If you look back in history.  Everytime there has been long blocks, and then a "restart" for one reason or another there has been a block found very shortly after, and the near 100 % will continue for a while.

I have seen this numerous times since December that this happens almost like clock work.


YES, saying that it does nothing, becuase of probability, can be an accurate statement, but when this happens too often, there is more to this than just probability.


I dont remember the date, but a while ago, Kano said he needed to do a pool restart due to a logging issue, and then the double restart cuased another issue, and he found why in the code this issue happened.  after that, blocks were normal again, near 100 %  but prior to that there were long blocks.. like now.

Persoanlly, I think there is an issue in some detail yet to be discovered, that after the pool runs for an extended period of time, or so many shares, blocks are not found as near 100% as they should. ( when you use "probability" alone )