As always, the preference is for full consensus. Rough consensus (which I personally would consider to be less than 10% opposition) is usable but far less desirable than unanimity. ...
Rough consensus has worked pretty well in the IETF. The only groups I know of that work by full consensus are Quaker meetings and criminal juries. The Quakers have a tradition that you "stand aside" if you disagree with the majority, but not so vehemently that you are willing to block everyone else. Quakers being Quakers, it works for them, but anywhere else it means that the group is often held hostage to one stubborn member. In juries, where someone's life is often in the balance, that's a good thing. When we're deciding how to word yet another comment to ICANN, it's not, so I agree that rough consensus is appropriate here. R's, John