistent pattern there where we have bad luck for that block if the pool hashrate is (10 + n*0.23-n)PH/s, where n is any positive integer. The block at 10.23PH/s was 20.5 hours, and the one at 10.45PH/s was over a full day. Lord knows what would happen if we actually hit 10.67PH/s, it could be a couple day long block.
I hope slush looks into this and can find the error in his code. These long blocks need to stop.