Alvin Plantinga, answered from the texts and cited to the page.
The Free Will Defense is a defense (a demonstration that God's existence is logically compatible with evil), not a theodicy (an account of God's actual reasons for permitting evil).123 The defense uses three premises: libertarian freedom (a free act is not sufficiently caused by anything external to the agent); the truth of counterfactuals of freedom (F-conditionals) about what every possible free creature would freely do under specified circumstances, known to God via middle knowledge; and that it is possible — possibly true — that every free creature God could have actualized suffers from transworld depravity: in every possible world in which they exist as free, they freely go wrong at least once.4
Granted those possibilities, God may have been unable to actualize a world containing free creatures who never go wrong, even though God is omnipotent. The probabilistic argument from evil is engaged through theistic skepticism: our cognitive access to God's possible reasons is severely limited, and the inference from 'we see no justifying reason' to 'there is no justifying reason' is unwarranted.5
Such an attempt to specify God's reason for permitting evil is what I earlier called a theodicy; in the words of John Milton it is an attempt to "justify the ways of God to man," to show that God is just in permitting evil. Augustine's kind of theodicy might be called a Free Will Theodicy, since the idea of rational creatures with free will plays such a prominent role in it.The Analytic Theist, p. 57
In Flew's piece, for example, the question is clearly whether the existence of the evil the world contains is logically compatible with the existence of an omnipotent, omniscient and wholly good God.Alvin Plantinga, p. 42
This suggestion on their parts is sometimes called the free will defense; in this chapter I shall explain and examine it. First of all, a distinction must be made between moral evil and physical evil. The former, roughly, is the evil which results from human choice or volition; the latter is that which does not.God and Other Minds, pp. 144–145
As it turned out, sadly enough, some of the free creatures God created went wrong in the exercise of their freedom; this is the source of moral evil. The fact that free creatures sometimes go wrong, however, counts neither against God's omnipotence nor against His goodness; for He could have forestalled the occurrence of moral evil only by removing the possibility of moral good.The Analytic Theist, pp. 59–60
Or so it is argued. The problem of evil as presented in this argument is a logical problem because it is about the consistency of (1) and (2). It is also an abstract problem in that it is not concerned with any particular kind or sort of evil. It does not depend on Auschwitz or the notorious Lisbon earthquake, nor on how much evil there is in the world.Alvin Plantinga, p. 226
The logical problem of evil is widely conceded disarmed by the FWD; the philosophical journal record from the 1970s onward shows Mackie-style logical-incompatibility arguments mostly abandoned even by atheologians. The evidential or probabilistic problem of evil remains very much alive — Rowe, Schellenberg, and the horrendous-evils revival (McCord Adams) press it in various forms. Among Christian philosophers, the Molinist apparatus is itself contested: Adams (1977) argues F-conditionals cannot have determinate truth-values and rejects middle knowledge; Swinburne accepts contingent F-conditionals but denies God can know them prior to creative decisions. The FWD does not strictly require Molinism in every formulation, but my preferred version uses it. Gale's freedom-canceling-control critique misreads what middle knowledge claims and overextends the analogy with human- manipulation cases.