Lewis is referring to the distinction between intentional action and unintentional reaction, the former being the freedom to volunteer inputs into the system beyond what the system itself would naturally produce. His theistic argument from reason involved recognizing on one hand that we necessarily presume we have such freedom for rational action, and on the other hand that we do not derive this freedom from an ultimately reactive reality nor from ourselves: consequently we should conclude supernaturalistic theism is true and deny that naturalistic atheism is true. (He refined this argument somewhat in response to criticism from the Catholic philosopher Anscombe, and presented the refined version in the 2nd edition of Miracles: A Preliminary Study in 1960, many years after The Great Divorce; but he may already have the revision in mind for TGD since that happened at the Socratic Club soon after MaPS was first published. I’m somewhat oversimplifying his argument for convenience; the revised version of the argument runs on a logical formality that’s hard to briefly explain, although I accept several versions of the argument. I also note that Lewis forgot to double-check the formal weakness the other way around, to see if it deducts theism out of the option list – having deducted atheism out he concluded theism by dichotomy, but strictly speaking he should have tested both options.)
Lewis doesn’t mean we exist prior to Nature, and his position does not necessarily imply this; he means our ability to rationally act is a direct spiritual gift from the father of spirits, Who is Himself eternal reality, the Most Real Reality, the independent ground of all reality. We act supernaturally in relation to Nature, although not in utter independence from Nature.
Lewis reconciles God’s foreknowledge with creaturely freedom, not because rationally free creatures exist extratemporally, but because God exists extra-temporally, immediately knowing by directly active experience all instances of any created system of space-time. Whether any rationally free creatures exist extratemporally or to what extents an extratemporal creature could exist, is beside that point. (I agree with Lewis on this.)
Thus as Lewis famously argued, no one thinks a person is less free to act because God presently sees the action they are choosing ‘now’; the same is true in regard to future actions.
I have always argued that Christian universalism at minimum means God persistently acts toward saving all sinners from sin. (Or rather that’s theistic universal salvation, but would be included in varieties of theistic universalism, including with uniquely Christian details if X-variety of Christianity is true.) That doesn’t mean in itself that God necessarily succeeds; a never-ending stalemate could theoretically be possible. That was in fact what I originally expected to find.
That qualification is not the same as denying that God can guarantee successful universal salvation, though. It becomes a question of God’s competency and/or a question of whether God reveals final success from an omniscient perspective. The latter is a question of publicly available (i.e. scriptural) revelation (not counting private assurances if any), and that’s an exegetical argument. As to the former, I often quip (paraphrasing Lewis from The Problem of Pain on a similar topic the other way around!) that it doesn’t take a specially robust faith to bet on God instead of the sinner being victorious. People are free to play against the Chessmaster (reffing Lewis again); people are not free to play more competently than the Chessmaster. Take that bishop if you insist, but He moves here, and here, and it is mate in three moves.
So I hardly need pre-existence of souls to grant either or both types of assurance about God’s persistence for all being ultimately victorious. I’m even doubtful pre-existence adds anything to the assurance – my assurance that God will save pre-existent rebel angels is not even slightly based on them being pre-existent to our own natural system, for example – but I would want to read Chris’ article before commenting on particulars there.
If by pre-existence Chris isn’t talking about temporal pre-existence of souls, but only about our ontological dependence on God in a relationship superior to our dependence upon the natural system we live in, then I might have no objections, arguing much the same thing myself. But such ontological intimacy is not something missing from my universalist puzzle.