I have previously expressed on Prosblogion my belief that Rowe’s argument — made most recently and thoroughly in Can God be Free? — against the existence of God from the impossibility of creating the best world founders on the concept of sufficient goodness and satisficing action and I don’t wish to revisit that issue. However, I do think there is an issue in the neighborhood which suggests a conclusion which will be unpopular with many: that God must create a world. The argument starts with a kind of dominance principle and I’ll call it the "swamping principle"–(SP) for short–because one category swamps the other.
(SP) For any two types of actions A and B if every token of A is better than any token of B, then God must bring about a token of A.
So let A = Creating a member of the set of sufficiently good worlds and let B = creating no world.
It looks, then, like on this assignment (SP) entails that God must create.
I don’t think this is a very interesting proposition given the operative sense of "must" and that I’m not an unrestricted libertarian.
What I find interesting is the following dialectic: Leibniz and Clarke agree that there’s a best possible world and dispute whether this is consistent with God being free and the nature of that freedom. Rowe denies that there is a best possible world and argues from that that there is no God, since a God would have to create the best. I agree with Rowe that there is no best possible world, but reject his argument on the grounds that God can satisfice. However, the satisficing response assumes that there is a class of action-types like A above which, together with (SP), entails that God must create which brings us back to the Leibniz/Clarke debate on God’s freedom.
There are lots of interesting threads leading out of this, but I wonder if anyone thinks they have a counter-example to (SP).