Well, in my vision (after going many rounds with a friend designing this), the whole enterprise would be decentralized, and as much as possible rules #2 and #3 would be drilled into the new recruits, and there would be term limits etc. To make it as hard as possible to take over the organization.

As for who makes a decision to go into a country ... well, if you're squads of people from Peru and Japan etc. and there's escalating sectarian violence in Iraq, then you are welcome to go there. It's all decentralized. It's sort of like "actors" in Erlang, each running their own program and enforcing on others.

The only thing that's sort of more centralized would be the due process for those peacekeepers or groups of peacekeepers who violate rule #1. Besides this, peacekeepers would try to screen out peacekeepers who travel to an area where they are found to have a significant stake in the outcome.

Of course, this force would not need any weapons of mass destruction, nuclear weapons, etc. All those could be gradually destroyed by the various nations. They would, however, have crowd control non-lethal weaponry.

My point is that from a utilitarian point of view, it's even better to kill all the belligerents early (after warning them) than let the thing escalate into, say, the Russian Revolution where 20 million people died in fighting. Just the chilling effect alone that "violence only backfires" (analogue of "crime doesn't pay") will deter many from joining the escalation. As opposed to now where the US "world police" merely takes one side and arms them to demolish the other.