aristotle - 12:10 am on Jun 29, 2013 (gmt 0)
The pattern for panda updates has been the garbage coming to the top and then it is skimmed off.
To me, this doesn't seem like a good way to do a roll-out. In fact, I'm not even sure I understand how it works. For example, how is the garbage skimmed off? Do defects become apparent as the roll out progresses, and then are quickly fixed? Isn't there some way to test everything beforehand. I don't understand.