I attended a talk given by W. Hugh Woodin regarding the Ultimate L axiom and I wanted to verify my current understanding of what the search for this axiom means. I find it to be a fascinating topic but the details are so far beyond my grasp.

Given the language of set theory, one can write down a multitude of first-order sentences. By Godel's Incompleteness Theorem, it is known that from the ZFC axioms one can only derive the truth-values of a (small) fragment of these sentences.

In the past, it was hoped (by Godel, among others) that the Large Cardinal Axiom hierarchy would provide an infinite ladder of axioms of increasing strength such that any first-order sentence in the language of set theory would be either provable or refutable from ZFC + LCA for some suitable LCA.

However, it is now known (?) that the LCA hierarchy (pictorially represented as the vertical spine of the set-theoretic universe V) is not enough to settle all such questions. In particular, there is an additional horizontal "degree of freedom" due to Cohen forcing: for instance, when it comes to CH, it is known (or merely believed?) that both CH and ~CH are consistent with the LCA hierarchy.

Now, let a "completion of ZFC" be an assignment of truth-values to every first-order sentence in the language of set theory, such that a sentence is true whenever ZFC proves that sentence; moreover, for the other sentences (i.e. those which are undecidable in ZFC) the assignment of truth-values must be consistent.

My understanding of Ultimate L is that it picks out a unique completion of ZFC as being the "correct" one; that is, even though Cohen forcing allows us to have models (and therefore completions) of both ZFC + CH and also of ZFC + ~CH, Ultimate L eliminates the horizontal ambiguity and provides us with a unique completion of ZFC in which the truth-values of first-order sentences only depend on the vertical LCA hierarchy.

Is my understanding correct? And how do we know that there are (infinitely) many different completions of ZFC in the first place? Could it be that there is no way to consistently assign truth-values to all first-order sentences, i.e. that no completion exists?

Also, how would we know that Ultimate L + LCA picks out a unique completion (as opposed to a class of completions)? And would it be a valid completion (does consistency of ZFC + Ultimate L follow from Con ZFC)?

I would appreciate answers to any of the above questions, as I can't find anything on this topic in the literature. Thank you!