The following argument is taken from a paper by Alexander Pruss. What I'm concerned about is the questions that it raises about God's nature; but before that I should explain the main point of the argument. Essentially, it is a variation of what is commonly known as the free will defense. I will call it the Prussian Free Will Defense. (That sounds like a chess strategy, doesn't it?) What the paper aims to show is that God's omnibenevolence and ability to make creatures with significant freedom does not preclude his creating creatures who freely choose to do evil. It essentially runs like this.
First off, the whole reason we are trying to make a free will defense is because we assume there is some moral principle of an omnibenevolent God which is incompatible with the actual existence of evil in the world. This is hopefully an accurate statement of how the atheist understands the situation to be. Let's call it the No Evil People (NEP) principle:
(NEP) There is some moral principle in the nature of an omnibenevolent, and omnipotent God which prohibits him from creating a creature who does something immoral.
We assume that significant freedom is a great good. That's how free will defenses work. This is what I mean by the phrase "significant freedom." Someone performs an action significantly freely just in case he freely refrains from a duty or freely refrains from doing something immoral. So someone who performs a significantly free action has significant freedom. Let's call the following the Significant Freedom Principle (SFP).
(SFP) It is not the case that God’s omnibenevolent nature contains a moral principle that would make it impossible for him to create a significantly free person.
Now, this is the general form of argument. What we argue for is the conditional C: If (NEP), then not-(SFP). If we take that as our first premise, and our free will defense says that (SFP) since we assume the great value of significant freedom, it follows that not-(NEP). And if not-(NEP) then there is no logical contradiction between God's creating a world with creatures which commit evil.
Here's how we argue for that conditional. First we need this Highly Plausible Principle (HPP):
(HPP) If it is logically impossible for someone to do something immoral, then he lacks significant freedom.
Classical theists usually understand God to be a necessary being, i.e. one which exists in all possible worlds. God is understood to be the creator of all things other than Himself. That means that necessarily, any contingent being is a creature of God. So if (NEP) is correct, and if it is logically necessary that every contingent being is a creature of God, then it is logically impossible that a contingent being does anything immoral. Thus, it is logically impossible that any person does anything immoral. Hence, by (HPP), it is impossible that any person have significant freedom. This implies that (SFP) is false, and thus our C is true. Modus tollens then, (NEP) is false.
I won't argue for the idea that God is a necessary being. I'd stand in the traditional theistic tradition by saying that any contingent entity can't be what is understood by the classical God. Besides, Pruss offers many arguments to show that C still holds should we understand God to be contingent. The main point is the conclusion that we have to draw from the Prussian Free Will Defense:
(conclusion) It is not the case that there is some moral principle in the nature of an omnibenevolent and omnipotent being which prevents him from creating a creature who does something immoral.
This raises some questions about the nature of omnibenevolence (can you see them?) which I hope to look at later.